Skip to content

Latest commit

 

History

History
 
 

README.md

Source codes of IOHexperimenter

[This folder] includes common used codes of IOHexperimenter. For details of IOHprofiler_problem and IOHprofiler_suite, please visit problem page and suite page.

IOHprofiler_transformation

IOHprofiler_transformationdefines methods of transformation applied onIOHprofiler_problem`. For the definition of methods, please visit wikipage

IOHprofiler_random

IOHexperimenter supplies methods of generating random numbers, which can be used to reproduce same experiments with the same random seed.

IOHprofiler_random is a class of random methods, you can construct it with a specified random seed as:

IOHprofiler_random(long seed);

And two random number generator methods are as follow.

  • double IOHprofiler_uniform_rand(), returns a random number from standard uniform distribution.
  • double IOHprofiler_normal_rand(), returns a random number from standard normal distribution.

IOHprofiler_observer

IOHprofiler_observer defines triggers of recording evaluations, and logger classes inherit it to set up the time of recording.

Four strategies of recording evaluations are available,

  • complete tracking, provides the highest granularity, by storing information for each function evaluation. Use set_complete_flag(true) to enable this strategy,
  • interval tracking, stores information for each $\tau$-th function evaluation. Use set_interval() to set $\tau > 0$ to enable this stragety.
  • target-based tracking, stores information for each iteration in which the best-so-far fitness improved. Use set_update_flag(true) to enbale this stragety.
  • time-based tracking, stores information when the user-specified running time budgets are reached. These budget are evenly spaced in log-10 scale.

IOHprofier_class_generator

IOHprofiler_class_generator supplied methods of registering and creating IOHprofiler_problem and IOHprofiler_suite.