Communications in Asteroseismology - Universität … · Communications in Asteroseismology Volume...

Communications in Asteroseismology

Volume 146June, 2005

Austrian Academy of Sciences PressVienna 2005

Editor: Michel Breger, Turkenschanzstraße 17, A - 1180 Wien, AustriaLayout and Production: Wolfgang Zima

Editorial Board: Gerald Handler, Don Kurtz, Jaymie Matthews, Ennio Porettihttp://www.deltascuti.net

COVER ILLUSTRATION: Selected sample screen shots of the programPeriod04.

British Library Cataloguing in Publication data.A Catalogue record for this book is available from the British Library.

All rights reservedISBN 3-7001-3505-X

ISSN 1021-2043Copyright c© 2005 by

Austrian Academy of SciencesVienna

Austrian Academy of Sciences PressA-1011 Wien, Postfach 471, Postgasse 7/4

Tel. +43-1-515 81/DW 3402-3406, +43-1-512 9050Fax +43-1-515 81/DW 3400

http://verlag.oeaw.ac.at, e-mail: [email protected]

Contents

Editorialby M. Breger 4

New variable and multiple stars in the lower part of the Cepheid instabilitystrip

by Y. Fremat, P. Lampens, P. Van Cauteren and C.W. Robertson 6

OV And, a new field RRab Blazhko star?by K. Kolenberg, E. Guggenberger, P. Lenz, P. Van Cauteren, P. Lampens

and P. Wils 11

Search for intrinsic variable stars in three open clusters: NGC 1664,NGC 6811 and NGC 7209

by P. Van Cauteren, P. Lampens, C.W. Robertson and A. Strigachev 21

High-degree non-radial modes in the δ Scuti star AV Cetiby L. Mantegazza and E. Poretti 33

Projected rotational velocities of some δ Scuti and γ Doradus starsby L. Mantegazza and E. Poretti 37

Application of the Butterworth’s filter for excluding low-frequency trendsby T.N. Dorokhova and N.I. Dorokhov 40

PODEX – PhOtometric Data EXtractorby T. Kallinger 45

Period04 User Guideby P. Lenz and M. Breger 53

Comm. in Asteroseismology

Vol. 146, 2005

Editorial

This issue is the biggest one we have published so far. Once again, you canfind a number of articles with the latest research results.

Apart from the many new exciting discoveries, we have included the manualfor PERIOD04 - the latest incarnation of our multiple-period-finding softwarepackage. This is the 25th anniversary of the first version. A long time ago, Iundertook a minisabbatical from the University Texas at Austin and went toVienna to write the program. The PERDET and MULTIPER versions weresimple Fortran programs not catering to the comforts of the users at all. Butthey worked and are still used in some places. With time, more simultaneousfrequencies could be considered, and amplitude as well as phase variabilitybecame an issue. With versions PERIOD90 and PERIOD95, the human brainbecame too small to remember all the options and how to call them, e. g.,Enter ’r’, then ’x’, etc.

So, Martin Sperl added a graphical interface and more options to the pro-gram, which became PERIOD98. This program was downloaded by many uni-versities and observatories. His publication in the Communications became themost widely quoted masters thesis we have had. His number of citations evenput some professors to shame! If your university persecutes you with the citationindex, learn from him.

For the last few years, we spent some of our time improving the programand (horrors!) allowing for near-infinite combinations of a near-infinite numberof simultaneously present frequencies. (Of course, this also allows you to flauntall laws of statistics and significance, if you wish.)

So Patrick Lenz programmed all the changes and did a wonderful job. I useddozens and dozens of new versions of the program on the Delta Scuti Networkdata that came in and suggested/debugged more and more features. Theprogram even determined that over the years FG Vir with its 80+ frequencieshas no orbital light-time corrections larger than a few seconds! I just hope thatPERIOD04 with all the new features needed by our research has not emulateda famous OFFICE suite ..... enough said.

Allow me a side comment. There exist a number of different programspackages and approaches to help analyze the periodic content in photometricdata. I have participated in a number of successful blind and semi-blind tests.The conclusion is that most programs and approaches work very well and give

M. Breger 5

correct as well as complete answers when used by cautious and experiencedastronomers. For Delta Scuti stars, the Merate and Vienna astronomers almostroutinely get the same answers with their different programs. Only a few ap-proaches (may I mention CLEAN or earlier versions of Maximum Entropy?) arecontroversial.

We hope you find PERIOD04 for WINDOWS, LINUX and Mac OS X useful.Please give us your suggestions for PERIOD08.

Michel BregerEditor


Vol. 146, 2005

New variable and multiple stars in the lower part of theCepheid instability strip

Y. Fremat1, P. Lampens1, P. Van Cauteren2 and C.W. Robertson3

1 Royal Observatory of Belgium, 3 Ringlaan, 1180 Brussels, Belgium2 Beersel Hills Observatory (BHO), Beersel, Belgium

3 SETEC Observatory, Goddard, Kansas, USA

Abstract

We obtained high-resolution spectra for 33 A-type stars during 4 consecutivenights. All our targets are Hipparcos program stars located in the lower partof the Cepheid instability strip which show hints for variability in radial velocityof a yet unidentified nature.

CCD photometry was acquired for some of the most promising candidatesand used to further interpret the variability of several targets. A first result isthe discovery of two new pulsating variable stars.

In addition, five new binary or multiple systems were identified (one SB1,three SB2 and one SB2 triple system).

1. Introduction

A very interesting region of the Hertzsprung-Russel diagram lies at the inter-section of the main sequence and the classical Cepheid instability strip, wherea variety of phenomena are at play in the stellar atmospheres.

These phenomena include magnetism, diffusion, rotation and convection,which are active in δ Scuti, SX Phe, γ Dor and roAp variable stars. Thelatter processes may boost or, on the contrary, inhibit the presence of chemicalpeculiarities (occurring in Ap, Am, ρ Puppis and λ Boo stars). The competitionbetween these processes and mechanisms thus leads to a large mix of stellargroups of different atmospheric composition which also behave in different wayswith respect to pulsation and binarity. To address the issue of the interactionsbetween chemical composition, pulsation and multiplicity, we aim to perform a

Y. Fremat, P. Lampens, P. Van Cauteren and C.W. Robertson 7

systematic study of the chemical composition of a sample of poorly known main-sequence A- and F-type stars in this region. For this purpose, high-resolutionspectroscopic observations were carried out at the Observatoire de Haute-Provence (OHP, France) in December 2004. We here give a brief account ofthe present variability status in the observed sample.

2. Description of the sample

We selected 38 targets from the Hipparcos catalogue. The following criteriahave been adopted : 1) brightness larger than magnitude 8; 2) spectral typeranging from A0 to F2; 3) showing some indication of radial velocity variabilityin the catalogue of Grenier et al. (1999) and 4) having less than 10 referencesin the bibliography recorded in the SIMBAD data base (CDS). Except for theHipparcos epoch photometry and because of the use selection criteria, verylittle is known about the targets. The distribution in spectral type of the sampleof 33 observed stars is plotted in Fig. 1 (based on Grenier et al. 1999).

A2 A3 A4 A5 A6 A7 A8 A9 F2Spectral Type

0

3

6

9

12

15

Num

ber o

f Sta

rs

Figure 1: Distribution in spectral type of the observed sample.

3. Observations

The spectroscopic observations were carried out at the 1.93 m telescope equippedwith the ELODIE spectrograph of the OHP (Baranne et al. 1996). High-resolution spectra were collected during 4 nights in 2004 (December 3–7). Eachtarget was observed 2 to 5 times in order to be able to detect rapid (periodsof order of a few hours) or slow (periods of order of a few days) line profile

8 New variable and multiple stars

variations (LPVs) and/or changes in radial velocity. However, due to circum-stances (bad weather conditions) some of our targets were only observed 2–3times successively, without the ability to reobserve at a later date. We adaptedthe time exposures to ensure a S/N ratio per pixel (at 5000 A) generally varyingfrom 80 to 100. The data were automatically reduced order by order at the endof the night using the INTERTACOS pipeline. INTERTACOS was also used toperform a cross-correlation with the F0 mask after each exposure.

For 4 promising candidates, we also acquired complementary CCD photom-etry in the period between December 2004 and February 2005. These observa-tions were carried out using a 40cm Newton equipped with a SBIG ST10XMEcamera (2 targets) at BHO, a 30cm telescope with SBIG ST8i camera (1 tar-get) at SETEC Observatory (Kansas, USA) and a 25 cm telescope with a SBIGST10XME camera (1 target) at BHO on one more occasion. Depending on thetarget’s magnitude, a B or V filter according to Bessell’s specifications (Bessell1995) was used.

-200 -100 0 100 200Radial Velocity (km/s)

0.93

0.96

0.99

1.02

Nor

mal

ized

Inte

nsity

3.4 3.5 3.6 3.7JD - 2453380

-1.92

-1.90

-1.88

-1.86R

elat

ive

B M

agni

tude

-200 -100 0 100 200Radial Velocity (km/s)

0.95

0.98

1.00

Nor

mal

ized

Inte

nsity

9.2 9.3 9.4 9.5 9.6JD - 2453359

-1.38

-1.35

-1.32

-1.29

-1.26

Rel

ativ

e B

Mag

nitu

de

HIP 40361

HIP 113790

Figure 2: The left panels show some cross-correlation functions obtained during con-secutive exposures while the right panels illustrate the corresponding light curves.

4. Detection of pulsation and/or multiplicity

The detection of the occurrence of pulsation and/or multiplicity is done by avisual inspection of the shape and the variations of the cross-correlation func-tions (CCFs) for each target. The identification of several peaks in the CCFs

Y. Fremat, P. Lampens, P. Van Cauteren and C.W. Robertson 9

and/or the existence of day-to-day changes in radial velocity are interpreted asthe signature of binarity or multiplicity, while the appearance and disappearanceof bumps in the peaks of the CCFs collected during two consecutive exposuresare interpreted as the signature of intrinsic variability.

We classified the observed targets according to their likelihood to be effec-tively intrinsically variable. To this purpose four classes were defined (VAR1,VAR2, VAR3, VAR4) among which the class VAR1 indicates a large probabil-ity. All VAR1 candidates were subsequently submitted to at least two (short)photometric runs in order to verify whether they also show short-period lightvariations (see new variable stars in Table 1). Fig. 1 illustrates the procedure asapplied to HIP 40361 and HIP 113790. Note that a simple frequency analysisof the Hipparcos epoch photometric data confirms the listed main periodicityfor HIP 113790.

Table 1: Newly detected photometric-spectroscopic variable stars as well as mul-tiple systems.

HIP HD Mag V Sp. Type Notes

New variable starsconfirmed by CCD photometry

9501 12389 7.99 A4V already known δ Scuti star40361 68725 6.94 F2 P ∼ 0.12 d

113790 217860 7.30 A8III P ∼ 0.05 d, multiperiodicunresolved Hipparcos variable

Newly detected multiple systems

3227 3777 7.44 A2III SB28581 11190 7.88 A2III SB2

46642 81995 7.35 A5 SB1115200 219989 7.35 A2IV SB2116321 221774 7.38 A4IV SB2

5. Conclusion

Among the 33 observed stars of our sample, 9 were found to be spectroscopicallyvariable on a timescale of several hours and are thus promising candidates forthe detection of pulsation(s). We classified these 9 candidates into two classesVAR1 and VAR2. For 4 of these targets CCD light curves in the B or V passband

10 New variable and multiple stars

were obtained. The new photometric data confirm the variable nature of 3 stars,one of which is the already known δ Scuti star HIP 9501 (Schutt 1991) andanother one is an unresolved Hipparcos variable star (ESA 1997). Five newbinary or multiple systems were further identified (one SB1, three SB2 and onetriple system). We are presently analyzing and exploiting the ELODIE spectrain order to determine fundamental parameters for each observed target. Theseresults will be published in more details in the near future. We further planto organize a large photometric campaign next season in order to monitor thelight variations of the new intrinsic variable stars HIP 40361 and HIP 113790.

Acknowledgments. This work is based on spectroscopic observationsmade at the Observatoire de Haute-Provence (France) and on CCD photo-metric observations collected at Beersel Hills and SETEC observatories. Partof the photometric data were acquired with equipment purchased thanks to aresearch fund financed by the Belgian National Lottery (1999). Ample use wasmade of the SIMBAD data base (Centre de Donnees Stellaires, Strasbourg).

References

Baranne, A., Queloz, D., Mayor, M., Adrianzyk, G., Knispel, G., Kohler, D., Lacroix,D., Meunier, J.-P., Rimbaud, G., Vin, A. 1996, A&A Supp. Ser., 119, 373

Bessell, M.S. 1995, CCD Astronomy 2, No. 4, 20ESA 1997, The Hipparcos and Tycho Catalogues ESA-SP 1200Grenier, S., Burnage, R., Faraggiana, R., Gerbaldi, M., Delmas, F., Gomez, A. E.,

Sabas, V., Sharif, L. 1999, A&A Supp. Ser., 135, 503Schutt R.L. 1991, AJ, 101, 2177


Vol. 146, 2005

OV And, a new field RRab Blazhko star?

K. Kolenberg1, E. Guggenberger1, P. Lenz1, P. Van Cauteren2, P. Lampens3,P. Wils4

1 Institut fur Astronomie, Turkenschanzstr. 17, 1180 Vienna, Austria2 Beersel Hills Observatory, Belgium

3 Koninklijke Sterrenwacht van Belgie, Ringlaan 3, 1180 Ukkel, Belgium4 Vereniging voor Sterrenkunde (VVS), Belgium

Abstract

Between 2004 November and 2005 January, we organized a small photometricobserving campaign dedicated to the RRab star OV And. This star lies veryclose to other stars in the field. NSVS (Northern Sky Variability Survey) datasuggested the presence of the Blazhko effect in this star. The new data revealthat the large discrepancies in pulsation amplitude of the star reported in theliterature can be reproduced by applying different reduction methods. We showthat only the data reduced with inclusion of the neighboring stars can be con-sidered trustworthy. Nevertheless, we find some indications that the Blazhkoeffect may exist in this star.

1. Introduction

The most intriguing subclass of the astrophysically important RR Lyrae starsconsists of stars showing the Blazhko effect, the phenomenon of amplitudeor phase modulation. These stars have light curves that are modulated ontimescales of typically tens to hundreds of days. Blazhko (1907) was the firstto report this phenomenon in RW Dra. The estimated incidence rate of Blazhkovariables among the galactic RRab stars (fundamental mode pulsators) is about20-30 % (Szeidl 1988; Moskalik & Poretti 2002).

Though the Blazhko effect was discovered nearly a century ago, its physicalorigin still remains a mystery. Two groups of possible models are proposed,the magnetic models and the resonance models, but neither of them provides asufficient explanation for the variety of behavior observed in Blazhko stars (seealso Kolenberg 2004 for a concise general overview).

12 OV And, a new field RRab Blazhko star?

The number of brighter Blazhko stars in the field is limited, and most ofthem have been known for a long time. For some, extensive photometric studieshave been published, often using data sets gathered over several decades (e.g.,Borkowski 1980, Kovacs 1995, Szeidl & Kollath 2000, Smith et al. 2003, Jurcsiket al. 2005). Long-term photometric surveys, often serving other aims thanstudying only stellar variability (e.g., microlensing surveys like MACHO - Alcocket al. 2003, and OGLE - Moskalik & Poretti 2002), have entailed the detectionof a multitude of new variable stars, among which also RR Lyrae Blazhkostars. As a by-product of the Robotic Optical Transient Search Experiment(ROTSE), the Northern Sky Variability Survey (NSVS) was established, offeringa temporal record of the sky over the optical magnitude range from 8 to 15.5.This offers data for detecting new RR Lyrae Blazhko stars in the field. One ofsuch suspected Blazhko candidates is OV And.

2. The target star OV And

In the sky plane the RR Lyrae star OV And (NSV 134, RA2000: 00:20:44.42,Dec2000: 40:49:41.6, V=10.4-11.0 according to GCVS 2003) lies close to afew other stars (Rossiger 1987), which makes it difficult to resolve with theclassical photometric technique involving aperture photometry. Together withits ’companion’ of almost equal brightness (at an angular separation of 7.5arcsec) the star was previously denoted as BD+40 0060 = NSV 134 (Nikulina1959) and was supposed to be a rapid irregular variable in the New Catalogueof Suspected Variable Stars (Kukarkin & Kholopov 1982). Photometric datafrom 1985-1986 revealed a light curve amplitude of about 0.5 mag, and led tothe conclusion that one of the stars must be an RR Lyrae star of type ’ab’,meaning that it pulsates in the fundamental radial mode (Rossiger 1987).

From CCD images obtained by Prosser (1988) it was discovered that infact there are three other fainter stars located near the variable and its bright’companion’. High-dispersion spectra obtained by the same author revealed alarge difference in radial velocities between OV And (about -194 km/s) and itsbright ’companion’ (-62 km/sec), indicating that the two brighter stars do notform a physical pair. In his photometric data, which allowed resolution of OVAnd and its ’companion’, no variability was detected in the ’companion’ nor inany of the other neighboring stars. Prosser’s differential observations (featuring‘variable minus ’companion’ magnitude’) point out an amplitude of 0.980 mag(σ = 0.004 mag) in the star’s light variation, essentially twice the amplitudefound by Rossiger. According to Prosser this “could mean that the adoptedcomparison star, BD+40 0056, may have some variation in brightness as well”.Neither author mentioned a Blazhko effect.

K. Kolenberg et al. 13

2.1 NSVS data of OV And

The NSVS data of OV And were gathered by ROTSE-I in 1999-2000, and covera total time span of 254.8 days.

A frequency analysis of the 155 NSVS measurements with Period04 (seeLenz & Breger 2005, this volume) yields the main frequency f to the expectedaccuracy. We adopt the period P = 0.470581 d (frequency f = 2.12503c/d) listed in the GEOS RR Lyrae database and determined from previousmeasurements.

Figure 1: NSVS data phased with the main frequency f = 2.12504 c/d. For visibilitywe chose phase 0.5 to be in the middle of the rising branch. Is the large scatter partlydue to a Blazhko effect?

In all our frequency analyses we adopt a 4σ significance criterion, σ being themean noise level in the Fourier periodogram at the relevant frequency (Breger etal 1993). After prewhitening with the known main frequency and its significantfirst 5 harmonics, a new frequency peak is found close to the main frequency,however only at a level of 3.5 σ. Such closely-spaced frequency peaks are atypical feature of Blazhko stars. However, due to its low signal-to-noise ratio,we are not (yet) convinced that this peak is related to a real frequency.

As the pixel size of the CCD used for the NSVS corresponds to about14 arcsec on the sky, the ROTSE-I cameras will always have measured thecombined magnitude of all five stars in the small group, resulting in a smalleramplitude. Hence, the Blazhko effect should become harder to detect. Thelight curve folded with the main period (Figure 1) shows a large scatter. Theaverage point to point photometric scatter of the NSVS data is reported to be±0.02 mag for bright, unsaturated stars (Wozniak et al. 2004).


Table 1: Log of the photometric observations of OV And. BHO: Beersel Hills Obser-vatory – vlt: Vienna Little Telescope.

Observatory Detector Reduction Software ObserversBHO SBIG ST10XME Mira AP (1) PVC, PLavlt Apogee AP9E MaxIm DL (2), IRAF (3) KK, EG, PLe(1) Produced by Axiom Research Inc. (2) Produced by Diffraction Limited.

(3) Image Reduction and Analysis Facility,

written and supported at the National Optical Astronomy Observatories (NOAO).

Date/HJD Telescope Observatory Filters Amount of data= 245 0000 Data Hours19.11.04/3329 25-cm BHO V 118 4.08924.11.04/3334 25-cm BHO V 32 1.07824.11.04/3334 80-cm vlt V 75 2.53925.11.04/3335 80-cm vlt B, V 77, 80 7.576, 7.64903.12.04/3343 40-cm BHO V 108 3.15620.12.04/3360 80-cm vlt B, V 91, 91 4.937, 4.93721.12.04/3361 80-cm vlt B 150 3.90111.01.05/3382 80-cm vlt B, V 38, 35 3.788, 3.82917.01.05/3388 80-cm vlt B, V 18, 21 2.627, 2.586Total time span: B 53.1 d, V 59.1 d.

