The optimisation dataset

The OptimizationDataset proposes several particular group names, namely DESIGN_GROUP, OBJECTIVE_GROUP, OBSERVABLE_GROUP, and CONSTRAINT_GROUP. This particular Dataset is useful to post-process an optimization history.

from __future__ import annotations

from gemseo.datasets.optimization_dataset import OptimizationDataset

First, we instantiate the OptimizationDataset:

dataset = OptimizationDataset()

and add some data of interest using the methods add_design_variable(), add_constraint_variable(), add_objective_variable(), and add_observable_variable() that are based on Dataset.add_variable():

dataset.add_design_variable("x", [[1.0, 2.0], [4.0, 5.0]])
dataset.add_design_variable("z", [[3.0], [6.0]])
dataset.add_objective_variable("f", [[-1.0], [-2.0]])
dataset.add_constraint_variable("c", [[-0.5], [0.1]])
dataset.add_observable_variable("o", [[-3.0], [8.0]])

as well as another variable:

dataset.add_variable("a", [[10.0], [20.0]])
dataset

GROUP	designs			objectives	constraints	observables.	parameters
VARIABLE	x		z	f	c	o	a
COMPONENT	0	1	0	0	0	0	0
1	1.0	2.0	3.0	-1.0	-0.5	-3.0	10.0
2	4.0	5.0	6.0	-2.0	0.1	8.0	20.0

We could also do the same with the methods add_design_group(), add_constraint_group(), add_objective_group(), and add_observable_group()

dataset = OptimizationDataset()
dataset.add_design_group(
    [[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]], ["x", "y"], {"x": 2, "y": 1}
)
dataset.add_objective_group([[-1.0], [-2.0]], ["f"])
dataset.add_constraint_group([[-0.5], [0.1]], ["c"])
dataset.add_observable_group([[-3.0], [8.0]], ["o"])
dataset.add_variable("a", [[10.0], [20.0]])
dataset

GROUP	designs			objectives	constraints	observables.	parameters
VARIABLE	x		y	f	c	o	a
COMPONENT	0	1	0	0	0	0	0
1	1.0	2.0	3.0	-1.0	-0.5	-3.0	10.0
2	4.0	5.0	6.0	-2.0	0.1	8.0	20.0

Then, we can easily access the names of the different input variables

dataset.design_variable_names

['x', 'y']

the names of the output variables

dataset.constraint_names, dataset.objective_names, dataset.observable_names

(['c'], ['f'], ['o'])

and the names of all variables:

dataset.variable_names

['a', 'c', 'f', 'o', 'x', 'y']

The OptimizationDataset provides also the number of iterations:

dataset.n_iterations

2

and the iterations:

dataset.iterations

[1, 2]

Lastly, we can get the design data as an OptimizationDataset view:

dataset.design_dataset

GROUP	designs
VARIABLE	x		y
COMPONENT	0	1	0
1	1.0	2.0	3.0
2	4.0	5.0	6.0

and the same for the other data groups,

dataset.constraint_dataset

GROUP	constraints
VARIABLE	c
COMPONENT	0
1	-0.5
2	0.1

dataset.objective_dataset

GROUP	objectives
VARIABLE	f
COMPONENT	0
1	-1.0
2	-2.0

dataset.observable_dataset

GROUP	observables.
VARIABLE	o
COMPONENT	0
1	-3.0
2	8.0