package lbfgs

  1. Overview
  2. Docs
Bound-constrainted optimization in many variables

Install 

dune-project
 Dependency

github.com Readme Changelog LGPL-3.0-only WITH OCaml-LGPL-linking-exception License Edit opam file Versions (5)

Authors

  1. C
    Christophe Troestler <Christophe.Troestler@umons.ac.be>

Maintainers

  1. C
    Christophe Troestler <Christophe.Troestler@umons.ac.be>

Sources

lbfgs-0.9.5.tbz
sha256=5c4f3c0fec90f589a1ea36cddc0a7be4a8a55d4d4dd3815150400af6c514242d
sha512=9663e317a25dae126c1066328ae120a7a9f7a383ee6669c56bf26fb2b6629bffcceec8a768556f6dc275a1069c2227cac4076cfdcc7e52d01e2bca55e8ec1846

Description

This library provide an algorithm to minimize and maximize functions of many variables on bounded or unbounded domains. This is a binding to L-BFGS-B, a library for Large-scale Bound-constrained Optimization.

Tags

science numerics optimization minimization maximization bound-constrained large-scale

Published: 18 Oct 2023

README

Build Status

OCaml bindings for L-BFGS

L-BFGS is an optimization algorithm in the family of quasi-Newton methods that approximates the Broyden–Fletcher–Goldfarb–Shanno (BFGS) algorithm using a limited amount of computer memory.

This library is a binding to Nocedal's implementation of L-BFGS-B which adds the possibility of setting bounds on the variables.

Install

The easiest way to install the library is to use opam:

opam install lbfgs

If you clone this repository, download Lbfgsb.3.0 and extract it in src/ (it should create src/Lbfgsb.3.0/). Then issue make and make install.

In case the right FORTRAN compiler for your platform is not automatically found, you can specify it explicitly by exporting the FORTRANC variable before invoking opam. For example

    export FORTRANC=/usr/bin/x86_64-w64-mingw32-gfortran.exe
opam install lbfgs

Documentation

If you cloned this repository, issue

dune build @doc

You can also consult the interface or online.

Dev Dependencies (1)

  1. lacaml with-test

Used by (2)

  1. oml < "0.0.7"
  2. owl-opt-lbfgs

Conflicts

None