We decided to start a photometric monitoring of OV And in order to checkwhether a Blazhko effect is indeed present.

3. Photometry of OV And

3.1 Observations

Photometric CCD observations were gathered from two different sites: the80-cm telescope at the Vienna Observatory (Austria) yielded Johnson B andV data, and the 25-cm and 40-cm telescopes at Beersel Hills Observatory(Belgium) provided data in the filter V (following Bessell’s specifications, Bessell1995) only.

A log of the observations is given in Table 1. We used GSC 2787 1798as the comparison star and GSC 2787 1777 as the check star. The errors onindividual data points varied between 0.010 mag and 0.030 mag as obtained


Table 2: Observed times of maximum (in Johnson B and V filters), and correspondingO-C values for the elements T0 = HJD24539026.478 and P = 0.470581 d from theGEOS RR Lyrae database. Errors in HJD(Max) are of order 0.001 d.

HJD(Max) O-C (±0.0014 d)245 3329.306 -0.0105245 3334.476 -0.0169245 3335.418 -0.0161245 3360.358 -0.0169245 3361.300 -0.0160

from the rms error on the differential magnitudes of the comparison star andthe check star.

3.2 Analysis of the data

On the CCD frames obtained with the 25-cm and 40-cm telescopes it wasimpossible to resolve OV And and its ’companion’, and hence the reduceddata represent the combined light. For the CCD frames obtained in Viennawe carried out the reductions twice: once with an aperture radius of 7 pixels(i.e., about 0.5-0.8*FWHM depending on seeing conditions) which allowed to(barely) separate both stars and once with an aperture radius of 19 pixels (about3.5*FWHM) including both stars.

Rossiger (1987) found the following elements from his observations:

HJD(Max) = 2446764.241(±0.001)+ 0.470568(±0.000002)E (d), (1)

confirmed by Prosser (1988) based on data taken one year later. AdoptingRossiger’s period, our observed times of maximum do no longer correspondwith his time T0 of maximum light. Adopting the period P=0.470581 d listedin the GEOS database, we still obtain a systematic shift in the (O-C) values(Table 2). As an updated ephemeris we propose:

HJD(Max) = 2453335.418(±0.001)+ 0.470580(±0.000001)E (d). (2)

3.2.1 How efficiently can the stars be resolved in the reduction?

For the night with the best seeing (JD 245 3361), we were able to separate OVAnd from its ’companions’, as illustrated in Fig. 3 displaying the ’OV And group’on representative frames for two different nights. The bright ’companion’ starindeed proved to be constant, as was already reported by Prosser (1988).


Figure 2: OV And (star to the E) and its immediate neighborhood on representativeframes obtained at the 80-cm telescope in Vienna for HJD 2453361 (left) and HJD2453335 (right). This figure demonstrates the effect of variable seeing. (North is up,East is left.)

For all other nights, the differential data of the ’companion’ star showedvariations in brightness. However, we found evidence that these variations arelinked to those seen in the OV And data and therefore must be caused bysome contribution of the light of OV And itself. In the same way, the variablecontribution of the ’companion(s)’ in the ’separated’ OV And data (due tovariable seeing conditions) may not be neglected, as already indicated by thelarger scatter in the data, and hence these reduced data have to be interpretedwith caution!

The peak-to-peak B amplitude for the night of JD 2453361 amounts to1.65 ± 0.03 mag. As we only obtained B data during this night, we can makeno statements on the V magnitude of OV And (without additional light).

3.2.2 Is OV And a Blazhko star?

Blazhko stars are typified by the occurrence of secondary frequencies closeto the main pulsation frequency and its harmonics. These frequencies arethought to be related to nonradial modes in the mainly radially pulsating RRLyrae star, modes that, according to the currently plausible models, would causethe observed amplitude and/or phase modulation.

A period analysis of the OV And B and V data, respectively with its ’com-panions’ and tentatively separated, and using Period04, always revealed themain frequency f and its harmonics, up to high (8th) order. However, we didnot find any evidence of a secondary frequency close to the main pulsationfrequency above the 4σ significance level.

The main feature of the Blazhko effect is the variation of the light curveshape and amplitude. Table 3(a) displays the total amplitude of the lightcurvein different nights. From these data we cannot unambiguously conclude on thepresence of a Blazhko effect in OV And yet; at the level of 2σ the amplitudes


Table 3: Variation of the (peak-to-peak) light curve amplitude in the Johnson B andV filter. The data in the upper panel are for OV And with its ’companions’; the lowerpanel data for the reduced data in which we attempted to separate the light variationof OV And. Note that the upper panel reproduces Rossiger’s (1987) values quite well,whereas the lower panel agrees better with Prosser’s (1988) values. However, we callfor caution in interpreting the amplitudes in the lower panel!

(a) OV And and ’companions’

Date AV (mag) ∆AV AB (mag) ∆AB

19.11.04 0.476 0.01524.11.04 0.463 0.01525.11.04 0.458 0.015 0.865 0.01520.12.04 0.460 0.015 0.894 0.01521.12.04 0.893 0.015

(b) OV And ’separated’

Date AV (mag) ∆AV AB (mag) ∆AB

24.11.04 1.015 0.02825.11.04 1.053 0.057 1.505 0.02820.12.04 1.179 0.057 1.596 0.02821.12.04 1.179 0.028 1.655 0.028

are stable. Though there may be an indication for a variation of the totalamplitude as suggested by the data in Table 3(b), we call for much cautionat this stage. We indeed cannot trust the values from Table 3(b) as it is verydifficult to obtain ”clean” measurements of OV And without influence of the’companion”s light (and other neighboring stars)!

4. Conclusions

• Like previous authors who used small telescopes, we experienced difficul-ties separating OV And from its neighbors on the sky. Reduction of dataof OV And should include all neighboring stars in order to be trustworthy.

• The smaller amplitude reported by Rossiger (1987), about half of theamplitude detected by Prosser (1988), can be explained by the fact thatRossiger included the bright ’companion’ in his analysis, rather than dueto the comparison star’s behavior, as Prosser (1988) assumed. Additionalstar light will reduce the total amplitude of the light curve, and alsoinfluences its shape. The NSVS data show an even smaller amplitude.The main reason for this, besides the inclusion of yet another star in thereduction, is that the ROTSE camera (unfiltered CCD) is more sensitive


Figure 3: All our V data phased with the main frequency. OV And was reduced withits ’companions’ included in the aperture. The amplitudes of the lightcurve are listedin Table 3(a) (see columns for V ). The data suggest a small change in the light curveshape and amplitude, but its significance needs to be proven.

in the red, the wavelength region where RR Lyrae stars emit less lightthan in V .

• For the ’isolated’ OV And, we detected a light curve amplitude of 1.6±0.1mag in the B filter. The light curve in the V filter is estimated to be1.1± 0.1 mag.

• The present data don’t confirm the additional frequency suggested by theNSVS data, but since we only observed 4 maxima we cannot disprove itsreality yet. Moreover, our observed light curves suggest that a Blazhkoeffect may be present. We plan a follow-up action on OV And for nextautumn.

5. The Blazhko Project

These observations were carried out in the framework of the Blazhko Project,an international collaboration, set up to join efforts in obtaining a better un-derstanding of the Blazhko phenomenon in RR Lyrae stars. The project wasfounded in Vienna and started its activities in the autumn of 2003. The startingpoint for improving the modeling is an extensive data set, consisting of spec-troscopy and photometry, of a limited sample of field RR Lyrae Blazhko stars,and also some non-Blazhko stars for comparison. Besides the larger campaigns,


some small photometric campaigns – like this one – are organized to fine-tuneour knowledge of the frequency behavior of field Blazhko stars.

All interested colleagues – professionals and amateurs – are warmly invitedto join in the observational campaigns of the Blazhko Project. For more infor-mation we refer to the website dedicated to the Blazhko Project(http://www.astro.univie.ac.at/˜blazhko/)or directly contact [email protected].

Acknowledgments. The Blazhko Project, KK and EG are supportedby the Austrian Fonds zur Forderung der wissenschaftlichen Forschung, projectnumber P17097-N02. KK thanks Jacqueline Vandenbroere and Anton Paschkefor information on the data reported on the GEOS and BAV websites.

The data were partly gathered in the framework of the Observatoriums-praktikum, during which students at the Institute for Astronomy are taught touse the telescope and CCD camera (see http://www.deltascuti.net/obsprak/and http://www.astro.univie.ac.at/˜blazhko/Winter.html). We thank allthe students who participated in the observations of OV And: Paul Beck, DanielBlaschke, Johannes Puschnig, Markus Hareter, Eveline Glassner, Pia Hecht,Elisabeth Fullenhals, Sophie Aspock, Vera Steinecker, Gerald Jungwirth, HannsPetsch, Jenny Feige, Adrian Partl, Mathias Jager, Petra Zippe, Marius Halosar,Florian Herzele. Part of the data was acquired with equipment purchased thanksto a research fund financed by the Belgian National Lottery (1999).

References

Alcock, C., Allsman, R.A., Becker, A., et al., ApJ 598, 597Bessell, M.S., 1995, CCD Astronomy 2, No. 4, 20

Blazhko, S., 1907, Astron. Nachr. 175, 325Borkowski, K.J., 1980, Acta Astron. 30, 393

Breger, M., Stich, J., Garrido, R., et al., 1993, A&A 271, 482GEOS RR Lyrae database: http://webast.ast.obs-mip.fr/people/leborgne/dbRR/

Kolenberg, K., 2004, CoAst 145, 16Kukarkin, B.V., Kholopov, P.N., 1982, New Catalogue of Suspected Variable Stars

Jurcsik, J., Sodor, A., V’aradi, M., et al., 2005, A&A 430, 1049Kovacs, G., 1995, A&A 295, 693

Lenz, P., & Breger, M., 2005, CoAst 146, 53Moskalik, P., & Poretti, E., 2002, A&A398, 213

Nikulina, T.G., 1959, Astron. Tsirk., 207, 14NSVS, Northern Sky Variability Survey, http://skydot.lanl.gov/nsvs/nsvs.phpProsser, C.F., 1988, Inform. Bull. Var. Stars, 3130, 1

Rossiger, S., 1987, Inform. Bull. Var. Stars, 2977, 1ROTSE, Robotic Optical Transient Search Experiment,

http://rotse1.physics.lsa.umich.edu/


Szeidl, B., 1988, in ’Multimodal Stellar Pulsations’, Kultura, eds. Kovacs, G.,Szabados, L., Szeidl, B., 45

Szeidl, B., Kollath, Z., 2000, ASP Conf. Ser. 203, 281Smith, H.A., Church, J.A., Fournier, J., et al., 2003, PASP 115, 43Wozniak, P. R., Vestrand, W. T., Akerlof, C. W., et al., 2004, The Astronomical

Journal, Vol. 127, 2436


Vol. 146, 2005

Search for intrinsic variable stars in three open clusters:

NGC 1664, NGC 6811 and NGC 7209

P. Van Cauteren1, P. Lampens2, C.W. Robertson3 and A. Strigachev4,2

1 Beersel Hills Observatory, Beersel, Belgium2 Royal Observatory of Belgium, 3 Ringlaan, 1180 Brussels, Belgium

3 SETEC Observatory, Goddard, Kansas, USA4 Institute of Astronomy, Bulgarian Academy of Sciences, Sofia, Bulgaria

Abstract

We report on new time-series CCD observations for three poorly studied openclusters from the STACC list. The collected light curves in the V filter of alarge number of stars per field have been examined in detail in search for short-periodic variable stars. In particular we looked for new δ Scuti-type candidates.First results for the open clusters NGC 1664, NGC 6811 and NGC 7209 arepresented.

1. Why study open clusters ?

The study of variable stars in open clusters presents various advantages for theunderstanding of pulsation physics:

• open clusters with variable members provide comparative asteroseismo-logical tests of stellar structure and evolution based on the implicationsof a common origin and formation of their stellar members;

• open clusters are ideally suited for performing high-accuracy differentialphotometry as well as ’CCD ensemble photometry’ based on a large sam-ple of comparison stars in the field-of-view.

Such studies moreover require long baseline differential photometry (to detectmultiple frequencies combined with small amplitudes) as well as complemen-tary multi-color photometry or high-resolution spectroscopy (to determine thebasic stellar parameters) if one wishes to take full profit of these observations

22 Search for intrinsic variable stars in three open clusters

(Freyhammer et al. 2001). We selected three poorly studied open clustersfrom the STACC list (Frandsen & Arentoft 1998) in order to perform a CCDsurvey using small telescopes (up to 1.3m) with the goal to search for intrinsicvariable stars. These open clusters are visible from the Northern hemisphereand are furthermore prime targets for the detection of δ Scuti-type candidatestars (Frandsen & Arentoft 1998).

On the other side, the presence of close neighboring stars in the (semi)-crowded fields of the central parts of open clusters may affect the quality ofthe results and this could imply the need for a different reduction technique(point-spread function fitting instead of the aperture photometry approach).

2. General information about the observed clusters

2.1 NGC 1664

Is an extended open cluster of the STACC list with a rich central region (Frand-sen & Arentoft 1998). A recent proper motion study was performed by Baum-gardt et al. (2000). The cluster membership was also investigated by Missana& Missana (1998). It is situated at a distance of about 1200 pc and has alog age of 8.47 (Mermilliod 1995).

2.2 NGC 6811

Is an open cluster of the STACC list with many candidates in the instabilitystrip (Frandsen & Arentoft 1998). A proper motion study was performed bySanders (1971), yielding 97 probable members. Glushkova et al. (1999) col-lected radial velocities in the field and identified seven new members. Fromtheir photoelectric photometric data they obtained a mean distance of 1040 pcand an age of 0.7 Gyr based on 88 identified cluster members (log age = 8.85).

2.3 NGC 7209

Is an open cluster of the STACC list with a poorly populated color-magnitudediagram (Frandsen & Arentoft 1998). Several studies of proper motions inthe cluster field exist: for example, Platais (1991) found 148 possible clustermembers down to mag 15 while recent proper motion studies were performed byBaumgardt et al. (2000) and by Dias et al. (2001) based on the TYCHO-2 cata-logue. Pena & Peniche (1994) proposed the existence of two clusters, NGC 7209a and b, with a different reddening, age and distance based on Stromgren pho-tometry of 54 stars in the cluster direction. Vancevicius et al. (1997) obtainedphotoelectric photometry in the Vilnius system for 96 probable members but donot support that conclusion: they estimated the mean reddening, the metalicity

P. Van Cauteren, P. Lampens, C.W. Robertson and A. Strigachev 23

and derived a median distance of 1026 ± 30 pc and an age of 0.45 Gyr (logage = 8.65).

3. Observations and photometric reduction

The logbook of all observations, the telescopes and CCD cameras used arelisted in Table 1. Except for the observatory Hoher List (HL) where a self-built camera was employed (2Kx2K; HoLiCam of the Observatorium HoherList, University of Bonn, Germany), the other cameras are from Photometrics(1Kx1K; Skinakas, Crete) or SBIG (various ST’s), equipping instruments withdifferent focal lengths. We typically made use of a field-of-view with a size of26′ × 28′ (e.g. BHO 40cm) or 34′ × 23′ (e.g. BHO 25cm).

Table 1: Log of the cluster observations

NGC Obs/Tel CCD Start End Nr/hrs Filter

1664 BHO40 ST10E 27-10-03 10-12-03 25.8 B,VBHO25 ST7 08-12-03 08-12-03 8.9 VHL100 HoLiCam 17-12-03 17-12-03 5.5 V

6811 BHO40 ST10E 22-03-03 27-09-03 31.8 B,VNAO70 ST8 03-05-03 03-05-03 3.3 V

7209 BHO40 ST10E 29-07-02 20-08-03 25.9 B,VBHO25 ST7 01-08-03 10-08-03 31.9 VSETEC ST8i 12-08-02 05-09-02 42.6 VSKI130 CH360 19-08-03 04-10-03 9.2 V

All the frames were treated following the classical basic reduction scheme:a median dark/bias was subtracted from each science frame and these werethen divided by a median flat-field. Aperture photometry was subsequentlyperformed using various packages. Consequently, we intentionally avoided thestars with close neighbors on the collected images. First, for all the frames ob-tained at BHO, HL, SETEC and Skinakas, the aperture photometric package ofMIRA AP was applied. Differential magnitudes with respect to a pre-selectedcomparison star were produced. For the frames acquired during one night in

0The MIRA AP software is produced by Axiom Research Inc.


2003 with the Schmidt 50/70 telescope of the National Astronomical Observa-tory of Bulgaria (Tsvetkov et al. 1987), we gathered the differential data usingthe MOMF code (Kjeldsen & Frandsen 1992). In a second step, CCD ensemblephotometry was applied in order to produce differential magnitudes with respectto a comparison magnitude using a(n) (inhomogeneous) set of comparison stars(Honeycutt 1992). The figures shown below are based on the latter data.

A V filter following Bessel’s (1995) specifications was used for the time-seriesphotometry. In addition, some nights were dedicated to the acquisition of theB-magnitudes with the aim to produce updated color-magnitude diagrams inthe near future.

4. Results

4.1 NGC 1664

4.1.1 Analysis

The observed field of NGC 1664 is illustrated by Fig. 1. We used the numberingof the (WE)BDA (Mermilliod 1995; cf. http://www.obswww.unige.ch/webda/)database for identification purposes. 9 stars received an internal numbering.284 stars were included in the reduction scheme and examined for rapid lightvariations. The CCD data consist of more than 300 differential magnitudes inthe mean for most targets, with actual numbers varying between 39 (a few)and 516 (many) depending on the position in the field.

4.1.2 Discussion of individual variable stars

We report the detection of two new short-periodic variable stars in the field:V1 and V2. Both are possible δ Scuti variable stars. A period analysis of theV-band data for star #100 = V1 indicates a main frequency of ≈ 4.48 c/dand a semi-amplitude of 0.04 mag in V. Star #265 = V2 is probably a newδ Sct star with a main frequency of ≈ 16.77 c/d and a semi-amplitude of 0.02mag only in V. The light curves of two individual nights are presented in Fig. 2.

4.2 NGC 6811

4.2.1 Analysis

The observed field of NGC 6811 is illustrated by Fig. 3. We used the Lindoff-numbering for identification purposes (same as in (WE)BDA). 230 stars receivedan internal numbering. 441 stars were included in the reduction scheme and


examined for rapid light variations. The CCD data consist of more than 200 dif-ferential magnitudes in the mean for most targets with actual numbers varyingbetween 56 (a few) and 355 (many) depending on the position in the field.


We report the detection of nine new variable stars in the field, of which sixare probably short-periodic variable stars of type δ Scuti and one is a sus-pected short-period variable. The light curves of the variable stars detected inNGC6811 are presented in Fig. 4. Stars #18 = V1 and #37 = V2 are two δ Scttype variables, with a main frequency of about 20 c/d and a semi-amplitude ofthe order of 0.01 mag in V. Star #70 = V3 is a δ Sct type variable with a mainfrequency of about 7.5 c/d and a semi-amplitude of 0.02 mag in V. The dataof star #39 = V4 indicate a frequency of about 8.4 c/d with a semi-amplitudeof 0.01-0.02 mag in V. Star #113 = V5 is a suspected variable of a similar typewith a possible frequency of 13 c/d of much smaller amplitude (< 0.01 mag).Two δ Scuti variable stars were also detected among the new designations:these are stars #489 = V6 and #491 = V7 (following the BDA-numbering)with frequencies of ≈ 9.4 c/d and possibly 15 c/d and semi-amplitudes of 0.04and 0.02 mag in V respectively. Two more variable stars of unknown type (V8and V9) were discovered at the field edges (Fig. 5, left): they were only ob-served very shortly (for about 3 hrs) and do not show the characteristics typicalof δ Scuti stars as their periods are longer. The right panel of Fig. 5 shows theobserved light variations with a large amplitude in both cases.

4.3 NGC 7209

4.3.1 Analysis

The observed field of NGC 7209 is shown in Fig. 6. We used the Mavers-numbering for identification purposes (same as in (WE)BDA). 215 stars receivedan internal numbering. 357 stars were included in the reduction scheme andexamined for rapid light variations. The CCD data consist of more than 900differential magnitudes for most targets with actual numbers varying between70 (a few) and 1100 (many) depending on the position in the field.


