Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of...
Transcript of Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of...
![Page 1: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/1.jpg)
Large Scale Bayesian Analyses of CosmologicalDatasets
Jens JascheGuilhem Lavaux, Fabian Schmidt, Florent Leclercq, Natalia Porqueres,
Doogesh Kodi Ramanah, Supranta Sarma Boruah, Dexter Bergsdal, Adam Andrews, Benjamin Wandelt
Artificial Intelligence in Astronomy
ESO, HQ, Garching, 26 July 2019
Visit us at: www.aquila-consortium.org
![Page 2: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/2.jpg)
A pretty good model...
Figure by Guilhem Lavaux
A set of “second order differential equations”. Weinberg (2009)
![Page 3: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/3.jpg)
A pretty good model...
Figure by Guilhem Lavaux
A set of “second order differential equations”. Weinberg (2009)
What is gravity?
![Page 4: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/4.jpg)
A pretty good model...
Figure by Guilhem Lavaux
A set of “second order differential equations”. Weinberg (2009)
What is gravity?What are the sources of gravity?
![Page 5: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/5.jpg)
A pretty good model...
Figure by Guilhem Lavaux
A set of “second order differential equations”. Weinberg (2009)
What is gravity?What are the sources of gravity?What are the initial conditions?
![Page 6: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/6.jpg)
(t = 0.73 Gyr)
The cosmic large scale structure...
(t = 1.97 Gyr)
(t = 13.83 Gyr)
(t = 5.97 Gyr)
... A source of knowledge!
![Page 7: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/7.jpg)
A large scale Bayesian inverse problem
Bayesian Forward modeling:
Structure formation model
Data model
Prior model Data model
Galaxy bias model
Final StateInitial State Data
See e.g. Neyrinck et al. 2014Ata et al. 2015
Lavaux & Jasche 2016
Jasche, Wandelt (2013)Lavaux, Jasche (2016)Jasche, Lavaux (2018)
![Page 8: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/8.jpg)
MCMC in high dimensions
HMC: Use Classical mechanics to solve statistical problems!
The potential :
see e.g. Duane et al. (1987)Neal (2012)Betancourt (2017)
![Page 9: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/9.jpg)
MCMC in high dimensions
HMC: Use Classical mechanics to solve statistical problems!
The potential :
The Hamiltonian :
see e.g. Duane et al. (1987)Neal (2012)Betancourt (2017)
![Page 10: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/10.jpg)
MCMC in high dimensions
HMC: Use Classical mechanics to solve statistical problems!
The potential :
The Hamiltonian :
see e.g. Duane et al. (1987)
Nuisance parameter!!!
Neal (2012)Betancourt (2017)
![Page 11: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/11.jpg)
MCMC in high dimensions
HMC: Use Classical mechanics to solve statistical problems!
The potential :
The Hamiltonian :
Nuisance parameter!!!
see e.g. Duane et al. (1987)Neal (2012)Betancourt (2017)
![Page 12: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/12.jpg)
MCMC in high dimensions
HMC: Use Classical mechanics to solve statistical problems!
The potential :
The Hamiltonian :
Nuisance parameter!!!
see e.g. Duane et al. (1987)Neal (2012)Betancourt (2017)
![Page 13: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/13.jpg)
MCMC in high dimensions
HMC: Use Classical mechanics to solve statistical problems!
The potential :
The Hamiltonian :
Nuisance parameter!!!
HMC beats the “curse of dimensionality” by:
Exploiting gradients
Using conserved quantities
Randomize and accept :
see e.g. Duane et al. (1987)Neal (2012)Betancourt (2017)
![Page 14: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/14.jpg)
The data model
BORG (Bayesian Origin Reconstruction from Galaxies)
Incorporates physical model into Likelihood (2LPT / PM)
Turn inference into initial conditions problem: Find !!!
Model sensitivity
Poisson Intensity:
Also see e.g. Jasche et al. 2015 ( arXiv:1409.6308 ) / Wang et al. 2014 (arXiv:1407.3451)
![Page 15: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/15.jpg)
The data model
BORG (Bayesian Origin Reconstruction from Galaxies)
Incorporates physical model into Likelihood (2LPT / PM)
Turn inference into initial conditions problem: Find !!!
