Use of the class `problem`#

The problem class represents a generic optimization problem. The user codes the details of such a problem in a separate class (the user-defined problem, or UDP) which is then passed to problem that provides a single unified interface.

Note

The User Defined Problems (UDPs) are optimization problems (coded by the user) that can be used to build a pygmo object of type problem

Some UDPs (optimization problems) are already provided with pygmo and we refer to them as pygmo UDPs.

For the purpose of this tutorial we are going to use a pygmo UDP called rosenbrock to show the basic construction of a problem, but the same logic would also apply to a custom UDPs, that is a UDP that is actually coded by the user.

Let us start:

>>> import pygmo as pg
>>> prob = pg.problem(pg.rosenbrock(dim = 5))
>>> print(prob) 
Problem name: Multidimensional Rosenbrock Function
    C++ class name: ...

    Global dimension:                       5
    Integer dimension:                      0
    Fitness dimension:                      1
    Number of objectives:                   1
    Equality constraints dimension:         0
    Inequality constraints dimension:       0
    Lower bounds: [-5, -5, -5, -5, -5]
    Upper bounds: [10, 10, 10, 10, 10]
    Has batch fitness evaluation:  false

    Has gradient: true
    User implemented gradient sparsity: false
    Expected gradients: 5
    Has hessians: false
    User implemented hessians sparsity: false

    Fitness evaluations: 0
    Gradient evaluations: 0

    Thread safety: constant

In the code above, after the trivial import of the pygmo package, we define a variable prob by constructing a problem from rosenbrock, a multidimensional problem constructed from its global dimensions. In the following line we inspect the problem We can see, at a glance, that the UDP rosenbrock is a five dimensional, single objective, box constrained problem for which neither gradients nor hessians nor sparsity information is provided in the UDP.

We also see that its fitness function has never been called hence the counter for fitness evaluations is zero.

All of the information contained in the problem print out can be retrieved using the appropriate methods, for example:

>>> prob.get_fevals() == 0
True

Lets check how a fitness computation increases the counter:

>>> prob.fitness([1,2,3,4,5])
array([14814.])
>>> prob.get_fevals() == 1
True

We may also get back the UDP, and thus access all the methods not exposed in the problem interface at any time via the extract() method:

>>> udp = prob.extract(pg.rosenbrock)
>>> type(udp)
<class 'pygmo.core.rosenbrock'>
>>> udp = prob.extract(pg.rastrigin)
>>> udp is None
True

Such an extraction will only work if the correct UDP type is passed as argument.

Use of the class problem

Use of the class problem#

Use of the class `problem`#