Although we did not (yet) systematically inspect all the material for this cluster,we already report the detection of three possible δ Scuti variable stars in thefield: stars #24 = V1 (= GSC 3605 2253), #55 = V2 (= GSC 3605 2309)and #59 = V3 (= GSC 3605 2247 1). Stars #24 and #55 are probably δ Scuti


stars with rather small semi-amplitudes (< 0.01 mag). The data of star #59clearly indicate a frequency of 7.7 c/d with a semi-amplitude of 0.01 mag in V.The observed light variations of the detected variabe stars in NGC7209 are plot-ted in Fig. 7. In addition, star #112 = V4 (= GSC 3605 2721 = SAO 51648),an infra-red source, appears to show small but irregular fluctuations.

We also noted the existence of slower fluctuations with observed amplitudesof the order of 0.1-0.2 mag in the V magnitudes of a few stars which were how-ever only occasionally observed: #91 (2 nights showing a (start of a) minimumcompared to the rest of the data), #122 (2 nights, ∆V = 0.3 mag), #179(1 night, ∆V = 0.6 mag) and #3612 (1 night, ∆V = 0.6 mag) (using theBDA-numbering).

5. Figures for three open clusters

Figure 1: Sample CCD frame of the open cluster NGC 1664. The newly detectedvariable stars are indicated.


940.4 940.6-0.56

-0.54

-0.52

-0.50

-0.48

-0.46

980.4 980.6Julian date ( 2 452 000+)

-0.56

-0.54

-0.52

-0.50

-0.48

-0.46

∆ m

ag (#

100

- com

p)

940.4 940.6

0.04

0.06

0.08

0.10

0.12

980.2 980.4 980.6Julian date ( 2 452 000+)

0.04

0.06

0.08

0.10

0.12

∆ m

ag (#

265

- com

p)

Figure 2: Light curves of NGC1664. Left panels: star #100. Right panels: star #265.

Figure 3: Sample CCD frame of the open cluster NGC 6811. The newly detectedvariable stars are indicated


747.2 747.4 747.6

0.14

0.16

0.18

0.20

0.22

0.24

879.4 879.6Julian date ( 2 452 000+)

0.14

0.16

0.18

0.20

0.22

0.24

∆ m

ag (#

18 -

com

p)

747.4 747.6

-0.84

-0.82

-0.80

-0.78

-0.76

-0.74

879.4 879.6Julian date ( 2 452 000+)

-0.84

-0.82

-0.80

-0.78

-0.76

-0.74

∆ m

ag (#

37 -

com

p)

747.4 747.6

-1.12

-1.10

-1.08

-1.06

-1.04

-1.02

862.4 862.6Julian date ( 2 452 000+)

-1.12

-1.10

-1.08

-1.06

-1.04

-1.02

∆ m

ag (#

70 -

com

p)

860.4 860.6

-0.48

-0.44

-0.40

-0.36

-0.32

879.4 879.6Julian date ( 2 452 000+)

-0.48

-0.44

-0.40

-0.36

∆ m

ag (#

39 -

com

p)

722.6 722.8

-0.46

-0.44

-0.42

-0.40

-0.38

747.4 747.6Julian date ( 2 452 000+)

-0.46

-0.44

-0.42

-0.40

-0.38

∆ m

ag (#

113

- com

p)

721.6 721.8

-0.92

-0.88

-0.84

-0.80

860.4 860.6Julian date ( 2 452 000+)

-0.92

-0.88

-0.84

-0.80

∆ m

ag (#

489

- com

p)

860.4 860.6

1.64

1.68

1.72

1.76

1.80

1.84

1.88

862.4 862.6Julian date ( 2 452 000+)

1.68

1.72

1.76

1.80

1.84

1.88

1.92

∆ m

ag (#

491

- com

p)

Figure 4: Light curves of NGC6811. From top to bottom and left to right:stars #18, #37, #70, #39, #113, #489, #491.


0.3 0.4 0.5 0.6

0.6

0.8

1.0

1.2

1.4

Star #288

0.3 0.4 0.5 0.6Julian date ( 2 452 762+)

1.5

2.0

2.5

3.0

3.5

4.0

∆ m

ag (v

ar -

com

p)

Star #507

Figure 5: Left panel: Extended field of the open cluster NGC 6811 (source: DigitizedSky Survey). The newly detected variable stars are indicated. Right panels: Lightcurves of NGC6811,V8-V9.

Figure 6: Sample CCD frame of the open cluster NGC 7209. The newly detectedvariable stars are indicated.


871.4 871.6

-0.82

-0.80

-0.78

-0.76

-0.74

917.2 917.4 917.6Julian date ( 2 452 000+)

-0.84

-0.82

-0.80

-0.78

-0.76

Dif

fere

ntia

l mag

(#24

- co

mp)

871.4 871.6

0.20

0.22

0.24

0.26

0.28

917.2 917.4 917.6Julian date ( 2 452 000+)

0.20

0.22

0.24

0.26

Dif

fere

ntia

l mag

(#55

- co

mp)

871.4 871.6

0.80

0.82

0.84

0.86

0.88

0.90

917.2 917.4 917.6Julian date ( 2 452 000+)

0.80

0.82

0.84

0.86

0.88

0.90

Dif

fere

ntia

l mag

(#59

- co

mp)

504.4 504.6

-0.48

-0.46

-0.44

-0.42

-0.40

-0.38

855.4 855.6Julian date ( 2 452 000+)

-0.48

-0.46

-0.44

-0.42

-0.40

-0.38

∆ m

ag (#

112

- com

p)

485.4 485.6

0.96

1.00

1.04

1.08

1.12

1.16

857.4 857.6Julian date ( 2 452 000+)

0.96

1.00

1.04

1.08

1.12

1.16

∆ m

ag (#

91 -

com

p)

Figure 7: Light curves of NGC7209. From top to bottom and left to right:stars #24, #55, #59, #112, #91.


6. Conclusions

We gathered new time-series CCD observations in two filters with the aim tosearch for short-periodic variable stars in three poorly studied open clustersfrom the STACC list: NGC 1664, NGC 6811 and NGC 7209. The time seriesdata are particularly extensive in the case of NGC 7209 (more than 100 hrs ofobservations). Among our first results we report the detection of two possibleδ Scuti variable stars in the field centered on NGC 1664, the detection of ninenew variable stars in the field centered on NGC 6811 of which six are probablyof type δ Scuti and one is a suspected δ Scuti star as well as the detection ofthree probable δ Scuti variable stars in the field centered on NGC 7209. Thesearch in NGC 7209 will be resumed as we did not (yet) systematically inspectall the material available. It is also our intention to publish the BV-photometryand to provide updated color-magnitude diagrams for these open clusters.

Acknowledgments. This work is based on CCD observations collected atBeersel Hills (BHO 40cm, BHO 25cm), Hoher List (1m Cass.), NAO Rozhen(Schmidt 50/70cm), SETEC (30cm Meade LX-200) and Skinakas (1.3m) obser-vatories. The Skinakas Observatory is a collaborative project of the Universityof Crete, the Foundation for Research and Technology – Hellas, and the Max-Planck-Institut fur Extraterrestrische Physik. We warmly thank Dr. P. Wilsfor assistance with the reductions. We are grateful to the respective directorsDr. K. Reif, Prof. K. Panov and Prof. I. Papamastorakis for the allocatedtelescope time. Part of these data were acquired with equipment purchasedthanks to a research fund financed by the Belgian National Lottery (1999).PL and AS acknowledge support from the Belgian Science Policy and fromthe Bulgarian Academy of Sciences (project BL/33/B11). PVC is grateful tothe Royal Observatory of Belgium for putting at his disposal material acquiredby the Fund for Scientific Research - Flanders (Belgium) (project G.0178.02).The Simbad (Centre de Donnees Astronomiques, Strasbourg, France) andthe (WE)BDA (Institut d’Astronomie, Lausanne, Switserland) databases wereextensively used.

References

Baumgardt, H., Dettbarn, C. & Wielen, R. 2000, A&AS 146, 251Bessell, M.S. 1995, CCD Astronomy 2, No. 4, 20Dias, W.S., Lepine, J.R.D. & Alessi, B.S. 2001, A&A 376, 441Frandsen, S. & Arentoft, T. 1998, The Journal of Astronomical Data, Vol. 4, No. 6Freyhammer, L. M., Arentoft, A. & Sterken, C. 2001, A&A 368, 580Glushkova, E.V., Batyrshinova, V.M. & Ibragimov, M. A. 1999, Astron. Lett. 25, 86Honeycutt, R.K. 1992, PASP, 104, 435Kjeldsen, H. & Frandsen, S. 1992, PASP 104, 413


Mermilliod J.-C. 1995, in ”Information and On-Line Data in Astronomy”,Eds D. Egret & M.A. Albrecht (Kluwer Academic Press, Dordrecht),127

Missana, M. & Missana, N. 1998, Astronomische Nachrichten 319, 187Pena, J. & Peniche, R. 1994, Rev. Mex. de Astronomıa y Astrofısica 28, 139Platais, I. 1991, A&AS 87, 557Sanders, W.L. 1971, A&A 15, 368Tsvetkov, M.K., Georgiev, T. B., Bilkina, B.P., Tsvetkova, A. G. & Semkov, E.H.

1987, Bolgarskaia Akademiia Nauk, Doklady, Vol. 40, No. 5, 9Vancevicius, V., Platais, I., Paupers, O. & Abolins, E. 1997, MNRAS 285, 871


Vol. 146, 2005

High-degree non-radial modes in the δ Scuti starAV Ceti

L. Mantegazza and E. Poretti

INAF - Osservatorio Astronomico di Brera, Via E. Bianchi 46, 23807 Merate, Italy

Abstract

High-resolution spectroscopic observations taken during 5 consecutive nights ofthe fast rotating δ Scuti star AV Ceti (v sin i = 188km/s) show the presence ofnon-radial modes with ` ∼ 10 − 16 and very short periods (ν ∼ 40− 70 c/d)

Introduction and Observations

The rapidly rotating δ Scuti star AV Ceti has been recently the target of asimultaneous photometric and low-resolution spectroscopic international cam-paign (Dall et al. 20031). Seven low-degree modes were detected in the fre-quency range 10-31 c/d.

We observed this star with the Coude Echelle Spectrograph attached at theESO CAT Telescope during 5 consecutive nights (October 24-30, 1994; RemoteControl from Garching headquarters) for about 2.5 consecutive hours per nightand 55 spectrograms were collected with exposure times of 10 min and a typicalS/N ratio at the continuum level of about 230. They had a resolution of 60000and cover the region 4494-4513 A. In this region we have two useful lines tostudy line profile variations: TiII 4501A and FeII 4508A. Fig. 1 shows theaverage continuum normalized spectrum (upper panel) and the pixel-by-pixelstandard deviation (lower panel).

1Added by editor (MB): The Astronomy and Astrophysics publisher used an incorrectname for the star (AV Cetei instead of AV Ceti) in the title and running head of the Dallet al. paper. The authors of the present publication note: Cetus,-i belongs to the secondlatin declination (as, for example, Cepheus,-i) : nominative and genitive cases are different(Cephei, Ceti). On the other hand, Doradus,-us belongs to the fourth declination and thegenitive is the same as the nominative (γ Doradus, not γ Doradi). For a correct use ofabbreviations of constellations see http://www.iau.org/Activities/nomenclature/const.html

34 High-degree non-radial modes in the δ Scuti star AV Ceti

Figure 1: Average normalized spectrum (top panel); pixel-by-pixel data standard de-viation (lower panel).

Data analysis and discussion

We can see that the line profiles have variations with a std.dev of about0.002 of the continuum level. A non-linear least-squares fit of a rotationallybroadened Gaussian intrinsic profile, taking also into account the limb darkening,on the two average line profiles supplies v sin i = 188± 1 km/s.

The line profile variations were analyzed by means of the Fourier DopplerImaging Technique (Kennelly et al. 1998; Hao 1998). Fig. 2 shows the two-dimensional power spectrum obtained considering all the data in a single timeseries and merging the information of the two spectral lines. We see that thereis a clear clump of modes in the region 10 ≤ ` ≤ 16, 35 ≤ ν ≤ 70 c/d. Becauseof the very severe aliasing, due to the poor temporal coverage, it is not possibleto derive a detection of individual modes.

A CLEANed version of this power spectrum is shown in Fig. 3 (gain=0.9,100 iterations). This figure should be considered only as indicative, and theindividual modes cannot be taken at face value, since the poor spectral win-dow makes a correct deconvolution problematic. However it clearly shows thatwe should expect a very complex pulsation pattern with several high-degree,high-frequency modes. Can this be due to the fast rotation? More intensive

L. Mantegazza and E. Poretti 35

Figure 2: Two-dimensional Fourier power spectrum.

Figure 3: CLEANed two-dimensional Fourier power spectrum.

36 High-degree non-radial modes in the δ Scuti star AV Ceti

observations are necessary to get a more exhaustive picture of the stellar pul-sation spectrum. We also note that the low-degree modes, photometricallydetected by Dall et al. (2003) do not produce relevant patterns in our powerspectrum. This does not necessarily mean that they are not present, since ourtechnique is more sensitive to high-degree modes (see also Mantegazza 2000).

References

Dall, T.H., Handler, G., Moalusi, M.B., Frandsen, S. 2003, A&A 410, 983Hao, J. 1998, ApJ 500, 440Kennelly,E.J. et al. 1998, ApJ 495, 440Mantegazza, L. 2000, ASP Conf. Ser. 210, 138


Vol. 146, 2005

Projected rotational velocities of some δ Scuti andγ Doradus stars

L. Mantegazza1, E. Poretti1

1 INAF - Osservatorio Astronomico di Brera, E. Bianchi 56, 23807 Merate, Italy

Abstract

We present the projected rotational velocities of some δ Scuti or γ Doradusstars as derived from high-resolution spectrograms. For 6 stars the values arethe first determinations.

In the past years during our campaigns of monitoring of some selectedδ Scuti stars (see for example Mantegazza & Poretti 2002 and referencestherein) we obtained spectra of a few δ Scuti or γ Doradus stars in order toestimate their projected rotational velocities. Most of the spectra were takenwith the Coude Echelle Spectrograph attached at the Coude Auxiliary Telescopeat La Silla Observatory (ESO) in the 90’s. These spectra have a resolution of60000 and cover the spectral range between 4490 and 4525 A; their typical S/Nin the center is usually better than 200 at the continuum level. Few more spec-trograms were taken with the FEROS spectrograph, then attached at the ESO1.5 m telescope of La Silla Observatory in the years 2001-2002. These spectro-grams have a resolution of 48000 and cover the spectral range 3600-9300A. Forthe present work the useful lines between 4400-4600A were considered. All thespectrograms were normalized to the continuum, defined by selecting a set ofcontinuum windows star by star and fitting them with a low-degree polynomial.The unblended lines were then fitted by means of a non-linear least-squaresroutine with a rotational profile convolved with a Gaussian intrinsic one takingalso into account the limb darkening. Limb darkening coefficients were de-rived from the paper by Diaz-Cordoves et al. (1995). The number of usefullines varies from 2 to 7, depending upon rotational velocity, spectral type andspectrograph. The typical rms uncertainties, resulting by averaging the valuesobtained form different lines and/or spectrograms, are between 1-2 km/s. The

38 Projected rotational velocities of some δ Scuti and γ Doradus stars

results are reported in the following table, where for each star we give name,spectral type, our v sin i, the number of spectrograms and a letter identifyingthe spectrograph (C=CES, F=FEROS). In the 3 successive columns we givefor comparison the values derived by Royer et al. (2002), Abt & Morrel (1995),and in the third those reported in the δ Scuti star catalog by Rodriguez et al.(2000) or in other recent papers. The references are given in the next one.There are only three γ Doradus stars in the sample: QW Pup, BT Psc, BUPsc.

The agreement among the estimates of different authors is generally satis-factory with a few exceptions. The most striking is that regarding QQ Tel. Weobserve that the estimate by Koen et al. (2002) is based on the correlationprofile (not directly on the line profiles), computed on spectrograms with alower resolution than ours (39000 vs. 60000), and maybe these are the causesof the discrepancy.

Acknowledgments.

References

Abt, H.A., Morrell,N.I. 1995, ApJS 99, 135Breger, M., Garrido, R., Handler, G. 2002, MNRAS 329, 531Diaz-Cordoves, J., Claret, A., Gimenez, A. 1995, A&AS 110, 329Koen, C., Balona, L., van Wyk, F., et al. 2002, MNRAS, 330, 567Mantegazza, L., Poretti E. 2002, A&A 396, 211Mathias, P., Le Contel, J.M., Capellier, E., et. al. 2004, A&A 417, 189Rodriguez, E., Lopez-Gonzalez, M.J., Lopez de Coca, P. 2000, ASP

Conf.Ser. 210,499Royer, F., Grenier, S., Bylac, M.O., Gomez A.E., Zorec, J. 2002 A&A 393, 897

L. Mantegazza and E. Poretti 39

Table 1: Measured v sin i and comparison with other determinations

HD Name Sp.Type v sin i previous valueskm/s (1) (2) others ref.

2724 BB Phe F2II 83 1F – – 83 (3)3326 BG Cet A5m 91 1C 110 98 98 (3)4849 AZ Phe A9.5III 92 1C – – –4919 ρ Phe F2III 84 1C – – –8511 AV Cet F0V 190 1F 212 195 141 (3)

“ “ 188 55C9065 WZ Scl F0IV 33 1C – – –

11413 BD Phe A1V 126 1C 139 – 124 (3)11522 BK Cet F0V 129 2C 133 120 120 (3)15634 TY For A9V:n 128 1C 146 141 141 (3)

“ “ 124 3F16723 BS Cet A7IV 52 1C – – –55892 QW Pup F0IV 56 2C 51 – –66853 BI CMi F2 78 1F – – 76 (4)

160589 V703 Sco A9V ≤10 13F – – ≤16 (3)182640 δ Aql F0IV 89 2C 91 85 –185139 QQ Tel F2IV 65 2C – – 45 (5)208435 BZ Gru F1III-IV 148 1C – – –214441 CC Gru F1III 123 1C – – –215874 FM Aqr A9III-IV 94 2C 110 98 100 (3)223480 BF Phe A9III 83 1F – – 80 (3)

“ “ 83 20C224638 BT Psc F0 19 16C – – 17 (6)224639 BH Psc F0 108 1F – – 110 (3)224945 BU Psc A3 58 26C – – 54 (6)(1) Royer et al. (2002)(2) Abt & Morrel (1995)(3) Rodriguez et al. (2000)(4) Breger et al. (2002)(5) Koen et al. (2002)(6) Mathias et al. (2004)


Vol. 146, 2005

Application of the Butterworth’s filter for excluding

low-frequency trends

T.N. Dorokhova, N.I. Dorokhov

Astronomical Observatory, Odessa National University,Shevchenko Park, Odessa 65014 Ukraine

Abstract

In the research of roAp stars low-frequency atmospheric and instrumental trendsof data can be removed by the application of Butterworth’s filter. Here we showthe validity of such a technique by using a data set in which we combined theobserved star’s data and an artificial frequency spectrum. It is shown that theclearing of the data in the required frequency region is acquired due to the re-duction of low-frequency noise, whereas the frequency solution practically doesnot change. The cleaning enables an easier detection of dominant frequenciesin the investigated frequency region.

During some years we tested low-frequencies filters for reducing atmosphericand instrumental trends from observations of roAp stars obtained under non-ideal photometric conditions. We examined such techniques by using artificialdata series in the manner described by Breger (1990).

The programs PERIOD (Breger 1990), FOUR (Andronov 1994) and Pe-riod98 (Sperl 1998) were used for the analysis. The SPE package by Sergeev(1992) was initially applied for data reducing and filtering.

We prepared a data set consisting of time series measurements of a com-parison star (10th mag) during three long nights (8-10 hours) and incorporatedthe artificial frequency spectrum with the parameters given in Table 1. Theresulting light curve is a combination of the observed atmospheric frequencydistribution with an ideal frequency spectrum. The frequency spectrum slightlychanged: the frequency at 112 c/d became 113 c/d (A ∼ 1mmag) and at 81c/d a frequency appeared with A ∼ 2mmag.

The spectral window and amplitude Fourier spectrum of the combined pat-tern are presented in Fig. 1.

T.N. Dorokhova and N.I. Dorokhov 41

Figure 1: The spectral window and amplitude spectrum of the combined data set.

