Skip to content

Latest commit

 

History

History
72 lines (54 loc) · 2.81 KB

README.md

File metadata and controls

72 lines (54 loc) · 2.81 KB
Raw

TulipCL

App for building a command-line executable of the Tulip.jl interior-point solver. All commands shown below are executed in the Tulip.jl/app/ directory.

Installation instructions

  1. Download and install Julia (version 1.3.1 or newer).

  2. Install PackageCompiler

    $ julia -e 'using Pkg; Pkg.add("PackageCompiler")'

  3. Instantiate the current environment

    $ julia --project=. -e 'using Pkg; Pkg.instantiate()'

  4. Build the command-line executable

    $ julia -q --project=.
julia> using PackageCompiler
julia> create_app(".", "tulip_cl", force=true, precompile_execution_file="precompile_app.jl", app_name="tulip_cl");
julia> exit()

    The executable will be located at tulip_cl/bin/tulip_cl.

    While building the command-line executable, a precompilation script is executed wherein a small number of problems will be solved, thereby producing a number of logs. This process is normal. For more information on how to build the app, take a look at the PackageCompiler documentation.

Using a different version of Tulip

Following the completion of step 3. above, this directory will contain a Manifest.toml that specifies the version of all packages, including that of Tulip. By default, this will be the most recently tagged version.

To build the executable with a different version/branch of Tulip follow these instructions, tag the particular version/branch before creating the app.

For instance, to try out a local development branch, running

julia> ]
pkg> dev ..

will use the current state of the Tulip.jl repository.

Finally, follow the last step in the above installation guide to generate the command-line executable.

Running the command-line executable

Once the build step is performed, the executable can be called from the command line as follows:

cd tulip_cl/bin
./tulip_cl [options] finst

where finst is the problem file.

For instance,

tulip_cl --Threads 1 --TimeLimit 3600 afiro.mps

will solve the problem afiro.mps using one thread and up to 1 hour of computing time.

Currently, possible user options are

Option name Type Default Description
Presolve Int 1 Set to 0 to disable presolve, 1 to activate it
Threads Int 1 Maximum number of threads
TimeLimit Float64 Inf Time limit, in seconds
IterationsLimit Int 500 Maximum number of barrier iterations
Method String HSD Interior-point method

For more information, run tulip_cl --help, or look at Tulip's documentation for more details on parameters.