• BFGS optimization algorithm implemented in C++, integrated in the high-precision arithmetic package ARPREC
• Download pBFGS source code
• Download NIPS 2012 Probabilistic Numerics workshop paper

Reference: "KL-based Control of the Learning Schedule for Surrogate Black-Box Optimization" Ilya Loshchilov, Marc Schoenauer and Michele Sebag. Eprint arXiv:1308.2655. 08/2013 http://arxiv.org/pdf/1308.2655v1.pdf

Comparative performance of KL-ACM-ES compared to high-precision BFGS and CMA-ES variants (see text) on the 20-dimensional Rosenbrock function fRos, f2Ros and f4Ros. The medium number of function evaluations (out of 15 runs) to reach the target objective value 1e-8 was computed and the corresponding run is shown. Markers show the objective value reached in each run after a given number of function evaluations..