For the case of roAp stars the frequency region of interest ranges from 70to 260 c/d Therefore, we are especially interested in the frequencies at 71, 81and 113 c/d.

Due to the high noise level, f4=113 c/d is undetectable and does not appeareven after 12 steps of the analysis.

After that we applied the low-frequency Butterworth’s filters of some dif-ferent cutoff frequencies and removed the low-frequency trends of the data.

We define the low frequency Butterworth’s filter in the following way:

|H(f)|2 ≈ (B/f)2M when f B

Table 1: Artificial frequency spectrum.

No frq.(c/d) Ampl Phase1 7 10 0.42 34 7 0.13 71 3 0.64 112 1 0

42 Application of the Butterworth’s filter

Figure 2: The amplitude spectrum of the data after removing the low-frequency trendby Butterworth’s filter degree M=2 and cutoff frequency B = 5%. The revealedfrequencies are: 71.01±0.01 c/d (A=1.9±0.13); 81.26 ±0.02 c/d (A=1.6±0.1);113.15±0.02 c/d (A=1.3±0.1).The first low frequency peak is at 34 c/d. Its am-plitude has decreased by 8 times but the value of frequency is practically unchanged.

The filter with such parameters is close to ideal, i.e. passes only the requiredregion of frequencies (see Otnes & Enochson 1978).

From experience we selected the degree M=2 and cutoff frequency B from2 to 5%. Here we present even more rigorous cutoff frequencies B=5% (Fig.2)and B=7% (Fig.3) studying the filters’ influence to the investigated frequencyregion.

As a result of filtering the low frequencies’ amplitudes became lower (thefrequencies are not changed). Figure 4. shows the noise spectrum for differentcutoff frequencies.

At 100 c/d the noise has been reduced by 20% for B=5%, by 35% forB=7% and up to 70% for B=10%. The reduction of the noise level enablesthe astronomer to find the intrinsic frequencies sooner and more reliable.

These simple examples show that the frequency solution for the investigatedregion is not distorted even by adopting a rather rigorous filter. The applicationof low frequency filters appears to be more efficient and a better procedure thancorrections with low degree polynomials.

T.N. Dorokhova and N.I. Dorokhov 43

Figure 3: The amplitude spectrum of the data after removing the low-frequency trendby Butterworth’s filter degree M=2 and cutoff frequency B = 7%. The revealed fre-quencies are: 113.16±0.02 c/d (A=1.1±0.1); 72.02±0.01 c/d (A=1.06±0.1); 81.26±0.02 c/d (A=1.0±0.1).

Figure 4: The noise spectrum for: solid – the unfiltered data; dotted – after the M=2degree and B=5% cutoff frequency filter; dash-dot-dot - after B=7% cutoff frequencyfilter; dashed – after B=10% cutoff frequency filter.

44 Application of the Butterworth’s filter

References

Andronov I.L. 1994, Odessa Astronomical Publications, 7, 49Breger M. 1990, CoAst, 20, 1Otnes R. K., Enochson L. 1978, Applied time series analysis V.1. Basic

Technologies, New York, p. 428Sergeev 1992, private communicationSperl 1998, CoAst, 111, 1


Vol. 146, 2005

PODEX – PhOtometric Data EXtractor

T. Kallinger

Institut fur Astronomie, Turkenschanzstr. 17, 1180 Vienna, Austria

Abstract

New reduction tools are required to handle the increasing amount of photomet-ric CCD data within a reasonable time. podex is an IDL based tool developedto optimize the time needed for extracting light curves out of photometric CCDdata. The GUI based semi-automatic reduction tool provides the possibilityto apply bias correction, flat fielding, and background modeling to the rawdata. After extracting magnitudes even in crowded fields, a color–dependentextinction correction and a conversion to relative magnitudes is performed. Theoutput of podex are completely reduced light curves of an arbitrary numberof stars on the CCD images. As the conditions between various nights andobserving runs change continuously and sometimes dramatically, full interac-tivity throughout the entire reduction procedure via a powerful graphical userinterface is a big advantage over all black-box procedures known to us.

1. Introduction

The motivation for developing podex originates in the increasing amount ofphotometric CCD data and in the need to reduce them as fast as possible andwith minimum effort. Therefore, a tool is needed providing all needed reductionsteps at once.

All black-box procedures known to us (like IRAF or MOMF) are compli-cated to handle and provide only the extraction of instrumental magnitudes.For a complete reduction also a color dependent extinction correction and theconversion to relative magnitudes is needed. podex provides it.

The podex source code can be found at

http://ams.astro.univie.ac.at/computer.php

46 PODEX – PhOtometric Data EXtractor

Figure 1: Graphical user interface of podex. top left: control field; middle left:The actual image is displayed in the raw data field. The stars of interest are markedwith cross–hair symbols on the first image. bottom left: status line; top right: Theapertures and background annulus of all used stars can be displayed in the aperturefield to control the proper definition of the aperture radii. middle right: The grey–scaled lines indicate the instrumental light curves of all used stars. The black symbolsrepresent the corresponding reference light curve. bottom right: The relative (orabsolute) light curve of the actual star is displayed in the light curve field.

2. Input

After starting IDL, type ”podex” to start the graphical interface of podex.The main input for podex is a file containing the list of CCD images to reduce(in FITS format). podex is able to handle compressed FITS images as well.If the images are not stored in the same directory as the list file (because thedata are on a CD or DVD, e.g.) define the full path in the list file and use thelist includes path option in the main window.

T. Kallinger 47

2.1 Preferences

The preferences of the actual session are controlled via the preference menu(click the Change Preferences button in the control field). There is also thepossibility to import a set of preferences from a file (Load Preferences). Thesession preferences are:

• FITS header keywords for date, Universal Time and integrationtime of the observations:Date and time are needed for calculation of the Julian date. Intensities arenormalized to ADU per, i. e. dividing by integration time.

• Right ascension (h:m:s) and declination (d:m:s) of the image cen-ter:Stellar coordinates at epoch 2000.0 are used for the heliocentric correction ofthe Julian date and the determination of the zenith distance.

• Geographic latitude (d:m:s) and east longitude (h:m:s) of the ob-servatory:Geographic coordinates are used for calculation of the zenith distance neededfor the photometric extinction correction.

• Filename of bias and flat field image:Bias correction and/or flat fielding can be activated by using the subtract bias

and/or divide by flat options in the main window.

• Center box (pixel) for image centering:Size of the box in which podex searches for the center of the star used todetermine the telescope pointing offset of consecutive images.

• Radius of the aperture (pixel):Aperture radius for the brightest used star on the image.

• Inner radius and width (pixel) of the background annulus.

• Scaling factor for individual apertures (pixel per mag):The aperture radius for fainter stars decreases by the given scale factor.

• Magnitude of star #1:In order to compensate transparency changes in consecutive nights, the mag-nitude of star #1 is used as a reference.

When using a preference file, the first line of the file indicates the actual workingpath. Click Save Preferences to save the actual preferences.

2.2 Coordinates

The pixel coordinates of the stars of interest are imported from a three columnfile. The first two columns contain the X– and Y–coordinates on the image. Thethird column an arbitrary color index (Johnson B-V, e.g.) for the correspondingstar. The color index is used for the color–dependent extinction correction. Ifno color index is known, the value should be zero.


The first star (star #1) in the list plays a special part. This star is usedfor determining coordinate offsets of consecutive images due to an insufficientpointing stability of the telescope. The star is also used as a reference lightsource (use the offset correction option in the main window) to calibrate relativemagnitudes to absolute magnitudes and to compensate transparency changesin consecutive nights. Therefore this star has to meet some special conditions.

• Star#1 should be a bright and constant star.

• Star#1 should not be located near the borders of the CCD frame.

• Inside the defined box for image centering no brighter star should be present.

podex provides the possibility to create a list of pixel coordinates. Whenmoving the cursor onto a target (in the raw data field) and clicking the leftmouse button, the center of the star is determined and marked by a cross-hairsymbol. Click the right mouse button, and the actual X/Y position of the cursorand the corresponding intensity (in ADU) are displayed in the status line. Usingthe Save Coordinate File button, the produced list of pixel coordinates (withcolor index equal to zero) is exported to a file.

3. What is going on?

The reduction process is started by clicking the START Reduction button.First, the bias image is subtracted from the raw image and the residual imageis divided by the flat field image (optional). After these corrections the imageis divided by integration time to normalize the intensity to ADU per second. Asub–image (the size is defined by the preference parameter center box) centeredon the pixel coordinates of star #1 is extracted from the image. The differencebetween the given coordinates for star#1 and the center of a 2–D Gaussian fitto the sub–image defines the pointing offset in X and Y direction of the actualimage. This offset is applied to all coordinates.

3.1 The first image

The first image is used to define the individual aperture radii for all stars listedin the coordinate list. As illustrated in Fig. 1 and Fig. 2, there is a clear relationbetween the brightness of a star and the optimum aperture size (determinedby minimization of the point–to–point scatter in the light curve). Followingincreasing concentric circles from the center of a star’s PSF to the border, thesignal–to–background scatter ratio grows smaller. When this ratio falls below acertain level, including further (less exposed) pixels decreases the signal to noiseratio of the resulting light curve. The turning point, where including pixels atthe edge of the PSF starts to decrease the resulting quality, is reached earlierfor faint stars than for bright stars.

T. Kallinger 49

For all stars listed in the coordinate file, the total amount of ADUs inside thedefault aperture (defined by the preference keyword aperture radius) gives thestellar intensity. The mean background intensity is subtracted from the stellarintensity (see next section), and the residuals are transformed into instrumentalmagnitudes. The instrumental magnitudes are used to define the individualaperture radii. If this option is not needed, set the preference keyword scalefactor to zero. In this case the default aperture radius is used for all stars.Instrumental magnitudes and the corresponding aperture radii are listed in theterminal window. This list can be used to acquire color indices when startingwith images obtained with different filters.

5 10 15 20aperture radius (pixel)

1

10

scat

ter /

min

imal

sca

tter

7.7mag12.8mag10.3mag 8.5mag11.8mag12.5mag

Figure 2: Point–to–point scatter versus aperture radius for stars with different mag-nitudes. The scatter decreases with increasing aperture radius, reaches a minimum,and starts to increase again. The size of the aperture with the least scatter in thelight curve depends on the brightness of the star. For fainter stars, smaller aperturesyield to optimum photometric quality.

3.2 Background correction

To use an annulus around the aperture to determine the background level issometimes problematic in crowded fields when one or more stars are located inthe background annulus. Nevertheless, podex uses such a background annulus


7 8 9 10 11 12 13 14magnitude

8

10

12

14

16

18

20

22

aper

ture

radi

us (p

ixel

)

Figure 3: Mean instrumental magnitude versus aperture radius associated to minimumpoint–to–point scatter in the resulting light curve. The solid line represents a linearregression, illustrating the high degree of linearity in the relation between apertureradius and stellar brightness. Here the individual aperture radius decreases with about2 pixel per magnitude.

but tries to ignore pixels exposed to stellar light (or cosmics). Hence the meanintensity level and standard deviation of all stars inside the background annulus(defined by the preference keywords inner radius and width) is determined. Ifa pixel intensity exceeds the mean level by more than three times the standarddeviation, the corresponding pixel is rejected. The mean intensity and standarddeviation of all accepted pixels is re–calculated. The procedure is repeated untilno pixels are rejected but only up to 5 times.

3.3 Extraction of instrumental light curves

After defining the individual aperture radii, the reduction procedure extracts thetotal intensity for all apertures, subtracts the corresponding mean backgroundlevel (multiplied by the number of pixels inside the aperture), and transformsthe residuals into instrumental magnitudes for all images given in the input file.

T. Kallinger 51

Figure 4: Left: Background annulus with containing a star. Right: Pixels exposed tostellar light are not used to determine the mean background level around the aperture(the star of interest is located inside the inner circle).

3.4 Color–dependent extinction correction

If a color index (CI) is given in the coordinate file and the option extinc. cor. inthe control field is selected a color dependent extinction correction is performedfor all instrumental light curves. First, the mean magnitude is subtracted fromall instrumental light curves. Then the extinction coefficient for an individuallight curve is computed as the slope of a linear regression in the Bouguer plot(magnitude versus airmass X). There is a linear relation between the extinc-tion coefficients and the corresponding CI. Again, a linear regression is used todefine the coefficients k0 and k1. The instrumental light curves are corrected by

magcorr = maginstrumental − X · (k0 + CI · k1).

If no color indices are given, the extinction correction is not necessary be-cause the differential extinction for different stars is too small to effect theresulting light curves. (The typical field of view of a CCD images is not largerthan a few arcminutes). The average extinction of the field of view is correctedby subtracting the reference light curve.

3.5 Reference light curve

The reference light curve corresponds to the weighted mean of all instrumentallight curves. The weight of an individual light curve is defined by the inverseD–value divided by the variance of the corresponding light curve. The D–valueis the ratio between the standard deviation and the point–to–point scatter ofa light curve. The D–value is used as an estimator for variability in the lightcurve. A light curve with a high D–value and/or a high variance has low weight


for the reference light curve. Clicking the Calc Va-Cp button, the referencelight curve is determined and subtracted from the light curve of star #1. Theresulting relative (or absolute, if the offset cor. option is used) light curveis displayed in the light curve field. The > (<) button switches to the lightcurve of the next (previous) star in the coordinate list. The reference lightcurve is determined without the star currently displayed. podex also providesthe possibility to exclude stars from calculation of the reference light curve.The stars to be encountered are selected in the use menu. Mean magnitude,variance, D–value, and weight of all light curves are listed in the terminalwindow. Mean magnitude, standard deviation, and D–value of the currentlydisplayed star are listed in the status line.

3.6 Outlier rejection

Outliers can be rejected from the light curve by moving the cursor onto thedata point in the light curve field and press the left mouse button. The recentlyrejected data point can be included again with the undo button.

4. Output

With the Save current star button the heliocentric Julian date, relative (orabsolute if the offset cor. option is used) magnitude, reference magnitude andthe airmass of the actual star are saved in a file. With the Save all starsbutton the heliocentric Julian date and the instrumental magnitudes of all starsare saved in a single file.

Acknowledgments. This work is supported by the Austrian Fonds zurForderung der wissenschaftlichen Forschung (FWF) within the project Stel-lar Atmospheres and Pulsating Stars (P14984), and the Bundesministeriumfur Verkehr, Innovation und Technologie (BMVIT) via the Austrian SpaceAgency (ASA).


Vol. 146, 2005

Period04 User Guide

P. Lenz, M. Breger

Department of Astronomy, Univ. Vienna, Turkenschanzstrasse 17,1180 Vienna, Austria

Abstract

Period04, an extended version of Period98 by Sperl (1998), is a softwarepackage designed for sophisticated time string analysis. In this article wepresent the User Guide for Period04.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . 55

1.1 Backwards compatibility . . . . . . . . . . . . . . . . . 56

1.2 Obtaining Period04 . . . . . . . . . . . . . . . . . . . . 56

1.3 Requirements . . . . . . . . . . . . . . . . . . . . . . . 56

2. The Graphical User Interface . . . . . . . . . . . . 57

2.1 The main window . . . . . . . . . . . . . . . . . . . . . 57

2.2 The menu bar . . . . . . . . . . . . . . . . . . . . . . . 57

2.2.1 The ’File’ menu . . . . . . . . . . . . . . . . . 57

2.2.2 The ’Special’ menu . . . . . . . . . . . . . . . 59

2.2.3 The ’Options’ menu (available in expert modeonly) . . . . . . . . . . . . . . . . . . . . . . . 61

2.2.4 The ’Help’ menu . . . . . . . . . . . . . . . . 61

2.3 The status bar . . . . . . . . . . . . . . . . . . . . . . . 62

2.4 The ’Time string’ tab . . . . . . . . . . . . . . . . . . . 62

2.5 The ’Fit’ tab . . . . . . . . . . . . . . . . . . . . . . . . 65

2.5.1 The Main tab . . . . . . . . . . . . . . . . . . 66

2.5.2 The ’Goodness of Fit’ tab . . . . . . . . . . . 70

2.6 The ’Fourier’ tab . . . . . . . . . . . . . . . . . . . . . 71

2.7 The ’Log’ tab . . . . . . . . . . . . . . . . . . . . . . . 75

2.8 Dialogs . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

2.8.1 The ’Weight selection’ dialog . . . . . . . . . . 76

2.8.2 The ’Subdivide time string’ dialog . . . . . . . 76

54 Period04 User Guide

2.8.3 The ’Combine substrings’ dialog . . . . . . . . 78

2.8.4 The ’Show time structuring’ dialog . . . . . . 79

2.8.5 The ’Calculate specific Nyquist frequency’ di-alog (available in expert mode only) . . . . . 80

2.8.6 The ’Adjust time string’ dialog . . . . . . . . . 80

2.8.7 The ’Set default label for deleted points’ dialog 82

2.8.8 The ’Relabel data point’ dialog . . . . . . . . 82

2.8.9 The ’Edit substring’ dialog . . . . . . . . . . . 83

2.8.10 The ’Calculate epoch’ dialog . . . . . . . . . . 83

2.8.11 The ’Recalculate residuals’ dialog . . . . . . . 84

2.8.12 The ’Predict signal’ dialog . . . . . . . . . . . 85

2.8.13 The ’Create artificial data’ dialog . . . . . . . 85

2.8.14 The ’Set alias-gap’ dialog . . . . . . . . . . . 86

2.8.15 The ’Show analytical uncertainties’ dialog . . . 86

2.8.16 The ’Improve special’ dialog . . . . . . . . . . 87

2.8.17 The ’Calculate amplitude/phase variations’ di-alog . . . . . . . . . . . . . . . . . . . . . . . 88

2.8.18 The ’Monte Carlo simulation’ dialog . . . . . . 89

2.8.19 The ’Calculate noise at frequency’ dialog . . . 90

2.8.20 The ’Calculate noise spectrum’ dialog . . . . . 91

2.9 Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

2.9.1 Time string plots . . . . . . . . . . . . . . . . 94

2.9.2 Phase plots . . . . . . . . . . . . . . . . . . . 95

2.9.3 Fourier plots . . . . . . . . . . . . . . . . . . 96

2.10 Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

2.11 Period04 preferences . . . . . . . . . . . . . . . . . . . 98

2.12 The expert mode . . . . . . . . . . . . . . . . . . . . . 100

2.13 The Data Manager (expert mode only) . . . . . . . . . 101

2.14 Overview of Shortcuts: . . . . . . . . . . . . . . . . . . 104

3. Getting Started . . . . . . . . . . . . . . . . . . . . 105

3.1 Tutorial 1: A first example . . . . . . . . . . . . . . . . 105

3.2 Tutorial 2: Least-squares fitting of data including a pe-riodic time shift . . . . . . . . . . . . . . . . . . . . . . 112

4. Using Period04 . . . . . . . . . . . . . . . . . . . . 118

4.1 Topics related to time string data . . . . . . . . . . . . 118

4.1.1 Importing time strings . . . . . . . . . . . . . 118

4.1.2 Exporting time string data . . . . . . . . . . . 120

4.1.3 Creating artificial data . . . . . . . . . . . . . 122

4.2 Topics related to Fourier calculations . . . . . . . . . . . 123

4.2.1 Calculation of Fourier spectra . . . . . . . . . 123

4.2.2 Significance of frequencies . . . . . . . . . . . 124

4.3 Topics related to least-squares calculations . . . . . . . . 126

4.3.1 Calculation of least-squares fits . . . . . . . . 126

4.3.2 Calculation of amplitude/phase variations . . . 127

P. Lenz and M. Breger 55

4.3.3 Using the periodic time shift mode . . . . . . . 128

4.3.4 Estimation of uncertainties . . . . . . . . . . . 129

4.4 General topics . . . . . . . . . . . . . . . . . . . . . . . 132

4.4.1 Using weights . . . . . . . . . . . . . . . . . . 132

5. Copyright notice . . . . . . . . . . . . . . . . . . . 135

5.1 Third party software . . . . . . . . . . . . . . . . . . . . 135

1. Introduction

Period04 is a computer program especially dedicated to the statistical analysisof large astronomical time series containing gaps. As its predecessor, Period98,the program offers tools to extract the individual frequencies from the mul-tiperiodic content of time series and provides a flexible interface to performmultiple-frequency fits.

