nloptr 2.0.0

A new version of the nloptr package has been released! The NLopt library has switched to CMake as build system. I am replacing Jelmer Ypma as maintainer of the package.

software
optimization
Author
Affiliation

Aymeric Stamm

Department of Mathematics Jean Leray, UMR CNRS 6629

Published

January 28, 2022

A new version of the nloptr package has been released! The NLopt library has switched to CMake as build system. I am replacing Jelmer Ypma as maintainer of the package.

nlopt

The NLopt library (Johnson 2021) is a free/open-source library for nonlinear optimization, providing a common interface for a number of different free optimization routines available online as well as original implementations of various other algorithms.

Starting from v2.5.0, NLopt has switched to cmake as build system, making the latest versions incompatible with the way nloptr used to interface the library.

nloptr

As of v2.0.0, we now provide an R interface to nlopt >= 2.7.0. This is achieved in different ways depending on the operating system you are using.

Windows

There is no dependency required for Windows users.

  • If you are running on R <= 4.1.x, installation of nloptr will rely on a valid internet connexion to download the nlopt 2.7.1 headers and static library from rwinlib and use it.
  • If you are running on R >= 4.2.x, installation of nloptr will directly link with nlopt 2.7.1 from the Rtools42 toolchain.

Linux and macOS

On Unix-like systems, installation of nloptr will go through the following steps:

Is pkg-config available on your PATH?

  • NO: compile nlopt 2.7.1 from included sources using cmake >= 3.15.0.

  • YES: can pkg-config detect a system build of nlopt >= 2.7.0?

    • NO: compile nlopt 2.7.1 from included sources using cmake >= 3.15.0.
    • YES: install nloptr using the detected system build of nlopt.

Hence, cmake >= 3.15.0 can become a system requirement for installing nloptr on your machine if you are running on a Unix-like system without pkg-config or without a system build of nlopt >= 2.7.0.

References

Johnson, Steven G. 2021. “The NLopt Nonlinear-Optimization Package.” https://nlopt.readthedocs.io/en/latest/.