Skip to content

Commit 9ce1832

Browse files
committed
Announce our July monthly meeting.
1 parent 9e16a1b commit 9ce1832

File tree

1 file changed

+38
-0
lines changed

1 file changed

+38
-0
lines changed

_data/meetinglist.yml

Lines changed: 38 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,41 @@
1+
- name: "CaaS Monthly Meeting"
2+
date: 2024-07-11 17:00:00 +0200
3+
time_cest: "17:00"
4+
connect: "[Link to zoom](https://princeton.zoom.us/j/94396923932?pwd=9MDrbK3nQgGYjqkEZILGO7ne4YczA9.1)"
5+
label: caas_11Jul2024
6+
agenda:
7+
- title: "Automatic Differentiation in Roofit for Fast and Accurate Likelihood Fits"
8+
speaker:
9+
image: "https://compiler-research.org/images/team/Vaibhav.jpg"
10+
name: "Vaibhav Thakkar"
11+
bio: |
12+
Vaibhav Thakkar is currently a research Intern at CERN, working with
13+
the compiler research group at Princeton University (US) on the
14+
project of compile-time automatic differentiation (Clad) in C++ for
15+
scientific computing applications. Previously, he earned my bachelor's
16+
in technology in Electrical Engineering from IIT Kanpur in India and
17+
worked as a software engineer for a couple of years after graduation.
18+
description: |
19+
RooFit is a widely used data analysis library in High-energy physics
20+
experiments at CERN, which requires minimizing likelihoods with
21+
thousands of free parameters spread over hundreds of likelihood
22+
components. To iteratively find the minimum of a likelihood, one
23+
generally has to know the gradient of the likelihood with respect to all
24+
free parameters. Previously, these were calculated using numerical
25+
differentiation, which requires varying one parameter at a time and
26+
reevaluating the full likelihood. Although complete reevaluation can be
27+
prevented by a caching mechanism, it incurs a significant overhead in
28+
bookkeeping. This talk discusses the recent integration of automatic
29+
differentiation (AD) in RooFit for these gradient computations using
30+
Clad, a compile-time AD library for C++ codebases. We also present the
31+
efficiency improvements with the recently released Atlas and CMS Higgs
32+
models.
33+
- title: Update
34+
speaker: Vassil Vassilev
35+
- title: Next meeting
36+
speaker: Vassil Vassilev
37+
link: 1 Aug 1700
38+
139
- name: "CaaS Monthly Meeting"
240
date: 2024-06-06 17:00:00 +0200
341
time_cest: "17:00"

0 commit comments

Comments
 (0)