Fundamentally, the program is composed of 3 modules:

• The Time String ModuleWithin this module the user administrates the time string data. Themodule contains tools to split a data set into substrings, combine datasets, set weights, etc.

• The Fit ModuleLeast-squares fits of a number of frequencies can be made in this module.Apart from basic fitting techniques, Period04 also contains the possibilityto fit amplitude and/or phase variations, or to take into account a periodictime shift. Furthermore, several tools for the calculation of uncertaintiesof fit parameters, such as Monte Carlo simulations, are available.

• The Fourier ModuleFor the extraction of new frequencies from the data, this module is pro-vided. The Fourier analysis in Period04 is based on a discrete Fouriertransform algorithm. We do not use a Fast Fourier Transform (FFT)algorithm as astronomical time string data sets usually are not equallyspaced.

Some tools or functions are only accessible when the program operates inthe so called ’Expert mode’. A detailed description of the expert mode can befound in section 2.12.

Period04 is project oriented and saves all data (time string data, Fourierspectra, frequencies and the log) in one central project file. The project file


itself is completely platform independent. This allows the user to switch betweenvarious operating systems. As the program also stores the current programsettings along with the data, it is easy to resume work on a project.

1.1 Backwards compatibility

Period04 is based on Period98 by Sperl (1998). Period98 project files (.p98) canbe read by Period04 without any problems. The default extension for Period04project files is .p04 though.

1.2 Obtaining Period04

Period04 is freely available and has been compiled for Linux, Windows andMacOSX. The program can be downloaded from the Period04 website:http://www.astro.univie.ac.at/ dsn/dsn/Period04/.The installers will guide you through the installation process.

1.3 Requirements

Period04 is a Java/C++ hybrid program. Therefore, to run the program aJava Runtime Environment (JRE) has to be installed. The JRE is alreadypreinstalled on every Mac OS X System. Windows and Linux users will haveto check if there already exists a JRE on their system. To do that simplyopen a shell or a command prompt and type ’java -version’ If the commandis not found no Java is installed. In this case you have to download it fromhttp://java.sun.com/getjava. It is free. If the ’java’ command exists, pleasecheck the version number. The Period04 binary works with all versions from1.4.2.


2. The Graphical User Interface

In this chapter the Period04 User Interface will be explained in detail.

2.1 The main window

The main window of Period04 consists of 3 parts: the menu bar at the top, thestatus bar at the bottom and a tabbed frame in the center. The tabbed framecontains 4 folders:

Time string a module for the management of time string dataFit a flexible interface for least-square fittingFourier a module for the calculation of Fourier transformsLog contains the protocol of all actions taken

The ’Time string’ folder is activated by default after start-up.

2.2 The menu bar

The menu bar gives access to 3 menus (File, Special and Help) in default mode,whereas in the expert mode an additional menu entry, Options, is available. SeeSection 2.12: ’The expert mode’ for more details on additional features.

2.2.1 The ’File’ menu

Figure 1: The ’File’ menu


• New ProjectThis command closes the current project and creates a new empty project.

• Load ProjectUse this command to open an existing Period04 (or Period98) projectfile. The file extension of Period04 projects is ’.p04’.

• Save ProjectSave the complete project (time string data, frequencies, Fourier spec-tra,...) into a file. If you save the project for the first time you will beasked to provide a name and a path for the file. Thereafter, each timeyou save, the changes will automatically be written to this file.

• Save Project AsProvides the possibility to save the current project under a different name.

• Recent project-filesGives instant access to up to 10 recently edited projects. The file pathsof projects are saved in a file named ’.period04-recentfiles’ which can befound in the user directory.

• ImportThis menu entry contains the following entries:

– Import time string:Imports a time string from a file. If the project already containstime string data the user will be asked whether he wants to erasethe old time string. Please see section 4.1.1 for further details onthis topic.

– Import frequencies:Import frequencies, amplitudes and phases from a file.

• ExportThe ’Export’ menu entry holds three items:

– Export time string:Saves selected data columns into a file. Please see section 4.1.2 forfurther details.

– Export frequencies:Saves frequency data into a file.

– Export log file:Saves the protocol into a file.


• Manage Data (available in expert mode only)This command opens the Data Manager (section 2.13), a tool to editproject data.

• Expert modeIf the ’Expert mode’ check box is selected, you will have access to addi-tional tools (see section 2.12). This option is by default deselected.

• QuitQuits the program. If the project has changed since the last backup theuser will be reminded to save the project before closing the program.

2.2.2 The ’Special’ menu

The special menu contains tools related to time string management, least-squares fitting and Fourier noise calculations.

Figure 2: The ’Special’ menu in expert mode

• Weight selectionStarts a dialog for activating weights. The chosen weight selection willbe used in Fourier calculations as well as in least-squares calculations.


Time string related menu entries:

• Subdivide time stringOpens a dialog to divide the time string into substrings.

• Combine substringsOpens a dialog to combine substrings.

• Calculate Nyquist frequency (available in expert mode only)Calculate the Nyquist frequency for a user-defined time range.

• Show time structuringShow some information on the time structuring of the current time string.

• Adjust time stringOpens a dialog for adjusting zero-point variations.

• Set default label for deleted pointsOpens a dialog to define the default name and the attribute of the sub-string in which deleted data points will be put.

• Delete selected pointsDeletes all points belonging to the currently selected time string irrevo-cably.

Menu entries related to least-squares fitting:

• Select all frequenciesSelects all frequencies in the Fit Module.

• Deselect all frequenciesDeselects all frequencies in the Fit Module.

• Clean all frequenciesCleans the frequency list in the Fit Module.

• Calculate epochsCalculates the times of epochs for every active frequency.

• Recalculate residualsRecalculate the residuals using the current selection of time points and auser-defined zero point.

• Predict signalPredict the magnitude or intensity at a specific time according to thecurrent fit.


• Create artificial dataOpens a dialog for creation of artificial data.

• Set alias-gapOpens a dialog to set the step size for frequency adjustments.

• Show analytical uncertaintiesOpens a window that displays the parameter uncertainties based on theassumption of an ideal case.

Menu entries related to Fourier calculations:

• Calculate noise at frequencyOpens a dialog to calculate the noise at a specific frequency.

• Calculate noise spectrumProvides a tool to calculate a noise spectrum.

2.2.3 The ’Options’ menu (available in expert mode only)

Figure 3: The ’Options’ menu

• Set fitting functionPeriod04 contains the possibility to change the fitting formula. Apartfrom the default calculation mode, a fit including a periodic time shiftcan be made.

2.2.4 The ’Help’ menu

• Period04 HelpLaunches the Period04 help system.

• TopicsGives access to certain Period04 topics in the help system.

• TutorialsLaunches the help system and shows the main page for the tutorials.


Figure 4: The ’Help’ menu

• ShortcutsOpens a page that lists all shortcuts that are supported by Period04.

• Period04 - HomepageLaunches a web browser and navigates to the Period04 homepage.

• Report a bugProvides instructions for the submission of bugs.

• CopyrightDisplays the copyright notice.

• AboutDisplays some information (i.e., the version number) about your copy ofPeriod04.

2.3 The status bar

The status bar shows additional messages to inform the user of the status of theprogram. In case of File I/O actions a progress bar is displayed that indicatesthe progress of the task.

2.4 The ’Time string’ tab

The Time String Module is dedicated to edit time string data. In the upperpart of the ’Time string’ tab buttons for loading and saving time string dataare located:


Figure 5: The status bar showing its default message.

Figure 6: The status bar showing the progress while loading a project.

• ImportImport time string data. If the current project is not empty, then the oldtime string data will be replaced and the remaining project data (frequen-cies, Fourier spectra and the log) will be erased.

• AppendAdd a time string to the time string of the current project.

• ExportSave time string data to a file.

The text box next to the buttons contains the full path-names of the filesfrom which the data have been imported. Below the text box some propertiesof the currently selected time string are shown:

• Points selectedThe number of points that are currently selected.

• Total pointsThe total number of points in the time string.

• Start timeThe lowest time value of the currently selected time string.

• End timeThe highest time value of the currently selected time string.

• Check box ’Time string is in magnitudes’If your data are in magnitudes instead of intensity this option shouldbe selected. In case of magnitudes the y-axis of time string plots will bereversed – it is also important for correct epoch calculation. By default theprogram assumes that the data are in magnitudes. It is possible to changethe default setting by editing the Period04 preferences (section 2.11).


Figure 7: The ’Time string’ tab.

The main part of the ’Time string’ folder is made up of 4 list boxes. Eachlist box represents an attribute for the time string data. If you did not read inany attributes then the lists will show the entry ’unknown’. The names of theattributes can be changed by clicking on the headings on the top of the lists.You may also change the default names by editing the preferences file.

In the lower part of the Time String Module there are several buttons:


• Edit substringClick on this button to edit the properties (name, color and weight) ofthe currently highlighted items (substrings) of the list. See section 2.8.9for further details.

• Display tableDisplays a list containing the full information on all data points of theselected time string.

• Display graphPlots the selected time string and, if the resolution of the viewport is highenough, shows the current fit as well.

The columns in the Time String Module provide a pop-up menu too. Thefollowing entries are currently available:

• Edit substring propertiesOpens a dialog to edit the properties of a substring.

• Select all substringsSelects all substrings in the current list box.

Figure 8: The pop-up menu for the list boxes in the Time String Module

2.5 The ’Fit’ tab

The ’Fit’ folder itself contains two tabs:

• MainThe ’Main’ tab contains the frequency list, a panel to define the settingsfor least-squares calculations and several buttons to start calculations.

• Goodness of FitThis tab provides tools for the calculation of uncertainties.


Figure 9: The ’Fit’ tab.

2.5.1 The Main tab

In the upper part of the ’Main’ folder buttons for loading or saving frequenciesare situated. If the frequencies are imported from an external file, the file to beimported has to be formatted as shown below:

F1 8.245592 0.036872 0.191925F2 8.866320 0.031595 0.201285

F3 (8.514142 0.009952 0.811823 )


If the frequency parameters are enclosed by parentheses, the frequency willappear inactive in the frequency list.

Furthermore, some information regarding the current fit is given:

• Selected frequenciesThe total number of active frequencies. A frequency is active (and willbe included into calculations) when the respective check box is selectedin the frequency list.

• Zero pointThe zero point of the magnitudes/intensities resulting from the last cal-culated fit.

• ResidualsThe residuals (χ) for the last calculated fit.

To the left of the import/export buttons and to the right of the informationlabels there is a toggle button, respectively.

Figure 10: The toggle buttons.

A click on the left hand side toggle button detaches the ’Time string’ tabfrom the main frame and places the ’Time string’ panel to the left of the mainframe, whereas a click on the right hand side toggle button causes the ’Fourier’tab to be detached and placed to the right of the main frame.

If your screen is wide enough this makes it possible to use Period04 withouthaving to switch between the different tabs.

Settings for least-squares fit calculations:

• Fitting formulaDisplays the formula that is being used for least-squares fits.


• Calculations based onThis defines the type of data (Original or Adjusted) to be used for thenext least-squares calculation.

• Use weightsDisplays a string that indicates which weights are being used for theleast-squares calculation. If no weights are used ’none’ will be displayed.

1 use the weights assigned to the substrings of attribute #12 use the weights assigned to the substrings of attribute #23 use the weights assigned to the substrings of attribute #34 use the weights assigned to the substrings of attribute #4p use point weightd use deviation weight

• Edit weight settingsOpens the weight-selection dialog.

The frequency list:

In the default mode the frequency list consists of input fields for frequency,amplitude and phase values and corresponding check boxes. If a check box isselected, then the respective non-zero frequency will be included in the nextleast-squares calculation.

Input options for frequency fields:

It is possible to enter harmonics or frequency combinations into frequencyfields as shown in the examples below:

=2f1 or =2*f1 for the second harmonic of f1.

=f1+f3 for a combination frequency which is defined asthe sum of f1 and f3.

=2f1-3f4 for a combination frequency which is defined asor =2*f1-3*f4 the difference of the second harmonic of f1 and

the third harmonic of f4.

Please note:A reference to a frequency, which is a combination itself, will be rejected by theprogram. If one of the frequencies listed in the combination is not active, thecombination is not active either.


If a frequency is a combination, a little button showing the letter ’C’ willappear between the frequency check box and the frequency input field. Clickon this button to show the frequency value instead of the combination stringand vice versa.

Figure 11: Example for an example combination frequency

Adding or subtracting the alias-gap value:

In order to add the alias-gap value to the frequency value, enter a trailing’+’. Type ’-’ if you want to subtract the alias-gap value. Multiple ’+’ or ’-’can also be used, e.g., ’34.11+++’ means 34.11 with three times the alias-gapvalue added. The default value for the alias-gap can be set by using the ’Setalias-gap dialog’ (section 2.8.14).

Extensions for the periodic time shift mode (available in expert mode only)

See also section 4.3.3: ’Using the periodic time shift mode’ for an instructionon how to activate this option. If the periodic time shift (PTS) mode is activethe frequency list is extended to contain input fields for the periodic time shiftparameters.

Figure 12: The frequency panel in the periodic time shift mode.


Furthermore, two additional buttons are available:

• Search PTS start valuesSearches for a good start value within a given range by means of MonteCarlo shots.

• Improve PTSStarts a least-squares calculation to improve the time shift parameters.For this calculation all three parameters of the periodic time shift will beimproved, while the parameters of all other frequencies will be kept fixed.

Please note:For the periodic time shift mode the buttons for showing the phase plot andthe calculation of amplitude and/or phase variations will not be available.

2.5.2 The ’Goodness of Fit’ tab

In this tab the uncertainties of the parameters can be calculated. See ’Estima-tion of uncertainties’ (section 4.3.4) for further details.

• InfoLaunches the Period04 help system and shows a page with some in-formation on how to use Period04 to estimate the uncertainties of fitparameters.

• Calculate LS uncertaintiesCalculates either correlated or uncorrelated uncertainties for the fittedparameters via a least-squares calculation.

• Monte Carlo SimulationOpens a dialog which allows for the calculation of uncertainties for thefitted parameters by means of a Monte Carlo simulation.

• Print listPrints the content of the text field.

• Export listOpens a file-selector to save the content of the text field to a file.


Figure 13: The ’Goodness of Fit’ Tab

2.6 The ’Fourier’ tab

The ’Fourier’ folder basically consists of two parts:

• a panel for the Fourier calculation settings

• a list box that contains all Fourier spectra calculated so far.


Figure 14: The ’Fourier’ tab.

The Fourier calculation settings panel

• TitleDefines the name of the Fourier spectrum.

• From / ToDefines the frequency range for the spectrum.


• Step rateDefines the accuracy of the frequency grid for the Fourier calculation.Three default options for the step rate quality are available and can beselected by the combo-box: High, Medium and Low. In order to set auser-defined value, select ’Custom’ and type the desired value into theinput field next to the step rate combo-box.Please note: The smaller the step rate the longer the calculation will take.

• NyquistDisplays the Nyquist frequency for the time string that is currently se-lected. This value should be a good estimate for the upper frequencylimit due to the sampling pattern for the current data set. The algorithmestimates the average time gap between neighboring points while ignoringlarge gaps.

• Use weightsDisplays a string that indicates which weights are being used for theFourier calculation. If no weights are used ’none’ will be displayed.

1 use the weights assigned to the substrings of attribute #12 use the weights assigned to the substrings of attribute #23 use the weights assigned to the substrings of attribute #34 use the weights assigned to the substrings of attribute #4p use point weightd use deviation weight

• Edit weight settingsOpens the weight-selection dialog.

• Calculations based onDefines the type of data to be used for the calculation: ’Original data’,’Adjusted data’, ’Residuals at original’ or ’Residuals at adjusted’. If ’Spec-tral window’ is chosen the program will calculate the spectral windowwhich is centered at zero frequency and reflects the pattern caused bythe structure of gaps in the time string.

• Compact modeThe output of the Fourier calculation can be quite extensive. To reducethe output ’Peaks only’ can be selected. In this case only local maximaand minima are saved. If ’All’ is selected, all data points of the spectrumare stored.

To start the calculation based on the settings in the Fourier Module eitherpress the button ’Calculate’ or type the shortcut ’Alt+C’.


The Fourier list

All calculated Fourier spectra are listed in the Fourier list box. For everyspectrum the title of the spectrum and, in parentheses, the frequency and am-plitude of the highest peak is given. By selecting a spectrum the settings forthat calculation will be loaded and displayed.

The buttons below the list provide the following functions:

• Rename spectrumChange the title of the highlighted Fourier spectrum.

• Export spectrumSave the data points of the highlighted Fourier spectrum into a file. Inorder to export a set of spectra the Data Manager should be used.

• Delete spectrumErase the highlighted Fourier spectrum.

• Display tableShow the table of data points for the highlighted Fourier spectrum.

• Display graphShow the highlighted Fourier spectrum as a plot on the monitor.

Please note:The functions of these buttons are also accessible by a pop-up menu.

Figure 15: The pop-up menu for the Fourier list.


2.7 The ’Log’ tab

Period04 logs all actions taken. It is possible to edit, print or save the log.

Figure 16: The ’Log’ Folder


2.8 Dialogs

2.8.1 The ’Weight selection’ dialog

In the ’Weight selection’ dialog the user selects the weights that should be usedfor Fourier and least-squares calculations. Multiple selections are possible. See’Using weights’ (section 4.4.1) for a detailed description of setting weights.

Figure 17: The ’Weight selection’ dialog

Period04 supports 3 types of weights:

• weights assigned to the substrings of an attributeIf you decided to give your substrings different weights in one or severalof the four attributes and you want to include them in the calculations,then select the appropriate attribute weight.

• point weightIf you read in a point weight for each of your data points and want touse them in your calculations, then select this weight.

• deviation weightIf this option is active the data points will be weighted based on theirresidual value. ’Cutoff’ defines a limit residual value. If the residual ofthe data point is within this limit, then its weight is set 1.0, otherwisethe weight will be calculated based on the residual and the cutoff value.

2.8.2 The ’Subdivide time string’ dialog

To split the time string into subgroups based on times of measurements select’Subdivide time string’ in the ’Special’ menu. Period04 offers two alternativeways of dividing a time string into substrings. The dialog box prompts the userto choose the desired type of subdivision:


Figure 18: The ’Subdivide data’ dialog

• Subdivide by gapsThe program will search for time gaps between consecutive points thatare larger than a user-defined value. The data set is then subdivided bythese gaps.

• Subdivide by fixed intervalsIf you want to divide the time string into substrings using a fixed timeinterval then select this option.

According to your selection one of the dialogs shown in Figs. 19 and 20 willopen. Below the dialogs an appropriate time string plot shows the differencebetween the two kinds of subdivisions applied to one and the same data set:

For a subdivision the following informations are needed:

• Size of the gapDefines the minimum gap size (using the same time unit as your timestring). If two consecutive points are separated more than the givenvalue then a subdivision will be made.

• Start timeOnly data points after this time value will be subdivided. Defaults to thelowest time value of the time string.

• Time intervalDefines the length of the interval for the subdivision.

• Choose the attribute you want to subdivideDefines the attribute (column) in which the subdivision should take place.

• Label prefixThe final labels of the substrings are constructed in the following way:[prefix] + [average time of the substring]

Here you can define this prefix.


Figure 19: Subdivide bygaps

Figure 20: Subdivide byfixed intervals

Figure 21: An examplefor ’Subdivide by gaps’

Figure 22: An examplefor ’Subdivide by fixed in-tervals’

• Use running counterIf this option is selected the program constructs the final label of thesubstrings in the following way:[prefix] + [a running counter]

• Decimal places to use from timeThe number of digits that will be used for the labels of the substrings.

2.8.3 The ’Combine substrings’ dialog

This is the inverse operation to ’Subdivide time string’ (section 2.8.2). In orderto combine a number of substrings first highlight the substrings you want to join.Then open the ’Combine substrings’ dialog (Special / Combine substrings).

Two values have to be defined:

• AttributeSelect the attribute in which the substrings should be combined.


Figure 23: The ’Combine substrings’ dialog.

• New nameEnter the name for the resulting substring in this field.

2.8.4 The ’Show time structuring’ dialog

This dialog displays some information on the structure of the time string. Forevery substring of the selected attribute the following data are given:

• Start / End time

• Length

• Mean time

• Number of points that belong to the substring

Figure 24: The ’Show time structuring’ dialog.


