GooFit: A library for massively parallelising maximum-likelihood fits

Fitting complicated models to large datasets is a bottleneck of many analyses. We present GooFit, a library and tool for constructing arbitrarily-complex probability density functions (PDFs) to be evaluated on nVidia GPUs or on multicore CPUs using OpenMP. The massive parallelisation of dividing up event calculations between hundreds of processors can achieve speedups of factors 200-300 in real-world problems.

Note:

Presented at the CHEP 2013 conference

programming
multiprocessor: graphics
data management
statistics

References(5)

Figures(0)

[1]

The RooFit toolkit for data modeling

Wouter Verkerke
(
- UC, Santa Barbara
)
,
David P. Kirkby
(
- UC, Irvine
)

- eConf C0303241 (2003) MOLT007,
- ,
•
e-Print:
- physics/0306116

[2]

NVIDIA CorpNVIDIA Practices Guide - CUDA toolkit 2.3 URL

CUDA C. Programming Best

[3]

GPU Computing Gems: Jade Edition

Bell N
,
Hoberock J

[4]

SIAM Journal on Optimization 1

Davidon W

[5]

CERN Program Library Long Writeup D506

James F