Bayesian Inference in High-Energy Physics

25 May 2022, 09:00 → 27 May 2022, 18:00 Europe/London

OC218 (IPPP)

Ian Vernon (Durham University), Jernej Kamenik (Jozef Stefan Institute, Ljubljana), Manuel Szewc (ICAS-UNSAM & IJS), Michael Spannowsky (IPPP, Durham University), Stephen Jones (IPPP, Durham), ezequiel alvarez (ICAS, UNSAM)

---

Bayesian inference methods consist in finding out probability distribution over unknowns given some observations and some knowledge of the model that connects them.  These techniques not only look very suitable for inquiring Nature in High-Energy Physics, but also they have been highly boosted and developed in the last decades through a broad spectrum of Machine Learning applications.  Our challenge is to reach the state-of-the-art of these techniques within High-Energy Physics applications to study their scope in better understanding non-perturbative phenomena, improving Monte Carlo techniques, improving experimental setups, modelling and eventually finding New Physics, and many other LHC-related disciplines, as well as other HEP areas.

The workshop consists in mainly three aspects, on which we aim to trigger discussions that construct connecting and developing bridges.  From one side, the known and in-research HEP cases, in theory, phenomenology and experiment,  in which inference methods provide relevant results.  From another side, how Monte Carlo simulations could be assisted through Bayesian methods.  And from the final side, we want to discuss with Computer Science specialists who foster the latest developments and tools on how these can be best used in HEP.

The workshop is in-person at the Durham IPPP (UK) and we aim to have ~25 participants.  Sessions will mix talks, coffee, and discussion slots.  If you are interested in attending the workshop please apply by registering here.

---

Follow us on twitter @IPPP_Durham

Wednesday, 25 May

1 Durham Locals Session: Intro to Bayesian Inference (for our local PhD / PostDoc / Staff)

Attendance from workshop participants is welcome but not expected. This session is aimed at introducing some of the basic ideas of Bayesian Inference to our local PhD/ PostDocs and interested staff members.

2 Workshop Official Start & Welcome

Results talks

Talks on existing papers, results, etc. by some of its authors

3 Mapping Machine-Learned Physics into a Human-Readable Space

Speaker: Taylor Faucett (UC, Irvine)

4 Bayesian Inference of QCD splittings with Invertible Networks

Speaker: Theo Heimel (U. Heidelberg, ITP)

5 How well do we know the neutron-matter equation of state at the densities inside neutron stars? A Bayesian approach with correlated uncertainties

Speaker: Jordan Melendez (The Ohio State University)

melendez_durham_2022.pdf

15:45 Coffee Break

6 A graphical multi-fidelity Gaussian process model, with application to emulation of expensive computer simulations

Speaker: Steffen Bass (Duke University)

Bass_IPPP_2022.pdf

7 Exploring phase space with Nested Sampling

Speaker: David Yallup (University of Cambridge)

dy_bayeshep_durham.pdf

Thursday, 26 May

Computer Science and Statistics: Mathematical aspects of Bayesian inference, by the Statistics group @ Durham

Talks on Computer Science tools

Convener: Prof. Ian Vernon (Durham)

8 Fully Bayesian parameter inference for Markov processes

Speaker: Darren Wilkinson (Department of Mathematical Sciences at Durham University)

djw-bayes-hep.pdf

9 Bayes linear strategies for emulation and history matching for complex computer models

Speaker: Michael Goldstein (Department of Mathematical Sciences at Durham University)

Hep_HM.pdf

10 Multilevel Emulation of Complex Computer Models, with application to EAGLE, a Galaxy Formation Simulation

Speaker: Ian Vernon (Department of Mathematical Sciences at Durham University)

Bayesian_Inference_for_HEP_May_2022.pdf

berger-sellke_pvalues_irreconcilability.pdf

11:00 Coffee break

11 Enforcing stationarity through the prior in vector autoregressions

Speaker: Sarah Heaps (Department of Mathematical Sciences at Durham University)

slides.pdf

12 Contemporaneous MCMC

Speaker: Louis Aslett (Department of Mathematical Sciences at Durham University)

220526.pdf

13:00 Buffet Lunch

Topics talks on Bayesian Inference in HEP

Talks on HEP topics from the Bayesian Inference point of view

13 Differentiable programming and its applications

Speaker: Lukas Heinrich (Technical University Munich)

14 HEP Simulator issues tackled with Bayesian Inference

Speaker: Manuel Szewc (IJS & ICAS-UNSAM)

ippp_talk_szewc.pdf

16:00 Coffee Break

15 Quantifying Uncertainty in Deep Learning

Speaker: Mike Walmsley (University of Manchester)

16 The inverse problem challenge in lattice QCD

Speaker: Alexander Rothkopf (University of Stavanger)

IPPP_Bayes_Lattice_v1.pdf

19:00 Conference Dinner

Friday, 27 May

Topics talks on Bayesian Inference in HEP

Talks on HEP topics from the Bayesian Inference point of view

17 Causal Inference

Speaker: Ezequiel Alvarez (ICAS, UNSAM)

18 Representation Learning for HEP

Speaker: Barry Dillon (University of Heidelberg)

19 Bayesian community criteria for anomaly evaluation

Speaker: Jernej Kamenik (Jozef Stefan Institute & Ljubljana University)

11:00 Coffee

Open discussion

Conference summary and open discussion