Finally the total observing time and the total mean time of the time stringare given. If ’Time unit is in days’ is selected, the length of the substrings andthe total observing time will be converted into hours, otherwise they will begiven using the same time unit as your time string.

The content of the text area can be printed and saved to a file by clickingon the appropriate button at the bottom of the dialog.

2.8.5 The ’Calculate specific Nyquist frequency’ dialog (available in expertmode only)

Sometimes it is useful to calculate the Nyquist frequency based on data pointswithin a specified time range. To open the dialog choose ’Calculate specificNyquist frequency’ from the ’Special’ menu.

Figure 25: The ’Calculate Nyquist frequency’ dialog.

Enter the desired time range and press ’Calculate’ to get the appropriateNyquist frequency.

2.8.6 The ’Adjust time string’ dialog

It might occur that your substrings have marginally different zero-point offsetsin magnitude/intensity. This can be caused by different instrumentation, ordifferent observational conditions for these substrings.

Zero-point variations entail additional noise in the low frequency domain.To reduce this noise source, one has to adjust the zero-point offsets. This canbe done through the ’Adjust time string’ dialog.

Before you open the dialog, however, you should calculate a fit as this rou-tine uses the residuals to calculate the zero-point offsets. Select the appropriate


Figure 26: The ’Adjust time string’ dialog.

attribute to show the set of given substrings. If weights should be used to cal-culate the average and the sigma values then select ’Use weights’.

In the list several columns are shown:

• LabelThe name of the substring.

• Observed or AdjustedShows the mean value of Observed or Adjusted, respectively.

• Sigma(Obs./Adj.)Shows the deviation of Observed or Adjusted.

• PointsThe number of points that belong to the given substring

• AdjustedIf this column contains the entry ’Yes’ some or all points of the specificsubstring have already been adjusted.

In order to adjust the data, select the substrings and press ’Adjust’. Theresult will be displayed. You may export or print the content of the text box bypressing the appropriate button.

Please note:Remember to select ’Adjusted data’ for all further calculations!


2.8.7 The ’Set default label for deleted points’ dialog

By default Period04 does not delete points directly, but adds them to a substringlabeled as ’deleted’ in the attribute ’Other’. As the user might prefer othersettings, the label and the position (that is the attribute) can be changed usingthis dialog. To open the dialog select ’Set default label for deleted points’ inthe ’Special’ menu.

Figure 27: The ’Set default label for deleted points’ dialog.

• Put deleted points as attributeDefines in which attribute the substring containing the deleted points willappear.

• Label to use for deleted pointsDefines the default name for the substring that contains the deletedpoints.

After pressing ’OK’ the new settings will take effect.

2.8.8 The ’Relabel data point’ dialog

The ’Relabel data point’ dialog appears through a click with the right mousebutton on a data point in a time string plot. It is also accessible via the pop-upmenu of the time string data table.

Figure 28: The ’Relabel data point’ dialog.


Using this dialog it is possible to change the labels for the chosen point ineach attribute. For clarification: the label refers to the name of the substringthat includes this point.

2.8.9 The ’Edit substring’ dialog

Figure 29: The ’Edit substring’ dialog.

The ’Edit substring’ dialog can be accessed through the ’Edit substring’buttons on the ’Time string’ tab or by using the pop-up menu of the attributelists. This dialog (Fig. 29) shows the current properties (name, weight andcolor) of the substring.

Name refers to the label of the substring that currently contains this point.Weight assigns the given weight to all points of the selected substring. See’Using weights’ (section 4.4.1) for more information on using weights for cal-culations. Color defines the color of this substring to be used for plots.

2.8.10 The ’Calculate epoch’ dialog

To calculate the epochs close to a given time open ’Calculate epoch’ in the’Special’ menu. Please note that this calculation is based on the last fit.

It is possible to calculate the time of either

• Maximum light (maximum intensity, respectively)

• Minimum light (minimum intensity, respectively)

• Zero point, the time when the fit crosses the average of the fitted function(1/4 period before the maximum light).

Note that the program already knows whether you are using magnitudes(maximum light at minimum value) or intensities.


Figure 30: The ’Calculate epoch’ dialog

The values for the epoch will be evaluated close to a given time, that hasto be entered into the respective input field.

After pressing ’Calculate’, the results will be displayed in the text box. Thefirst column is the frequency number, the second column shows the value forthe frequency and finally, in the last column the time of the epoch is displayed.You may export or print the content of the text box by pressing the appropriatebutton at the bottom of the dialog.

2.8.11 The ’Recalculate residuals’ dialog

For the purpose of recalculating the residuals using an other zero-point, select’Recalculate residuals’ in the ’Special’ menu.

Figure 31: The ’Recalculate residuals’ dialog

Enter a new zero-point value into the input field and press ’OK’ to start thecalculation.


2.8.12 The ’Predict signal’ dialog

If you want to calculate the signal at a given time, open the ’Predict signal’dialog (Special / Predict signal).

Figure 32: The ’Predict signal’ dialog

The dialog displays the start time of the currently selected time string. Be-low this line a specific time can be entered at which the signal will be calculatedbased upon the last fit. After pressing ’Calculate’ the result will be shown.

2.8.13 The ’Create artificial data’ dialog

Period04 also supports the creation of equally spaced artificial data. If desiredeven a periodic time shift can be included. You may also refer to ’Creatingartificial data’ (section 4.1.3) for a step-by-step guide about how to use thisfeature.

Figure 33: The ’Create artificial data’ dialog

• Start-time / End-timeDefines the time range for which data should be generated.

• Read time ranges from filePress this button to read in multiple time ranges from a file. Such a fileshould contain two columns: in the first column the start times shouldbe given and in the second column the respective end times.


• StepPeriod04 will create data points that are equally spaced in time. Thisfield defines the step size in time.

• Leading/trailing timeIf this option is non-zero the time range(s) will be extended by the givenvalue.

Before the generation of the artificial data set starts, a file name has to bespecified in which the output will be written. Press ’Write data to the followingfile’ and define the name and the path of the file in the file-selector dialog. Ifthe file already exists the user will be prompted whether he wants to appendthe data to the existing file or to replace the file.

2.8.14 The ’Set alias-gap’ dialog

The alias-gap dialog allows for the definition of the step size for frequency ad-justments. This value will be added or subtracted from a frequency when youenter a trailing ’+’ or ’−’ to the respective frequency in the frequency list ofthe Fit Module.

To open this dialog select ’Set alias-gap’ from the ’Special’ menu. Afterchanging the old value and pressing ’OK’ the new settings will take effect.

Figure 34: The ’Set alias-gap’ dialog

2.8.15 The ’Show analytical uncertainties’ dialog

This dialog displays the parameter uncertainties calculated from formulae whichassume an ideal case. See Breger et al. (1999) for the derivation of uncertaintiesfor frequency, phase and amplitude based on a mono periodic fit. If crossterms can be neglected the equations can also be applied for each pulsationfrequency separately. See ’Estimation of uncertainties’ (section 4.3.4) for moreinformation.


Figure 35: The ’Show analytical uncertainties’ dialog

You may export or print the content of the text area by pressing the appro-priate button at the bottom.

Please note:This option is only available when the standard fitting formula is being used.

2.8.16 The ’Improve special’ dialog

The ’Improve special’ dialog appears when you press the ’Improve special’ but-ton in the Fit Module.

Figure 36: The ’Improve special’ dialog

The dialog consists of three list boxes that contain all currently availablefrequency, amplitude and phase parameters. Select the parameters you wantto improve and press ’OK’ to start the least-squares calculation. Unselectedparameters will be kept fixed.


2.8.17 The ’Calculate amplitude/phase variations’ dialog

This dialog opens up when you press the ’Calculate amplitude/phase variations’button in the Fit Module. It offers the possibility to calculate amplitude and/orphase variations. Please also see ’Calculating amplitude/phase variations’ (sec-tion 4.3.2) for a step-by-step guide on how to use this tool.

Figure 37: The ’Calculate amplitude/phase variations’ dialog

Calculation settings:

• Attribute to use:Defines the attribute that contains the time string subdivision that shouldbe used for the calculation.

• Type of variations:Select either ’amplitude variations’, ’phase variations’ and ’amplitude andphase variations’ here.

• Parameters to improve:By default option ’all ampl. & phases’ is selected. That means that forthis calculation the frequencies will be kept fixed and all amplitudes and


phases are redetermined. To specify a different selection click on ’special’and make your choice in the selection dialog that will be opened.

Press ’Calculate’ to start the calculation using the defined settings. Theresults will be displayed in the text field. You may print or export the results toa file by clicking the appropriate button at the bottom.

2.8.18 The ’Monte Carlo simulation’ dialog

This dialog is shown when you click on Monte Carlo Simulation in ’Goodnessof Fit’ tab in the Fit Module. It provides an interface for estimating the uncer-tainties of fit parameters by means of Monte Carlo simulations.

Figure 38: The ’Monte Carlo simulation’ dialog

For the Monte Carlo simulation the program generates a set of time strings.Each data set is created as follows:

• The times of the data points are the same as for the original time string.

• The magnitudes (or intensity) of the data points are calculated from themagnitudes predicted by the last fit plus Gaussian noise.

For every data set a least-squares calculation will be made. Based on thedistribution of fit parameters the program calculates the uncertainties of theparameters.

• Number of processesA ’process’ consists of the creation of time string data and a least-squarescalculation to determine the fit parameters. A low number of processesresults in a bad estimate of the uncertainties. Therefore, a high numberof processes is highly recommended to obtain reliable results.


• Uncouple Frequency and Phase UncertaintiesFrequency and phase uncertainties are correlated. If this option is selectedthe time string will be shifted in time so that the ’average time’ is zero.If this condition is fulfilled, the frequency and phase uncertainties are nolonger correlated. This check box will only be visible if this option mightbe useful, that is when both frequency and phase parameters are fittedusing the standard fitting formula.

• Use system time to initialize random generatorIf this option is selected the random number generator that is used tocreate the time string data sets will be initialized with the current systemtime.

Please see ’Estimation of uncertainties’ (section 4.3.4) for further informa-tions.

2.8.19 The ’Calculate noise at frequency’ dialog

In order to check the significance of a detected frequency it is necessary tocalculate the signal to noise ratio. For this purpose select ’Calculate noise atfrequency’ in the ’Special’ menu.

Figure 39: The ’Calculate Noise at Frequency’ dialog


• Calculate noise at FrequencyDefines the frequency for which the noise should be calculated. Thefrequencies that are stored in the frequency list in the Fit Module areaccessible by the combo-box below the frequency field. By default thelast frequency of the frequency list is activated.

• Box sizeDetermines the size of the frequency range to be used for the noise cal-culation: [frequency − boxsize

2, frequency + boxsize

2]

• Step rateDefines the accuracy of the frequency grid for the Fourier calculation.Three default options for the step rate quality are available and can beselected by the combo-box: High, Medium and Low. In order to set auser-defined value, select ’Custom’ and type the value into the input fieldnext to the step rate combo-box.Please note: The smaller the step rate the longer the calculation will take.

• Calculations based onDefines the type of data to be used for the calculation: ’Original’ data,’Adjusted’ data, ’Residuals at original’, ’Residuals at adjusted’ or ’Spectralwindow’.

Click on ’Calculate’ to start the calculation. The results of the noise calcula-tion will be shown in the text field below. If the frequency has been selected viathe combo-box then even a value for the signal to noise ratio will be calculated.

You can export or print the content of the text box by pressing the appro-priate button.

2.8.20 The ’Calculate noise spectrum’ dialog

Period04 also provides a tool to calculate a noise spectrum. The correspondingdialog can be accessed via the Special menu.

This dialog is very similar to the ’Calculate Noise at Frequency’ dialog.However, for the noise spectrum the range and the step size have to be defined:

• Frequency range: from/toDefines the frequency range for the noise spectrum.

• SpacingIn order to construct a noise spectrum the noise will be calculated forcertain frequencies in equidistant frequency steps. The ’spacing’ definesthe size of these frequency steps.


Figure 40: The ’Calculate Noise Spectrum’ dialog

Click on ’Calculate’ to start the calculation. The resulting noise spectrumdata will be displayed in the text box below. To export or print the content ofthis text box press the appropriate button at the bottom of the dialog.

2.9 Plots

Visual inspection of the data is very important. Period04 supports plotting onthe monitor of time string data, Fourier data and phase plots.

The menu bar for plot windows generally contains the following commands:

• The Graph menu:

– Print GraphOpens a printer dialog for printing the plot using the current view-port.

– Export Graph As (EPS/JPG)Save the current plot as an image in eps or jpg format.

– CloseClose the plot window.


• The Colors menu (not available for Fourier plots)Every attribute has substrings which are assigned to a color. Using thismenu option it is possible to switch between the different coloring of thefour attributes.

• The Data menu (not available for Fourier plots)Here you can select the type of data to be plotted: ’Observed’, ’Adjusted’,’Residuals at observed’ and ’Residuals at adjusted’.

• The Zoom menu:

– Display allResets the zoom.

– Select viewportOpens a dialog to define the graph ranges.

– Freeze viewport (available for time string plots only)If this option is selected the current viewport will be maintained,unaffected of changes in the selection of substrings. Zooming func-tionality is not affected.

– BackUndo the last zooming action.

• The Help menuprovides a link to the help page for the appropriate plot.

The status bar

In the status bar of plot windows the current coordinates of the cursor willbe displayed, provided that the mouse cursor is situated within the plot domain.

Editing plots

The Scientific Graphics Toolkit that is being used for the plots allows toedit some properties of the design:

• LabelsChange the text, font, color or position of the plot title and/or the axistitles.

• AxesChange label font, tic properties or the range.

To open the appropriate dialog, make a right-click on the appropriate labelor axis to activate it, then a left-click to open the dialog.


2.9.1 Time string plots

To plot the currently selected time string, press the button ’Display graph’ inthe ’Time string’ tab.

Figure 41: An example for a time string plot.

In this mode the menu bar is extended by an additional menu:

• The Display menuIn this menu the ’Display header’ option can be selected. If this optionis activated a header will be shown providing some information on thecurrent fit (number of frequencies, residuals, ...).

Editing the contents of time string plots:

• Relabel pointsIf you want to change the substring a particular point belongs to, clickon that point with the right mouse button. The ’Relabel data point’dialog (section 2.8.8) will be shown. To prevent a wrong identification,the dialog does only appear when the point is well separated from otherpoints.

• Deleting pointsThere are two ways to delete points:

– Press ’Ctrl’ and right-click on the point you want to delete.

– Draw a rectangle using the right mouse button. In contrast to zoom-rectangles this rectangle is colored red. After releasing the buttonthe user will be prompted whether he wants to delete the pointswithin the rectangle or not.


2.9.2 Phase plots

To show the phase plot press the ’Phase diagram’ button in the Fit Module.By default the first found inactive frequency is being used for the plot. If noinactive frequency can be found a default frequency value of 1.0 is chosen.Please note: this option is not available within the periodic time shift mode.

Figure 42: A typical phase plot.

Phase plots provide the two additional options which are accessible via the’File’ menu:

• Export phasesOpens a time string export dialog which allows for saving the phases alongwith other data.

• Export binned phasesSame as ’Export phases’ apart from the fact that binned phases are ex-ported instead of phases.


The frequency panel

• The frequency combo-boxDefines the frequency that is used for the plot. The combo-box containsall inactive frequencies that have been found in the Fit Module. Touse a user-defined frequency value select ’Other Frequency’ and type thefrequency into the input field next to the combo-box.

• Use binningIf this option is selected the data will be averaged for discrete phaseranges. The input field next to the check-box defines the number of bins.

2.9.3 Fourier plots

To plot the currently selected Fourier spectrum, press the button ’Display graph’spectrum in the Fourier Module.

Figure 43: A common Fourier plot.

In the Fourier mode the menu bar is extended by one item:

• The Display menu

– Display PowerIf this option is selected power will be plotted instead of amplitude.

– Display headerIf this check-box is activated a header will be displayed containingthe title of the spectrum and the values of the highest peak.


2.10 Tables

Period04 allows to inspect the data in table format too.

To display the time string table, press ’Display table’ in the ’Time string’tab. The table lists all selected data points along with information on residuals,weights and the affiliation to substrings.

Figure 44: The time string table.

Within the Fourier Module it is also possible to inspect the values for fre-quency, amplitude and power of a Fourier spectrum in form of a table. Toopen this table select the respective Fourier spectrum and click on the button’Display table’.

As the tables are available only for informative purposes the table itself can-not be edited.

The menu bar for table windows contains the following entries:

• The Table menu

– Print TableUse this command to print the complete table.

– Export TableSave the content of the table to a file.

– CloseClose the table window.


Figure 45: The data table for a Fourier spectrum.

• The Help menuProvides a link to the appropriate help page.

Within time string tables a pop-up menu is available:

• Relabel pointGives access to the ’Relabel data point’ dialog (section 2.8.8) for theselected time point.

• Delete pointDeletes the currently highlighted time point.

2.11 Period04 preferences

A new feature of Period04 is the preferences file. It is located in the user di-rectory and it is named ’.period04-pref’. At start-up Period04 searches for thepreferences file and reads in the default settings as defined in the file.

If the file is not found, the program shows a message reporting the problemand creates a new file. The program also checks the file for defects. If the fileseems to be damaged or incomplete Period04 will ask the user if he wants todelete the preferences file and to generate a new one or to proceed using theinternal default values.

By editing the ’.period04-pref’ file the user can set the following values:

• default program mode (standard mode or expert mode)


• lower / upper bound for the default Fourier frequency range

• default compact mode for Fourier calculations

• default alias-gap value

• default scale (magnitudes or intensity)

• default names for the set of attribute names

• default file-format for time string files

The user may also add own comments provided that the line starts with’#’. This character advises Period04 not to read that line.

The default preferences file in detail:

## .period04-pref

## This is the Period04 preferences file. You can set

# your personal settings by replacing the default values.# If this file is not found, Period04 creates a new one

# using its original settings.##------------------------------------------

PROGRAM_MODE DEFAULT#

# start the program in the given mode# possible values: DEFAULT

# EXPERT#------------------------------------------DEFAULT_FOURIER_LOWER_FREQUENCY 0.0

## defines the default lower boundary

# of the frequency range for# Fourier calculations#------------------------------------------

DEFAULT_FOURIER_UPPER_FREQUENCY 50.0#

# defines the default upper boundary# of the frequency range for

# Fourier calculations#------------------------------------------DEFAULT_FOURIER_COMPACT_MODE ALL

## defines the default setting for

# Fourier calculations:# ALL = all data points of a Fourier# spectrum are saved

# PEAKS_ONLY = only the maxima and# minima of the spectrum are saved

#------------------------------------------DEFAULT_ALIAS_STEP DEFAULT

## defines the default alias step rate:# DEFAULT = (1./365.)

# a user-defined number


#------------------------------------------

DEFAULT_SCALE_SETTING MAG#

# defines the default scale setting:# MAG = magnitudes# INT = intensity

#------------------------------------------DEFAULT_NAME_SET[0] Date

DEFAULT_NAME_SET[1] ObservatoryDEFAULT_NAME_SET[2] Observer

DEFAULT_NAME_SET[3] Other## this defines the default attribute names

#------------------------------------------DEFAULT_TIMESTRING_FILE_FORMAT AUTO

## this defines the default file format for# reading in time strings.

# AUTO = try automatic determination# You may also enter a user-defined string,

# composed of the following letters:# t ... time

# o ... observed# a ... adjusted# c ... calculated

# w ... point weight# r ... residuals(observed)

# R ... residuals(adjusted)# 1 ... attribute #1# 2 ... attribute #2

# 3 ... attribute #3# 4 ... attribute #4

# 5 ... weight(attribute #1)# 6 ... weight(attribute #2)

# 7 ... weight(attribute #3)# 8 ... weight(attribute #4)# i ... ignore

2.12 The expert mode

If the expert mode of Period04 is activated, additional tools are available suchas:

• the Data Manager (see section 2.13)

• the possibility to calculate fits considering periodic time shifts (see sec-tion 4.3.3)

• a tool to calculate the Nyquist frequency using specific time-ranges (seesection 2.8.5)

The expert mode can be activated by selecting the check-box ’Expert mode’in the ’File’ menu. The expert mode setting is being saved in every Period04


project file and will be reactivated automatically when the project is openedagain.

