kornelz / localoptimizationea

NumBBO/CoCO Framework in Python (Experimental Part)

Prerequisites

The simplest way to check the prerequisits is to go directly to Getting Started below and give it a try. Then act upon failure, as in this case probably one of the following is lacking:

• Python is installed (version >=2.6). If this is not the case, check out Anaconda, as this provides additionally all Python packages necessary to run the COCO postprocessing as well as the ipython shell. The postprocessing needs currently Python < 3.0, i.e. 2.6 or 2.7.

• make is installed and works with one of the MakeFiles provided in folder code-experiments/build/c. You might type make within the build/c folder to see whether this works.

• A C compiler, like gcc, which is invoked by make.

Getting Started

See Getting Started here for the first steps. Then

• Copy (and rename) example_experiment_for_beginners.py or example_experiment.py to a place (and name) of your choice. Modify this file to include the solver of your choice (instead of random_search) to the benchmarking framework. Do not forget to also choose the right benchmarking suite and the corresponding observer.

• Execute the modified file either from a system shell like

      python example_experiment_new_name.py
  

  or in ipython (highly recommended for debugging, see Anaconda at prerequisites) with the run command like

      ipython
      Python 2.7.11 ...
      
      In [1]: run example_experiment_new_name.py
      
      Benchmarking solver...
      [...]
      bbob-biobj done (3300 of 3300 problems benchmarked),...
  

  or (with example_experiment.py only) in python or ipython like

      >>> import example_experiment_new_name as ee
      >>> ee.main()
  

See here for the remaining steps to investate the resulting data.

Details

• python do.py run-python builds, installs (locally) and tests the cocoex Python module, and runs python code-experiments/build/python/example_experiment.py. You will see something like this:

    AML	['code-experiments/src/coco_generics.c', 'code-experiments/src/coco_random.c', 'code-experiments/src/coco_suite.c', 'code-experiments/src/coco_observer.c', 'code-experiments/src/coco_runtime_c.c'] -> code-experiments/build/python/cython/coco.c
    COPY	code-experiments/src/coco.h -> code-experiments/build/python/cython/coco.h
    COPY	code-experiments/src/bbob2009_testcases.txt -> code-experiments/build/python/bbob2009_testcases.txt
    EXPAND	code-experiments/build/python/README.md to code-experiments/build/python/README.txt
    EXPAND	code-experiments/build/python/setup.py.in to code-experiments/build/python/setup.py
    PYTHON	setup.py install --user in code-experiments/build/python
    RUN	python coco_test.py in code-experiments/build/python
    RUN	python example_experiment.py in code-experiments/build/python

• The Python interface code, file interface.c, has been generated with Cython. Under normal circumstance this C-file is delivered with the library and does not need to be generated by the user.

---

^1: you will see something like this:

    Benchmarking solver '<function random_search' with budget=10 * dimension, Fri Jan 15 18:23:35 2016
    Simple usecase ...
    bbob-biobj done (1650 of 1650 problems benchmarked), Fri Jan 15 18:23:38 2016 (0.05 min).