CT update
description
Transcript of CT update
![Page 1: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/1.jpg)
CT update
J. HustonMichigan State University
(for CTEQ-TEA collaboration: S. Dulat, J. Gao, T-J Hou, P. Nadolsky, J. Pumplin, D. Stump,
C-P Yuan) PDF4LHC meetingCERN Dec13, 2013
![Page 2: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/2.jpg)
Two recent CT papers
![Page 3: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/3.jpg)
Compare Lagrange Multiplier and Hessian uncertainties for gg->Higgs
curves are LM calculationsof global fit c2 vs Higgs swith (blue) and without (red)‘Tier 2 penalty’
The blue (red) points are theHessian determination of theof the PDF uncertainty with(without) the Tier 2 penalty
Tier 2 penaltyprevents the fitto any one experiment fromdegrading tooMuch
all predictions atNNLO using m=mH
90% Uncertaintiesfrom CT10
![Page 4: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/4.jpg)
Lagrange Multiplier MethodThe Lagrange Multiplier technique is more robust than
the Hessian method in that it does not depend on any assumptions about quadratic behavior of c2
3.45%/2.15%=1.60,close to expected1.645
![Page 5: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/5.jpg)
sH vs c2
contour forPDF+as errorsat 90%CL(0.118+/-0.002)
contour forPDF+as errorsat 68%CL
no Tier 2 penalties
![Page 6: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/6.jpg)
sH vs c2
now with Tier 2 penalties
major effect is at 14 TeV
memorize thesetwo extremepoints for later
![Page 7: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/7.jpg)
PDF+as uncertainties LM estimates of PDF(+as) uncertainties slightly larger than
Hessian determinations, but close, especially for the combined PDF+as errors
Scaling the 90%CL error from the CT10 eigenvector set by a factor of 1.645 agrees well with the LM more exact determination
Note that validates theprescription of addingthe PDF and as errors in quadrature
e.g. 4.7%/1.645=2.85%
![Page 8: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/8.jpg)
CT10H (extreme sets) Choose extreme PDF sets in slide 6 (red squares)
upper uncertainty corresponds to as=0.1194 while lower uncertainty to 0.1167
other two red squares give PDF uncertainty at as=0.118
Note that the two extreme PDF setssaturate the gluondistribution in the x-range relevant for gg->Higgs
sets are availableat CTEQ-TEAwebsite
see also Jun’s talk onMETA-PDFs
![Page 9: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/9.jpg)
CT10 eigenvectors: brief review
Each eigenvector has components along the directions of all of the free parameters in the global fit (26 for CT10NLO)
Sometimes (as in CTEQ6.1) an eigenvector will lie mostly along one particular direction; for CTEQ6.1 and eigenvector 15, that direction saturates the uncertainty for the high x gluon distribution
But not reproduced for example in CT10
![Page 10: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/10.jpg)
A new study (useful to experimentalists)
Largest eigenvalue: nothing much happening
blue band is total PDF uncertainty; curves are contribution from eigenvector
![Page 11: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/11.jpg)
Eigenvector 12
has a fair amount of the gluon uncertainty in thex range for gg->Higgs (as well as at higher x)
![Page 12: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/12.jpg)
Eigenvector 18
Has much of the high x gluon uncertainty
![Page 13: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/13.jpg)
In general…
![Page 14: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/14.jpg)
NLO and NNLOEigenvectors have
similar behavior at NLO and NNLO
Eigenvector 12
NNLO
NLO
![Page 15: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/15.jpg)
Summary
Lagrange multiplier study indicates that Hessian errors scale from 90%CL to 68%CL
for gg->Higgs (not necessarily for other processes)
addition of PDF and as errors in quadrature borne out
LHC data for jets, photons, DY, tT being incorporated into new CT global fits at NLO and NNLO looking forward to full NNLO calculations for
inclusive jet and tT differential cross sections
![Page 16: CT update](https://reader035.fdocuments.us/reader035/viewer/2022062310/56816031550346895dcf4e77/html5/thumbnails/16.jpg)
CMS: 8 TeV jet resultsSMP-12-012-pas