It is possible to start Period04 in the expert mode by default. For thispurpose you have to edit the Period04 preferences file which is located in youruser directory. Change the setting for ’PROGRAM MODE’ from ’DEFAULT’to ’EXPERT’ as shown in this excerpt:

PROGRAM_MODE EXPERT

## start the program in the given mode

# possible values: DEFAULT# EXPERT

From now, the program will always start in expert mode right away.

2.13 The Data Manager (expert mode only)

The Data Manager is a tool that combines all import and export data optionsPeriod04 is capable of. It can be accessed via the menu bar (File / Managedata) or using the shortcut ’Ctrl+M’, provided that the expert mode is acti-vated (see section 2.12).

Figure 46: The ’Data Manager’

The Data Manager dialog is divided into 2 columns: the left column pro-vides options for import, export and rearrangement of data. The right columnshows some information on the current project. The project information panelwill hide if an option in the left column has been selected. Instead, an other


panel will be displayed, showing additional options for the selected action.

The project information panel:

• Time string data:

Clean Delete the complete time string.Total points The total number of time points.Points selected The number of currently selected time points.

• Frequency data:

Clean Delete all frequencies.Active frequencies The number of selected frequencies.

• Fourier data:

Clean Delete all Fourier spectra.Fourier spectra The number of spectra in the Fourier list.

• Project:

Save Save the current project.Save as Save the current project using an other name.

If the project has changed since the last backup, a red warning label will bedisplayed.

Options related to time string data:

• Append a new time string:Choose this option if you want to add time string data to the currentproject.

• Replace the old time string:Erases the current time string and reads in a new time string. This optiondoes also provide the possibility prevent the automatic deletion of currentfrequency data, Fourier spectra and the log.

• Rearrange time string data:This option makes it possible to interchange time string data columns.You may choose one of the following actions:

– replace ’Observed’ by ’Residuals(Observed)’

– replace ’Observed’ by ’Residuals(Adjusted)’


– replace ’Observed’ by ’Calculated’

– replace ’Observed’ by ’Adjusted’

By default the ’Adjusted’ column will also be replaced by the chosenvalues whereas the ’Calculated’ values will be set zero.

• Export time string data:Choose this option if you want to export time string data from the currentproject to a file.

Options related to frequency data:

• Import frequencies:Import frequency data from a file. Make sure that the file is formattedas shown in this example:

F1 12.716215 22.38070 0.953766F2 12.154121 4.034121 0.146566F3 24.227962 4.264320 0.688949

F4 23.403372 3.816048 0.002060F5 (9.6562750 3.649441 0.998882 )

Unselected frequencies should be placed within parentheses.

• Export frequencies:Export frequencies from the current project to a file.

Options related to Fourier data:

• Export Fourier spectra:This option makes it possible to export either all or a specific selection ofFourier spectra. According to the selection of the user, the program willname the files either using the titles of the spectra or a composition of aprefix and a running number.


Figure 47: The ’Export Fourier spectra’ interface.

2.14 Overview of Shortcuts:

General Shortcuts:

Ctrl + N Create a new projectCtrl + O Load an existing projectCtrl + S Save the active projectCtrl + T Import/append a time stringCtrl + A Adjust dataCtrl + F Import a set of frequenciesCtrl + M Launches the data manager (available in expert mode only)Ctrl + W Open the weight-selection dialogCtrl + Q Quit the programF1 Launch the Help System

Shortcuts only available in the ’Time string’ tab:

Alt + E Export time string dataAlt + D Display time string

Shortcuts only available in the ’Fit’ tab:

Alt + C Improve amplitudes and phasesAlt + I Improve all parameters

Shortcuts only available in the ’Fourier’ tab:

Alt + C Calculate Fourier transformationAlt + D Display the selected Fourier spectrum


3. Getting Started

The best way to learn how to use Period04 is by doing. For this reason wecreated two example time strings which will be processed and discussed in thetutorials below. The data files can be downloaded from the Period04 homepage:http://www.astro.univie.ac.at/ dsn/dsn/Period04/

3.1 Tutorial 1: A first example

In this tutorial, we will use Period04 to examine a data set and determine thefrequencies by using Fourier analyses to give us rough values of the frequenciesand the Fit Module to refine these frequencies. Note that the Fourier analysiscannot by itself solve the problem since it is a single-frequency method.

1. Start the program Period04.You may wish to use the file ’Empty Period04 file.p04’ which is part of thedownloaded data package. The tab ’Time string’ is selected and active.You will see four empty columns.

2. Import the data set.We will now load the data file. It has the name ’Tutorial1.dat’. Clickon the button ’Import time string’ (left, near top). A window opens andasks for the location of the data. Find the proper directory on your harddisk and click on the file name. Click on the button ’Import’.

Figure 48: The ’Time string import’ dialog.


A window opens and asks for the properties of each column. Since thefirst data column is the time and the second columns contains the mag-nitude variations, everything is fine. Click on ’OK’.(If not, under ’Column #1’ etc. you can select the contents of each col-umn.)

You now have 1254 data points loaded with observing times ranging from2720.81478 to 2740.92739. Do not worry about the ’unknown’ labels ineach of the four columns: it just means that you have not organized yourdata into groups.Save your data now (File, Save Project as) under, say, ’First’. It is nowstored as First.p04.

3. Look at the data.In the same ’Time String’ window, click on ’Display Graph’ at the bottomright. A new window opens with the light curves of the selected data,which, by default, was all the data. You notice that the data strings arespaced one (or more) days apart.

Let us examine a (any) single night. With your mouse select a part ofthe data by drawing a rectangle around it. You may wish to increase thescale of your selection by drawing more rectangles. If you make a mis-take, open the ’Zoom’ dialog (top of ’Time string plot’) and use an option.

The single-night data indicates a variation lasting about 0.1d, with non-repetitive light curves. This may already be a sign of multi periodicity.Notice too that within each night, the data are taken about every 0.003dor 5 minutes apart. This figure is approximate because the coverage dif-fers from night to night. The sampling theorem suggests that periodsshorter than 10 minutes should not be determined with such a data set.To put it differently, the Nyquist frequency is about (0.5 ∗ 1/0.003) or167 cycles/day, abbreviated as c/d.Furthermore, the data were taken one (or more) nights apart with day-time gaps. Consequently, 1 c/d aliasing is expected.

To summarize:

(a) we suspect that frequencies near 10 c/d exist,

(b) the Nyquist frequency should be near 167 c/d, and

(c) 1 c/d aliasing may exist.


4. Perform a Fourier analysis of the data: Spectral WindowMake sure that you have selected all the data. Click on the ’Fourier’ tab.A new window opens. Let us now enter all the necessary parameters.

Title: Spectral Window

From: 0

To: 5(remember that the spectral window is centered on 0 c/dand calculates the pattern caused by the observing gaps).

Use weights: Keep ’none’

Calculations based on: Spectral window

Compact Mode: All(since there are no huge gaps in the data)

Now press the central button: Calculate.

Let us look at the answers. In the line above the ’Calculate’ button, wesee that the highest peak occurs at frequency 0 with an amplitude of 1.This has to be the answer for a spectral window.

Click on ’Display Graph’ on the bottom right. A plot window opens. Youcan see the 1 c/d structure. Keep it in mind for the frequency search ofthe stellar variations. The true frequencies of the star should also showthe pattern, but centered on the true frequencies.

Figure 49: The spectral window.


Close Fourier graph: Spectral Window.

5. Perform a Fourier analysis of the data: Periodic content of dataYou are still in the Fourier window. If not, click on the ’Fourier’ tab. Letus now enter all the necessary parameters:

Title: All data, incorrect zero-point(the choice of this title will become clear below)

From: 0Now see the Nyquist Frequency (167.778). Use this.

To: 167


Calculations based on: Original data

Compact Mode: All


Now a window opens asking you to select the zero-point shift. It allowsyou to subtract the average brightness.

(a) INCORRECT OPTION:

Let us pick the incorrect option for the present data set. In the presentcase, we select ’No’. This means that we assume that the measured av-erage is not the true stellar average - this can indeed happen.

Let us look at the answers. In the line above the ’Calculate’ button, wesee that the highest peak occurs at frequency 0 with an amplitude of0.4875. No, this is not the spectral window. It is a consequence of theincorrect zero-point!

Do not include this frequency. Answer ’No’ to the question.

Click on ’Display Graph’ on the bottom right. A plot window opens. Wesee two patterns, one centered on the frequency 0, the other one at 10.Let us examine the structure at 0 in more detail: open the ’Zoom’ dialogto the top of the plot and use option ’Select viewport’. Enter:


Frequency min: -0.1

Frequency max: 5

Keep the chosen amplitudes. Click ’OK’.

Figure 50: An example for using a wrong zero-point

You see the peak with an amplitude of 0.49 (twice the incorrect zero-point value) near frequency 0. You also see the spectral window patternassociated with this peak (Fig. 50).

Close the graph and redo the Fourier analysis with the correct zero-point.

(b) CORRECT OPTION:

We redo the calculation (calling it ’All data’) and say ’Yes’. Now thehighest amplitude occurs at 10.0011883 with an amplitude of 0.20124.

Answer the question: ’Do you want to include this frequency?’ with ’Yes’.It is now entered in the ’Fit’ window.

Look at the Fourier diagram again (button ’Display Graph’ at bottomright). A nice pattern of peaks around the frequency 10 is visible. Adecision to try out this frequency for a fit appears reasonable. Let us doit.

Close the plot.


Figure 51: The Fourier spectrum using the correct zero-point

6. A One-frequency fit.Click on the ’Fit’ tab (top). A window opens. You see the previouslysuggested frequency.

(a) First calculation.

Select the first frequency F1 by clicking into the square to the left ofF1. A check mark appears. Click on ’Calculate’ at the bottom left. Youobtain the following result almost immediately:Amplitude = 0.202266723 and Phase = 0.955286.Near the top right you will see: Selected frequencies = 1. Yes, that istrue. Zero point: 0.2426. Yes, that is close to the average value al-ready suggested by the program before the Fourier analysis. Residuals:0.070878. We want to minimize this quantity, but we do not know whatthe minimum value will be.

(b) Improve the frequency.

This option should be used carefully. Let us apply it (button bottommiddle). The frequency becomes 9.99988955, Amplitude = 0.202731263,Phase = 0.501472. More importantly, the residuals have improved slightly.

7. Perform a Fourier analysis of the residualsLet us see if the residuals contain more periodicities. Click on the ’Fourier’tab.


Title: Residuals, 1 frequency

From: 0

To: 167


Calculations based on: Residuals at original (Note!)

Compact Mode: All


A frequency of 14.5008529 and amplitude 0.0994119445 are found. In-clude the frequency (for the next fit).

Examine the plot. It looks very good.

8. A Two-frequency fit.Click on the ’Fit’ tab (top). Now select both F1 and F2. Probably youonly need to click into the square to the left of F2 to see check marksnext to both frequencies.

Click on ’Calculate’. The residuals decrease. Click on ’Improve all’. Weobtain frequencies of 10 and 14.5, amplitudes of 0.2 and 0.1, respectively,and essentially zero residuals. That’s it.

Do you want to see how the fit looks? Click on the ’Time string’ taband select ’Display Graph’. The program cannot show the fit for severalnights. Therefore, select a single night (rectangle....). You now see theexcellent fit.


Figure 52: The final fit.

3.2 Tutorial 2: Least-squares fitting of data including a periodic time shift

This tutorial provides a basic introduction in how to find a multiple frequencysolution to a data set, taking into account a periodic time shift. Such a periodictime shift could be the result of orbital light-time effects.

1. Start Period04.

2. Import the data file.To read in the data set, press the button ’Import time string’. A file-selector dialog will be opened. Navigate to the directory that containsthe tutorial time strings and select the file ’Tutorial2.dat’.

In the next dialog, we are going to specify the properties of the columnsin the data file. The first column of our data file denotes time, whereasthe second column contains the observed magnitudes. Period04 shouldalready have assumed that, so just click on ’OK’.

Now press ’Display graph’ and examine the time string. You will observethat a strong beating is present, which could be the result of two closefrequencies.

3. Extract the first frequency.Click on the ’Fourier’ tab. In the ’Fourier Calculation Settings’ panel,enter a title for the new Fourier spectrum. As you can see, the Nyquistfrequency of this time string is 139.806 cycles/day. Extend the upper


limit of the frequency range to the lower integer part of this value (139).Before you start the calculation, make sure that the option ’Original data’is selected. Now press ’Calculate’.

A dialog will show up asking whether the zero point of magnitudes /intensities should be subtracted. Press ’Yes’. After the calculation hasfinished, the highest peak of the new spectrum is reported. Click on ’Yes’to copy the values of the frequency peak into the Fit Module.

In the ’Fourier’ tab, press ’Display graph’ to inspect the plot of the Fourierspectrum.

4. Calculate a first fit.Switch to the ’Fit’ tab. You will notice that the new found frequencyhas not yet been selected. Select frequency F1 and press ’Calculate’ toimprove amplitude and phase. Then, to improve all parameters press ’Im-prove all’.

To check how good this solution fits the data, move to the ’Time string’tab and press ’Display graph’. As you see, there is still some work to do.

5. Find and fit further frequencies.Switch back to the ’Fourier’ tab and calculate the next spectrum. Fromnow on the Fourier calculations should be based on the ’Residuals at orig-inal’, so make sure that this option is selected.

In the Fit Module, select the newly found frequency and press ’Calculate’.Then click on ’Improve all’ to find the best least-squares solution.

Figure 53: The preliminary Four-frequency fit.

To detect further frequencies proceed as stated above. Extract two morefrequencies. The residuals will continue to decrease. Fig 53 shows thefrequency parameters you should obtain.


6. Examining the dataFinally, after extraction of four frequencies the least-squares solution fitsquite well. Maximize the plot window and examine each night carefully.You will notice that during the first and the last nights of the time stringthe data points are shifted slightly to the right of the fit, whereas for thenights in the middle of the data set the points are situated slightly leftof the fit. This may indicate that a periodic time shift is present with aperiod that is approximately equal to the total length of the data set intime, about 100 days. That corresponds to a frequency of 0.01 cycles/day.

Well, let us see if we can find any further frequencies. Switch to theFourier Module and extract the next frequency:Frequency = 8.23515582 cycles/day and Amplitude = 0.00092585 mag-nitudes.

This is quite interesting, isn’t it? This frequency is quite close to the firstdetected frequency (F1). The difference is only 0.010095 c/d which isroughly the value for the frequency of the periodic time shift estimatedby visual inspection. It seems likely that the new frequency is an artefactcaused by a periodic time shift.

Now let us check whether our suspicion, viz. the presence of a periodictime shift, can be confirmed. Leave the new frequency (F5) unselectedand do not change your four-frequency solution.

7. Activate the periodic time shift (PTS) mode.

Figure 54: The frequency panel in the periodic time shift mode.

Fitting a periodic time shift can only be done if Period04 runs in expert


mode. To activate the expert mode, set the option ’Expert mode’ in theFile menu selected. A new menu entry, Options, will appear on the menubar. This menu contains the entry ’Set fitting function’ which providesthe alternative ’Standard formula with periodic time shift’. Select thisoption. Now the program is enabled for calculating least-squares fitsincluding a periodic time shift. You will notice that the Fit Module hasslightly changed.

8. Determining the periodic time shift parameters.For non-linear fitting good starting values are essential. In general, initialvalues for the periodic time shift parameters can be estimated from visualinspection of the data, as we did before. Period04 also provides a toolto search for starting values within a user-defined range of frequenciesand amplitudes by means of Monte Carlo shots. Press the button ’SearchPTS start values’ to use this option.

Figure 55: The ’Search PTS start values’ dialog.

The lower frequency limit is calculated from the time base of the dataset. You should not search for frequencies with lower values. The reasonis that for such frequencies the time base of the data is too short to allowa reliable determination of the periodic time shift.The number of shots refers to the number of initial parameter values thatare being tested. We will keep the default values. However, we will de-select ’Use system time to initialize random generator’ in order to allowthe users to compare their results with the results given here. Press ’OK’to start the calculation.

After the calculation has finished, the best set of starting values for theperiodic time shift parameters will be displayed (Frequency = 0.00975cycles/day, Amplitude = 0.00104 days). Now let us improve these pa-rameters by clicking on ’Improve PTS’.


Figure 56: The fit after the determination of the periodic time shift parameters.

Finally, improve all frequencies together with the periodic time shift pa-rameters by clicking on ’Improve all’. Do not use ’Calculate’, as for aproper fit of a periodic time shift, the frequencies also have to be rede-termined!

9. Extraction of further frequenciesSwitch to the ’Fourier’ tab and calculate a new Fourier spectrum. Press’Display graph’ and have a look at the plot. It is obvious that the de-tected frequency peak is not significant. Therefore, our analysis will stopat this point.

Figure 57: The Fourier spectrum shows no significant peak.


Please note:Let us suppose that you have found a statistically significant frequency.In this case you should use ’Improve all’ to obtain a new least-squaressolution. Furthermore, you do not have to use the ’Search PTS start val-ues’ tool again, as you already have good starting values for the periodictime shift which can be improved.

10. The final solutionYour final fit using the four extracted frequencies and including a periodictime shift should be:

Figure 58: The final solution.

The residual noise is 0.000999 magnitudes.

Click on the ’Time string’ tab and press the button ’Display graph’. Youwill notice that the solution fits the data very well. After having appliedthe periodic time shift, the data points are distributed uniformly on bothsides of the fit.

Now let us compare the final parameters to the values that had been usedto generate this time string:

# Frequency Amplitude Phase

------------------------------------------------------PTSF 0.01 0.001 0.5F1 8.24559 0.036872 0.19192

F2 8.86632 0.031595 0.20128F3 8.51414 0.009952 0.81182

F4 7.42476 0.008187 0.76091------------------------------------------------------

Desired residual noise: 0.001

Note the good agreement. The deviation from the initial values is causedby noise.


4. Using Period04

Within this section some of the key features of Period04 are discussed. In manycases short step-by-step guides are given for easier comprehension.

4.1 Topics related to time string data

4.1.1 Importing time strings

1. Start the importThere are several different ways to start the import of a time string,e.g. you can use the ’Import’ or ’Append’ buttons in the ’Time string’folder. ’Import’ erases old project data and then loads the new time string.’Append’ adds the time string to the current project without changing itotherwise.

In addition, there are two more general approaches:

• The ’Import → time string’ command in the ’File’ menu.(Shortcut ’Ctrl+T’)This command checks whether the project already contains timestring data. If no time points are found the file-selector dialog isshown, otherwise the program will display the dialog shown in Fig 59.

Figure 59: The selection dialog for time string imports.

’Append a new time string’ simply adds a data set to your existingtime string without any further changes in the project. If you select’Replace the old time string’ you have the possibility to instruct Pe-riod04 not to erase the frequency data and/or the log.

• Importing a time string through the Data Manager (section 2.13).


After having chosen one of the upper actions a file-selector is shown.By default all files with the extension ’dat’ are shown. To display fileswith another extension, please select ’All files’ in the file-selector dialog.Navigate to the appropriate directory and select the file containing timestring data.

2. Set the file-formatThe next step is to define the columns in your data file. For this purposethe file-format dialog is opened.

Figure 60: The file-format dialog.

The dialog displays the first few lines of the selected file and provides aninitial guess for the column identification determined from the structureof the columns in the file. The dialog shows the first four columns of thefile. In order to see the other columns, please use the scroll bar at thebottom of the text box.

The following data can be read in by the program:

• Time

• Observed

• Adjusted

• Residuals to Observed

• Residuals to Adjusted

• Calculated

• Point weight


• 4 attributes and 4 weights assigned to these attributes *)

• Ignore (column will not be read in)

*) The names of the attributes depend on the program settings and canbe changed by clicking on the heading buttons in the Time String Mod-ule. The default names are ’Date’, ’Observatory’, ’Observer’ and ’Other’.In order to change the default names, the Period04 preferences file hasto be edited (see section 2.11).

3. Finally, press ’OK’

Please note:Lines starting with ’#’, ’;’ or ’%’ will be treated as comment lines and areignored.

Additional options:

It is also possible to select multiple files in the file-selector dialog. In thiscase, the program will ask the user for the file-format of the first data file. Afterthe specification of the file-format, a dialog will show up that allows to selectwhether the program should apply the same file-format for all data files or toshow the file-format dialog for every file.