Model sensitivity
Poisson Intensity:
Also see e.g. Jasche et al. 2015 ( arXiv:1409.6308 ) / Wang et al. 2014 (arXiv:1407.3451)
![Page 16: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/16.jpg)
The data model
BORG (Bayesian Origin Reconstruction from Galaxies)
Incorporates physical model into Likelihood (2LPT / PM)
Turn inference into initial conditions problem: Find !!!
Poisson Intensity:
Also see e.g. Jasche et al. 2015 ( arXiv:1409.6308 ) / Wang et al. 2014 (arXiv:1407.3451)
![Page 17: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/17.jpg)
The data model
BORG (Bayesian Origin Reconstruction from Galaxies)
Incorporates physical model into Likelihood (2LPT / PM)
Turn inference into initial conditions problem: Find !!!
Model sensitivity
Poisson Intensity:
Also see e.g. Jasche et al. 2015 ( arXiv:1409.6308 ) / Wang et al. 2014 (arXiv:1407.3451)
![Page 18: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/18.jpg)
Adjoint coding
The problem: is a computer program, not analytic!!!
Finite differencing not feasible, too high-d!!!
Any computer program is a sequence of elementary operations
➔ Use chain rule
Result:
Sensitivity Matrix of your computer model
We need the adjoint, so do a transpose
Matrix cannot be stored, so use operator formalism
line by line derivative of your computer code!!BEWARE of if-switches!!!
![Page 19: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/19.jpg)
Adjoint coding
The problem: is a computer program, not analytic!!!
Finite differencing not feasible, too high-d!!!
Any computer program is a sequence of elementary operations
➔ Use chain rule
Result:
Sensitivity Matrix of your computer model
We need the adjoint, so do a transpose
Matrix cannot be stored, so use operator formalism
line by line derivative of your computer code!!BEWARE of if-switches!!!
![Page 20: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/20.jpg)
Adjoint coding
The problem: is a computer program, not analytic!!!
Finite differencing not feasible, too high-d!!!
Any computer program is a sequence of elementary operations
➔ Use chain rule
Result:
Sensitivity Matrix of your computer model
We need the adjoint, so do a transpose
Matrix cannot be stored, so use operator formalism
line by line derivative of your computer code!!BEWARE of if-switches!!!
![Page 21: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/21.jpg)
Adjoint coding
The problem: is a computer program, not analytic!!!
Finite differencing not feasible, too high-d!!!
Any computer program is a sequence of elementary operations
➔ Use chain rule
Result:
Sensitivity Matrix of your computer model
We need the adjoint, so do a transpose
Matrix cannot be stored, so use operator formalism
line by line derivative of your computer code!!BEWARE of if-switches!!!
![Page 22: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/22.jpg)
Adjoint coding
The problem: is a computer program, not analytic!!!
Finite differencing not feasible, too high-d!!!
Any computer program is a sequence of elementary operations
➔ Use chain rule
Result:
Sensitivity Matrix of your computer model
We need the adjoint, so do a transpose
Matrix cannot be stored, so use operator formalism
line by line derivative of your computer code!!BEWARE of if-switches!!!
![Page 23: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/23.jpg)
BORG3: A Modular statistical programing engine
Build flexible data models
Hierachical Bayes and block sampling
Jasche & Lavaux (2019, A&A)
![Page 24: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/24.jpg)
www.aquila-consortium.org
A detailed and physicallyplausible model
ofThe Nearby Universe
![Page 25: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/25.jpg)
New insights into the nearby universe
Jasche & Lavaux (2019, A&A)
Application: BORG – 2M++ (Lavaux & Hudson (2011, MNRAS))
● Domain: (677.7 Mpc/h)3
● IC f luctuation elements: 2563
● Simulation particles: 5123
● LSS model: Particle Mesh Solver
Some samples from the Markov Chain...
Leclercq et al. (2019, in prep)
![Page 26: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/26.jpg)
Inferred mass density
Leclercq et al. (2019, in prep)
Preliminary results!Inferred mass density in super-galactic plane:
![Page 27: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/27.jpg)
Estimating Cluster masses
Coma Mass Profile
Jasche & Lavaux (2019, A&A)
![Page 28: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/28.jpg)
Estimating Cluster masses
Coma Mass Profile
Virgo Mass ProfileJasche & Lavaux (2019, A&A)
![Page 29: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/29.jpg)
Estimating Cluster masses
Coma Mass Profile
Shapley Mass Profile
Virgo Mass ProfileJasche & Lavaux (2019, A&A)
![Page 30: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/30.jpg)
Dynamics of the Nearby Universe
Jasche & Lavaux (2019, A&A)
Leclercq et al. (2019, in prep)
Radial velocities Counts of DM streams
![Page 31: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/31.jpg)
Dynamics of the Nearby Universe
![Page 32: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/32.jpg)
www.aquila-consortium.org
Going deeper:
Analyzing
Sloan Digital Sky Survey III
DR12 (z < 0.7)
![Page 33: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/33.jpg)
Application to SDSS III DR12
First Attempt!!!
