CUTEst.jl documentation

This package provides an interface to CUTEst, the Constrained and Unconstrained Test Environment with safe threads for nonlinear optimization.

This package uses NLPModels.jl, but it also gives direct access to the CUTEst functions.

CUTEst brief history

CUTEst has been around for a while. It started as CUTE, then CUTEr, then CUTEr2, and finally CUTEst. The original project can be used independently of Julia.

CUTEst works by decoding a .SIF file into other files and objects so that a user compiles links that to his code. It also gives the option of doing that for you, in which case you have to send some code to it's folder, and ask for the compilation.

CUTEst gives you about 100 methods to access the objective and constraints functions, as well as their derivatives in many different formats. It also gives access to the problem's information, like number of variables, constraints, the initial point, the bounds, an so on.

Installing

On Linux, you'll need to install wget and gfortran, and possibly have to fix the location of libgfortran.so.

On Ubuntu 14.04, you can do this with

sudo apt-get install wget gfortran
sudo ln -s /usr/lib/gcc/x86_64-linux-gnu/$(gfortran -dumpversion | cut -f1,2 -d.)/libgfortran.so /usr/local/lib

On Ubuntu 16.04,

sudo apt-get install wget gfortran
sudo ln -s /usr/lib/x86_64-linux-gnu/libgfortran.so.3 /usr/local/lib/libgfortran.so

On Archlinux, do

sudo pacman -S wget gfortran

On Windows:

CUTEst does not work on Windows. To use CUTEst from Windows:

Install Windows Subsystem for Linux (WSL);

then in WSL:

Install gfortran;
Install Julia for Linux.

Note, Julia for Linux can be started from VSCode for Windows. In VSCode:

Open a terminal and type WSL;
Start Julia.

The following command should automatically download and install CUTEst and its dependencies (Use ] to enter pkg> mode).

pkg> add CUTEst

Usage

Check the tutorial for complete usage.

The simplest use of CUTEst is through the interface of NLPModels.jl.

using CUTEst, NLPModels

nlp = CUTEstModel("ROSENBR")
println("x0 = $(nlp.meta.x0)")
println("fx = $( obj(nlp, nlp.meta.x0) )")
println("gx = $( grad(nlp, nlp.meta.x0) )")
println("Hx = $( hess(nlp, nlp.meta.x0) )")
finalize(nlp)

x0 = [-1.2, 1.0]
fx = 24.199999999999996
gx = [-215.59999999999997, -87.99999999999999]
Hx = [1330.0 480.0; 480.0 200.0]

Check the NLPModels API for details.

You can pass parameters to sifdecoder by giving additional arguments to CUTEstModel. For instance, to change NH from model CHAIN, use

using CUTEst

for nh = 50:50:200
  nlp = CUTEstModel("CHAIN", "-param", "NH=$nh")
  println("nh = $nh, nnzh = $(nlp.meta.nnzh)")
  finalize(nlp)
end

nh = 50, nnzh = 153
nh = 100, nnzh = 303
tput: No value for $TERM and no -T specified
tput: No value for $TERM and no -T specified
nh = 150, nnzh = 1203
nh = 200, nnzh = 603

Working with CUTEst directly

We also have implemented function to allow access to the CUTEst functions directly: a core API which is only a wrapper for CUTEst. For more information see the section core.