This package provides an interface to CUTEst, a repository of constrained and unconstrained nonlinear programming problems for testing and comparing optimization algorithms, derived from the abstract model on NLPModels.jl.
If you use CUTEst.jl in your work, please cite using the format given in CITATION.cff.
This package will automatically install the CUTEst binaries for your platform.
The gfortran
compiler is required to compile decoded SIF problems.
Users on all platforms except Windows must install it to use CUTEst.jl
.
For Windows users, a small artifact containing gfortran.exe
is installed automatically.
No other Fortran compiler is supported.
The following command installs the CUTEst binaries and the Julia interface:
pkg> add CUTEst
After installation, you can create instances of NLPModels
models using the CUTEstModel
constructor:
using CUTEst
nlp = CUTEstModel{Float64}("BYRDSPHR")
This model supports the same functions as other NLPModels, for example:
using NLPModels
fx = obj(nlp, nlp.meta.x0)
gx = grad(nlp, nlp.meta.x0)
Hx = hess(nlp, nlp.meta.x0)
cx = cons(nlp, nlp.meta.x0)
Jx = jac(nlp, nlp.meta.x0)
Problems can also be instantiated in single and quadruple precision:
using CUTEst, Quadmath
nlp_single = CUTEstModel{Float32}("BYRDSPHR")
nlp_quadruple = CUTEstModel{Float128}("BYRDSPHR")
To retrieve the list of available SIF problems, use the function list_sif_problems
:
using CUTEst
available_problems = list_sif_problems()
If you want to retrieve only problems with specific properties, you can use the function select_sif_problems
:
using CUTEst
filtered_problems = select_sif_problems(; min_var=10, max_var=100, only_linear_con=true)
You can check an Introduction to CUTEst.jl on our site.