![Page 34: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/34.jpg)
Application to SDSS III DR12
First Attempt!!!
Kalus, Percival et al. (2018, MNRAS)
![Page 35: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/35.jpg)
Foreground contaminationsForegrounds hamper cosmological inference ( see e.g. Leistedt & Peiris (2014))
Star densities
Sky fluxes
Jasche & Lavaux (2017, A&A)
![Page 36: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/36.jpg)
Designing robust likelihoods
Porqueres et al (2019, A&A)
Map of patches on the sky Extruded into 3d volume
![Page 37: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/37.jpg)
Designing robust likelihoods
Porqueres et al (2019, A&A)
Map of patches on the sky Extruded into 3d volume
![Page 38: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/38.jpg)
Application to SDSS III DR12
Lavaux et al (2019, in prep)
Preliminary results!Real data application LOWZ + CMASS
Ensemble mean density Posterior power-spectrum
Application: BORG – SDSS III● Domain: (4000 Mpc/h)3
● IC f luctuation elements: 2563
● Simulation particles: 5123
● LSS model: Lagrangian Perturbation Theory
![Page 39: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/39.jpg)
Independent test of inferred mass
CMB Lensing:
● Correlation Planck 15 vs. BORG SDSS3 convergence map
Preliminary results!
Jasche et al (2019 in prep)
![Page 40: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/40.jpg)
What are the foregrounds?
Lavaux et al (2019, in prep)
![Page 41: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/41.jpg)
Cosmological parameter via the A/P test
Kodi Ramanah et al (2019, A&A)
Using the Alcock-Paczynski cosmological test
![Page 42: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/42.jpg)
Cosmological parameter via the A/P test
Kodi Ramanah et al (2019, A&A)
![Page 43: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/43.jpg)
Cosmological parameter via the A/P test
Kodi Ramanah et al (2019, A&A)
Application to SDSS III mock data:
![Page 44: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/44.jpg)
Cosmological parameter via the A/P test
Source of Information (Kodi Ramanah et al (2019, in prep)):● Complete use of modes● Exploitation of higher order statistics
Kodi Ramanah et al (2019, A&A)
SDSS III BAOs
BORG / ALTAIR
![Page 45: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/45.jpg)
Summary & Conclusion
BORG combines physical modeling with data science:
Dynamical modeling accounts for non-Gaussian statistics
Flexible data modeling via HMC and block sampling
Solves complex high dimensional statistics problems
Scientific results:
Characterization of initial conditions
Accurate & Detailed reconstructions of the DM field
Complementary mass estimates
Dynamical reconstructions
Inference of cosmological parameter
![Page 46: Large Scale Bayesian Analyses of Cosmological Datasets€¦ · Large Scale Bayesian Analyses of Cosmological Datasets Jens Jasche Guilhem Lavaux, Fabian Schmidt, Florent Leclercq,](https://reader034.fdocuments.us/reader034/viewer/2022042118/5e97a335a666cb7ec93295f1/html5/thumbnails/46.jpg)
The end...
Thank You!
Selected Bibliography:
Porqueres et al A&A 624, 115 (2019).
Ramanah et al A&A 621, 69 (2019).
Jasche, J. & Kitaura, F. S. MNRAS 407, 29–42 (2010).
Jasche, J., Kitaura, F. S., Li, C. & Enßlin, T. A. MNRAS 409, 355–370 (2010).
Jasche, J. & Wandelt, B. D. MNRAS 425, 1042–1056 (2012).
Jasche, J. & Wandelt, B. D. MNRAS 432, 894–913 (2013).
Jasche, J., Leclercq, F. & Wandelt, B. D. JCAP 01, 036 (2015).
Lavaux, G. & Jasche, J. MNRAS 455, 3169–3179 (2016).
Visit us at: www.aquila-consortium.org