DesignSpace creation and manipulation

In this example, we will see how to create and how to manipulate an instance of DesignSpace.

from __future__ import annotations

from gemseo.api import configure_logger
from gemseo.api import create_design_space
from numpy import array

configure_logger()

<RootLogger root (INFO)>

Create a design space

The user can create an instance of the DesignSpace using the API and the create_design_space() function.

design_space = create_design_space()

Add design variables

The user can add new design variables using the DesignSpace.add_variable(). In the following example, we add the x variable in the design space. We also define the lower and upper bound of the variable. It is then possible to plot the DesignSpace instance either using a print statement or by using the logger.

design_space.add_variable("x", l_b=array([-2.0]), u_b=array([2.0]), value=array([0.0]))

print(design_space)

Design space:
+------+-------------+-------+-------------+-------+
| name | lower_bound | value | upper_bound | type  |
+------+-------------+-------+-------------+-------+
| x    |      -2     |   0   |      2      | float |
+------+-------------+-------+-------------+-------+

The user can also add design variables with dimension greater than one. To do that, the user can use the size keyword:

design_space.add_variable(
    "y", l_b=array([-2.0, -1.0]), u_b=array([2.0, 1.0]), value=array([0.0, 0.0]), size=2
)
print(design_space)

Design space:
+------+-------------+-------+-------------+-------+
| name | lower_bound | value | upper_bound | type  |
+------+-------------+-------+-------------+-------+
| x    |      -2     |   0   |      2      | float |
| y[0] |      -2     |   0   |      2      | float |
| y[1] |      -1     |   0   |      1      | float |
+------+-------------+-------+-------------+-------+

By default, each variable infers its type from the given values. One may also specify it with the var_type keyword

design_space.add_variable(
    "z",
    l_b=array([0, -1]),
    u_b=array([3, 1]),
    value=array([0, 1]),
    size=2,
    var_type="integer",
)
design_space.add_variable(
    "w",
    l_b=array([-2, -5]),
    u_b=array([3, 1]),
    value=array([2, -2]),
    size=2,
    var_type=["integer", "integer"],
)
print(design_space)

Design space:
+------+-------------+-------+-------------+---------+
| name | lower_bound | value | upper_bound | type    |
+------+-------------+-------+-------------+---------+
| x    |      -2     |   0   |      2      | float   |
| y[0] |      -2     |   0   |      2      | float   |
| y[1] |      -1     |   0   |      1      | float   |
| z[0] |      0      |   0   |      3      | integer |
| z[1] |      -1     |   1   |      1      | integer |
| w[0] |      -2     |   2   |      3      | integer |
| w[1] |      -5     |   -2  |      1      | integer |
+------+-------------+-------+-------------+---------+

Note

Some optimization algorithms may not handle integer variables properly. For updated information about the optimization algorithms that handle integer variables, refer to Optimization algorithms.

For additional information on how GEMSEO handles integer variables, refer to How to deal with design spaces.

Remove design variables

The user can also remove a variable in the design space using the DesignSpace.remove_variable() method:

design_space.remove_variable("x")
print(design_space)

Design space:
+------+-------------+-------+-------------+---------+
| name | lower_bound | value | upper_bound | type    |
+------+-------------+-------+-------------+---------+
| y[0] |      -2     |   0   |      2      | float   |
| y[1] |      -1     |   0   |      1      | float   |
| z[0] |      0      |   0   |      3      | integer |
| z[1] |      -1     |   1   |      1      | integer |
| w[0] |      -2     |   2   |      3      | integer |
| w[1] |      -5     |   -2  |      1      | integer |
+------+-------------+-------+-------------+---------+