Techn. Fakultät Website deprecated and outdated. Click here for the new site. FAU-Logo

CUDA Quasi-Newton Optimization


This is a C++ library for general-purpose, unconstrained non-linear optimization on the GPU using NVIDA's CUDA. It implements the L-BFGS (“Limited-memory Broyden-Fletcher-Goldfarb-Shanno“) method, a popular Quasi-Newton optimization method with a low memory footprint.

The source code package already comes with some example applications and test cases. We also use this library in our GPU-accelerated super-resolution framework for non-linear optimization. More details about this application are available Opens internal link in current windowhere



  • 03/04/2013 Initial version released (1.0)
  • 04/23/2013 Bug fix to support CUDA 5.x


  • CUDA L-BFGS library is available Initiates file downloadhere
  • Please follow the instructions in README.txt on how to build, install and use the library


This work is licensed under a Creative Commons Attribution 3.0 Unported License. (CC-BY)
The authors do not assume liability for errors contained in or for damages arising from the use of the software.

There is no publication available that covers this library exclusively, but you may consider citing the paper it was introduced in:

Articles in Conference Proceedings