If the columns of most of your data files have a very special format, youmay find it useful to define a default file-format for reading in time strings. Inthis case the file-format dialog will always propose the given default file-formatinstead of trying to provide a guess about the file structure. This can be doneby editing the Period04 preferences file (see section 2.11).

4.1.2 Exporting time string data

1. Open the export dialogTo start the export you can either press the button ’Export time string’,use the shortcut ’Alt+E’ or, if the program is running in expert mode,start the Data Manager and select the option ’Export time string data’.In any case the following dialog will show up:

2. Set the output formatThe dialog contains two columns: the left text box lists all data columnsthat can be exported. To the right there is another box, which definesthe structure of the columns in the resulting file. Now select the columnsto be exported from the list of available columns. Click on the button


Figure 61: The time string export dialog.

’Add’ to add these columns to the list on the right. If you want to removea column that you have added before, simply select the column name inthe ’Columns to export’ list and press ’Remove’. Finally, when you havefinished your selection, press ’OK’.

Figure 62: The time string export dialog with an example of a selection.

In the example shown in Fig. 62, we have chosen the columns ’Time’,’Observed’, ’Calculated’, ’Observer’ and the weights assigned to the sub-strings of attribute ’Observer’ for export.


3. Choose a file nameNow, a file chooser will ask you for a name for the file. Enter a file nameand press ’Export’ to start the export of the data.

4.1.3 Creating artificial data

It is easy to generate an equally-spaced artificial data set using Period04. First,make sure that there is a set of active frequencies – or at least one activefrequency – in the frequency list of the Fit Module. The current fit will be usedfor the calculation of magnitude/intensity of the artificial data.

1. Select ’Create artificial data’ in the Special menuThe ’Create artificial data’ dialog will appear.

Figure 63: The ’Create artificial data’ dialog.

2. Setting the time range(s)There are two options to define the start and end times of the time seriesthat will be generated:

• Definition of start and end time by the given input fields.

• Using the button ’Read time ranges from file’ for reading the defi-nitions of start and end times from a file.

The latter option has the advantage that artificial data for an arbitrarynumber of different start and end times can be created right away. Pleasemake sure that the file obeys the following format conditions:

• first column: start time

• second column: end time

3. Calculation settingsNow you have to define the step size. By default a value of 1/(20∗Fmax)is given, which provides a good sampling even for the highest frequency.


The input field ’Leading/trailing time’ gives you the possibility to extendthe time ranges by a certain value.

4. Start calculation and output of dataClick on the button ’Write data to the following file ...’ and choose a filein which the data should be written. If the file already exists, you will beprompted whether you want to append the data to the file, replace thefile or to cancel the task.

4.2 Topics related to Fourier calculations

4.2.1 Calculation of Fourier spectra

Let us suppose that you have already read in a time string.

1. Click on the ’Fourier’ tab.As you can see, in the upper part of the window the settings for theFourier calculation can be defined.

2. Choose a title for the Fourier spectrum.

3. Define the frequency range for the spectrum.Make sure that the upper frequency bound does not exceed the Nyquistfrequency.

4. Choose a step rate.This value defines the step size in frequency. Usually the step rate quality’High’ provides a good sampling.

5. If you want to use weighted data, set and activate the appropriate weights.See section 4.4.1 for more information on this topic.

6. Choose the type of data you want to use.If this is the first frequency to be extracted from the data set, ’Originaldata’ is the right choice. For subsequent frequencies you would select‘Residuals at original’. If you adjusted the zero-points of the time stringselect ’Adjusted’ instead of ’Original’.

7. Select the compact mode.If you want to reduce the output (the number of data points) of theFourier calculation, select ’Peaks only’. If this option is selected only thelocal extrema of the spectrum will be saved.

8. Press ’Calculate’ to start the calculation.A dialog will be opened and you will be asked whether you want to


subtract the zero point of the magnitudes/intensities. Press ’Yes’. Whilethe program is calculating the ’Calculating’ tab is being shown. It reportsthe progress of the calculation and shows the chosen settings. By clickingon the button ’Cancel Calculation’ the calculation can be stopped.

9. After the calculation has been finished, a dialog is opened that informsyou about the value of the highest peak and asks you if you want toinclude this frequency. Press ’Yes’ to add the frequency to the frequencylist in the ’Fit’ tab.

10. Inspect the Fourier spectrum visually by clicking on ’Display Graph’. Thenext step is to make a least-squares fit. See ’Calculation of least-squaresfits’ for further details.

11. For extracting further frequencies proceed as stated above except thatyou choose ’Residuals at original’ instead of ’Original data’.

4.2.2 Significance of frequencies

Every measurement is affected by noise that may be generated by many sources(e.g., observations, instrumentation, undetected frequencies). For Fourier spec-tra, noise has the annoying effect that a peak that has been found may not bereal.

The signal to noise ratio

Therefore, a criterion is needed to ensure that the signal is a real feature.Empirical results from observational analyses by Breger et al. (1993) and nu-merical simulations from Kuschnig et al. (1997) have shown that the ratiobetween signal and noise in amplitude should not be lower than 4.0 for highsignificance.

What is the signal?As signal we define either the amplitude of the respective peak in the Fourierspectrum or the amplitude of the least-squares solution for the peak.

What is the noise?The noise is defined as the average amplitude in a frequency range that enclosesthe detected peak. The noise may be calculated before or after prewhiteningthe peak. The choice is beyond the scope of this manual.


Checking the significance of a frequency

Let us assume that you have calculated a Fourier spectrum and detected anew peak.

1. Select ’Calculate noise at frequency’ in the ’Special’ menu.By default the program loads the last frequency value from the frequencylist in the Fit Module into the frequency field of the dialog. If you want tocalculate the noise for an other frequency, type the new frequency valueinto the frequency field.

2. Define the box size for the noise calculation.This value specifies the frequency range [frequency− boxsize

2, frequency+

boxsize2

] which is being used for the calculation of the noise.

3. Choose a step rate.Usually the step rate quality ’High’ provides a very good sampling.

4. Select the type of data to use for the noise calculation.In most cases ’Residuals at original’ is what you need, unless you workwith adjusted data.

5. Press ’Calculate’ to start the calculation.After the calculation has been finished, the noise result will be displayedin the text area in the lower part of the window. If the frequency hasbeen selected from the combo-box then even the signal to noise ratio willbe shown.

Period04 also contains the possibility to calculate a noise spectrum. Seesection 2.8.20 for more details.


4.3 Topics related to least-squares calculations

4.3.1 Calculation of least-squares fits

This is a step-by-step guide for calculating least-squares fits using the standardfitting formula:

f(t) = Z +∑

i

Ai sin (2π (Ωit + Φi))

1. Click on the ’Fit’ tab.

2. After you have extracted a frequency by means of a Fourier calculation,the frequency and amplitude are added to the frequency list. You mayalso enter own frequency values.

3. Activate the new frequency by selecting the check box at the beginningof the frequency line.

4. Select the type of data you want to use. By default ’Original data’ isselected.

5. If you want to use weighted data, set the appropriate weights. See sec-tion 4.4.1 for more information on this topic.

6. Press ’Calculate’ to improve amplitude and phase.

7. Press ’Improve all’ to improve all parameters, including the value of thefrequency.

Period04 does also allow for more sophisticated least-squares fits:

• Improve special:If you want to define a specific selection of parameters that should beimproved, then choose this option.

• Calculate amplitude/phase variations:Use this if you suspect that the amplitude and/or phase of a parameteris variable in different substrings.

• When the expert mode is active it is possible to calculate least-squares fitsthat include a periodic time shift. See section 4.3.3 for more informationon this topic.


4.3.2 Calculation of amplitude/phase variations

If you suspect that the amplitude and/or phase of a parameter is variable withindifferent substrings (e.g., representing different years), you might consider touse this tool.

1. Prerequisites:Let us assume that you have a set of frequencies and a certain subdivisionof your time string in the attribute ’Date’. By calculating a least-squaresfit for each substring separately you realized that the amplitude of afrequency might be variable.

Figure 64: In this example a time string has been subdivided in attribute ’Date’.

2. Press the button ’Calculate amplitude/phase variations’ in the Fit Mod-ule.

3. In the dialog, select the attribute in which the subdivision of the timestring is located.

4. Choose the type of variation: ’amplitude variation’, ’phase variation’ or’amplitude and phase variations’.

5. Select the frequency that shows the amplitude variation.

6. Decide whether you want to improve ’all amplitudes and phases’ whilekeeping the frequencies fixed, or to define a special selection of variableparameters. If you prefer the latter, then click on ’special’ to open adialog for the definition of the selection.

7. Press ’Calculate’ to start the calculation.


8. After the calculation has finished the results are displayed in the text boxbelow.

4.3.3 Using the periodic time shift mode

In order to account for light time effects, Period04 does also allow to includea periodic time shift in your least-squares calculations. In this case the fittingformula is extended in the following way:

f(t) = Z +∑

i

Ai sin (2π (Ωi [t + αpts sin (2π (ωptst + φpts))] + Φi))

The periodic time shift parameters are labeled with the subscript ’pts’. Zdenotes the zero point of f(t) and Ai, Ωi and Φi the parameters of frequency i.

Please note:This calculation mode can only be selected when the expert mode is activated.

Let us assume that you already read in a data set and extracted somefrequencies. The visual inspection seems to indicate that the times are shiftedwith a certain period.

1. Select the periodic time shift modeFirst, activate the expert mode by selecting ’Expert mode’ in the ’File’menu. You will notice that a new menu, ’Options’, has appeared. Thismenu contains the entry ’Set fitting function’ which provides two choices:Standard formula and Standard formula with periodic time shift. Clickon the latter.

2. Set parameters for the periodic time shiftBy inspection of your data you might already have a rough estimate offrequency and amplitude for the periodic time shift. You can either enterthese values into the respective field directly, or use the ’Search PTS startvalues’ button to search for a good start value for the periodic time shiftparameters within a user-defined range of frequencies and amplitudes.

The lower frequency limit is calculated from the time base of the dataset. You should not search for frequencies with lower values since forthese frequencies the time base of the data is too short to allow a reliabledetermination of the periodic time shift.The number of shots refers to the number of initial parameter values thatare being tested.


Figure 65: The ’Search PTS start values’ dialog.

3. Improving the parameters and the fitThe next step is to improve the periodic time shift parameters by means ofa least-squares fit. In order to start this calculation, press ’Improve PTS’.Please note that the common frequency parameters will be kept fixedduring this calculation.

Now it is time to make a least-squares fit that considers all parametersas variable. For this purpose press ’Improve all’. Do not use ’Calculate’,as for a proper fit of a periodic time shift, the frequencies also have tobe redetermined!

Please note:If the start values are not good enough it might occur that the least-squaresalgorithm gets trapped in a false local minimum. Generally, nonlinear fittingrequires some judgment from the user!

Please see section 3.2 for a tutorial that explains the same matter using anexample time string.

4.3.4 Estimation of uncertainties

Period04 provides several tools to calculate the uncertainties of the parametersof a fit:

• Calculation of uncertainties from the error matrix of a least-squares cal-culation

• Monte Carlo Simulation


• Uncertainties calculated from analytically derived formulae assuming anideal case.

Please note:The errors of frequency and phase are correlated. However, by an appropriatechoice of a zero point in time the uncertainties for frequency and phase can bedecoupled. This is the case when

N∑

i=1

ti = 0

(Breger et al., 1999). It is very likely that your data set does not fulfill thiscondition. Therefore, Period04 provides the possibility to shift the data setby the required value in time, for the purpose of determining the uncorrelatedparameter uncertainties when the standard fitting formula is being used.

1. Calculation of uncertainties from the error matrix of a least-squarescalculationPeriod04 applies the curfit routine from Bevington (1969), which is aLevenberg-Marquardt non-linear least-squares fitting procedure. As aby-product of least-squares fits an error matrix is available from whichparameter uncertainties can be calculated. In some cases though (i.e.,when the error matrix is ill-conditioned) this method does not provide agood estimate of the uncertainties. In order to ensure that the calculateduncertainties are reliable, Period04 performs checks for these cases.

The output of common least-squares fits are correlated uncertainties.When the standard fitting formula is being used, Period04 additionallyoffers the possibility to calculate uncorrelated uncertainties.

To calculate the uncertainties of the fit parameters, press ’Calculate LSuncertainties’ in the ’Goodness of Fit’ tab. If you improved frequenciesand phases simultaneously, a dialog will ask you whether you want to un-couple the uncertainties of frequency and phase (in other words: whetheryou want to calculate correlated or uncorrelated uncertainties). After youmade your choice, the uncertainties will be displayed in the text box.

2. Monte Carlo SimulationMonte Carlo simulations are a very reliable way to determine parameteruncertainties. The principle idea is to repeat an experiment (in our casethe optimization routine) on a generated set of samples.For the Monte Carlo simulation Period04 generates a set of time strings.Each data set is created as follows:


• The times of the data points are the same as for the original timestring.

• The magnitudes of the data points are the magnitudes predicted bythe last fit plus Gaussian noise.

For every data set a least-squares calculation will be done. Based on thedistribution of fit parameters the program calculates the uncertainties ofthe parameters.

A short step-by-step guide for making Monte Carlo simulations:

• The Monte Carlo simulation will use the same settings that havebeen used for the last fit. So if you want to determine the uncer-tainties for all parameters, you will have to make a least-squarescalculation with all parameters variable first.

• Now click on ’Monte Carlo Simulation’ in the ’Goodness of Fit’ tab.In the dialog you have to define some settings for the simulation:

Figure 66: The ’Monte Carlo Simulation’ dialog.

Number of processesA ’process’ consists of the creation of time string data and aleast-squares calculation to determine the fit parameters. A lownumber of processes results in a bad estimation of the uncer-tainties. Therefore, to obtain reliable results, a high number ofprocesses is necessary.

Uncouple Frequency and Phase UncertaintiesIf this option is selected, the time string will be shifted in timeso that the ’average time’ is zero. In this case the frequencyand phase uncertainties are no longer correlated. This checkbox will only be visible if this option might be useful, i.e., when


both frequency and phase parameters are fitted using the stan-dard fitting formula.

Use system time to initialize random generatorIf this option is selected the random number generator that isused to create the time string data sets will be initialized withthe current system time.

• Press ’OK’ to start the calculations. Depending on the number oftime points and the number of processes this might be a quite timeconsuming task.

3. Uncertainties calculated from analytically derived formulae assum-ing an ideal caseBased on some assumptions one can derive a formula for the uncertain-ties in frequency, amplitude and phase. See Breger et al. (1999) for thederivation based on a mono periodic fit. If cross terms can be neglectedthen the following equations can also be applied for each pulsation fre-quency separately:

σ(f) =

√

6

N

1

πT

σ(m)

a

σ(a) =

√

2

Nσ(m)

σ(φ) =1

2π

√

2

N

σ(m)

a.

N is the number of time points, T is the time length of the data set,σ(m) denotes the residuals from the fit and a refers to the amplitude ofthe frequency.

To show these parameter uncertainties, select ’Show analytical uncertain-ties’ in the ’Special’ menu.

Please note:This option is only available when using the standard fitting formula.

4.4 General topics

4.4.1 Using weights

Period04 does also handle weighted data. It recognizes three different types ofweights:


• Point weightEach measurement has its own weight.

• Deviation weightFiltering of poor data points.

• Attribute weightsWeights assigned to substrings of each of the four attributes (Date, Ob-server,...)

In order to use weights the user has to set or read in weights AND to ac-tivate them for calculations. This can be done in the ’Weight selection dialog’which can be accessed by the ’Special’ menu or via the shortcut ’Ctrl+W’.

Figure 67: The ’Weight selection’ dialog

Using point weights:

As ’point weight’ we denote weights that are assigned to each of your datapoints. Point weights can only be read in along with time string data. Toinclude these data proceed as described here:

1. Click on the button ’Import time string’ and select the file that containsyour data set in the file-selector dialog.

2. Now another dialog appears. It will show the columns in your data fileand asks you to label these columns correctly. Specify the column thatcontains the point weight data as ’Pnt.weight’ and select appropriatelabels for the other columns too – at least ’Time’ and ’Observed’ arerequired. Press ’OK’.


3. Although your data set has been read in, the program does not yet in-clude the point weights in calculations. They have to be activated in the’Weight selection dialog’ first. So open the dialog, select ’Point weight’and press ’OK’.

From this point the program will use the weighted data until you deselectthis option.

Using deviation weights:

The deviation weights are calculated from the residuals of data points:

residual = |observed − calculated|

deviation weight =

1.0 if residual < cutoff(

cutoffresidual

)2

if residual ≥ cutoff

In order to use deviation weights open the ’Weight selection dialog’, select’Deviation weight’ and set an appropriate cutoff value.

Using attribute weights:

Each of the four list boxes in the ’Time string’ tab represents an attribute.The default names of these attributes are ’Date’, ’Observatory’, ’Observer’ and’Other’. If the time string has been subdivided in at least one attribute it ispossible to give each substring a certain weight.

1. If you have not done so already, read in a time string.

2. Subdivide the time string. Let us assume that you divided the time stringinto several substrings in the attribute ’Date’.

3. Select a substring for which you want to change the weight. Open the’Edit substring properties’ menu by pressing the ’Edit substring’ button atthe bottom of the ’Date’ list box. You can also access this dialog directlyvia the pop-up menu.

4. Type the new weight value into the respective field and press ’OK’.

5. To enable the weights assigned to substrings of attribute ’Date’, openthe ’Weight selection dialog’, select ’Date’ and press ’OK’. Now theseweights will be included in all calculations.


5. Copyright notice

Copyright c©2004-2005 Patrick Lenz, Institute of Astronomy, University of Vi-enna

Permission to use, copy, modify, and distribute this software and its doc-umentation for any purpose is hereby granted without fee, provided that theabove copyright notice, author statement and this permission notice appear inall copies of this software and related documentation.

THE SOFTWARE IS PROVIDED ’AS-IS’ AND WITHOUT WARRANTYOF ANY KIND, EXPRESS, IMPLIED OR OTHERWISE, INCLUDING WITH-OUT LIMITATION, ANY WARRANTY OF MERCHANTABILITY OR FIT-NESS FOR A PARTICULAR PURPOSE. IN NO EVENT SHALL THE INSTI-TUTE OF ASTRONOMY OR THE UNIVERSITY OF VIENNA OR PATRICKLENZ BE LIABLE FOR ANY SPECIAL, INCIDENTAL, INDIRECT OR CON-SEQUENTIAL DAMAGES OF ANY KIND, OR ANY DAMAGES WHATSO-EVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHEROR NOT ADVISED OF THE POSSIBILITY OF DAMAGE, AND ON ANYTHEORY OF LIABILITY, ARISING OUT OF OR IN CONNECTION WITHTHE USE OR PERFORMANCE OF THIS SOFTWARE.

Period04 is based on Period98 (Copyright c©1996-1998 by Martin Sperl).

5.1 Third party software

Period04 makes use of the following third party software:

• Scientific Graphics Toolkit (SGT) - an open source library of Java graph-ics classes.The SGT is provided by NOAA (National Oceanic and Atmospheric Ad-ministration) for full, free and open release and is available athttp:// www.epic.noaa.gov/java/sgt/sgt download.shtml.

• EpsGraphics2D - an open source package for creating high quality EPSgraphicsCopyright c©2001-2004 by Paul James Muttonhttp://www.jibble.org/epsgraphics/

• JavaHelpTM2.0 01Copyright 2003 Sun Microsystems, Inc.http://java.sun.com/products/javahelp


Acknowledgments. We are grateful to M. Sperl for valuable com-ments. This work was supported by the Austrian Fonds zur Forderung derwissenschaftlichen Forschung (Project P17441-N02).

References

Bevington, P. R., 1969, Data reduction and error analysis for the physical sciences,McGraw-Hill (New York)

Breger M., Handler G., Garrido R., et al., 1999, A&A, 349, 225Breger et al., 1993, A&A 271,482Kuschnig et al., 1997, A&A 328, 544Sperl, M. 1998, CoAst 111,1

Communications in Asteroseismology - Universität … · Communications in Asteroseismology Volume...

Documents

Transcript of Communications in Asteroseismology - Universität … · Communications in Asteroseismology Volume...