Create sensitivity analysis

create_sensitivity_analysis() is a top-level function to create a sensitivity analysis from a sensitivity analysis class name, e.g. "MorrisAnalysis".

from __future__ import annotations

from gemseo.problems.uncertainty.ishigami.ishigami_discipline import IshigamiDiscipline
from gemseo.problems.uncertainty.ishigami.ishigami_space import IshigamiSpace
from gemseo.uncertainty import create_sensitivity_analysis

There are two ways of using create_sensitivity_analysis().

The first one is to perform a sensitivity analysis from a collection of disciplines and an uncertain space:

analysis = create_sensitivity_analysis("MorrisAnalysis")
uncertain_space = IshigamiSpace()
discipline = IshigamiDiscipline()
samples = analysis.compute_samples([discipline], uncertain_space, n_samples=0)
indices = analysis.compute_indices()
indices

    INFO - 16:21:57: *** Start MorrisAnalysisSamplingPhase execution ***
    INFO - 16:21:57: MorrisAnalysisSamplingPhase
    INFO - 16:21:57:    Disciplines: IshigamiDiscipline
    INFO - 16:21:57:    MDO formulation: MDF
    INFO - 16:21:57: Running the algorithm MorrisDOE:
    INFO - 16:21:57:      5%|▌         | 1/20 [00:00<00:00, 491.31 it/sec]
    INFO - 16:21:57:     10%|█         | 2/20 [00:00<00:00, 832.20 it/sec]
    INFO - 16:21:57:     15%|█▌        | 3/20 [00:00<00:00, 1109.80 it/sec]
    INFO - 16:21:57:     20%|██        | 4/20 [00:00<00:00, 1325.01 it/sec]
    INFO - 16:21:57:     25%|██▌       | 5/20 [00:00<00:00, 1526.53 it/sec]
    INFO - 16:21:57:     30%|███       | 6/20 [00:00<00:00, 1703.16 it/sec]
    INFO - 16:21:57:     35%|███▌      | 7/20 [00:00<00:00, 1840.30 it/sec]
    INFO - 16:21:57:     40%|████      | 8/20 [00:00<00:00, 1912.91 it/sec]
    INFO - 16:21:57:     45%|████▌     | 9/20 [00:00<00:00, 2029.61 it/sec]
    INFO - 16:21:57:     50%|█████     | 10/20 [00:00<00:00, 2135.16 it/sec]
    INFO - 16:21:57:     55%|█████▌    | 11/20 [00:00<00:00, 2213.25 it/sec]
    INFO - 16:21:57:     60%|██████    | 12/20 [00:00<00:00, 2285.41 it/sec]
    INFO - 16:21:57:     65%|██████▌   | 13/20 [00:00<00:00, 2361.86 it/sec]
    INFO - 16:21:57:     70%|███████   | 14/20 [00:00<00:00, 2432.08 it/sec]
    INFO - 16:21:57:     75%|███████▌  | 15/20 [00:00<00:00, 2492.75 it/sec]
    INFO - 16:21:57:     80%|████████  | 16/20 [00:00<00:00, 2544.89 it/sec]
    INFO - 16:21:57:     85%|████████▌ | 17/20 [00:00<00:00, 2606.97 it/sec]
    INFO - 16:21:57:     90%|█████████ | 18/20 [00:00<00:00, 2657.15 it/sec]
    INFO - 16:21:57:     95%|█████████▌| 19/20 [00:00<00:00, 2702.70 it/sec]
    INFO - 16:21:57:    100%|██████████| 20/20 [00:00<00:00, 2704.87 it/sec]
    INFO - 16:21:57: *** End MorrisAnalysisSamplingPhase execution ***

MorrisAnalysis.SensitivityIndices(mu={'y': [{'x1': array([-0.60047199]), 'x2': array([0.51230435]), 'x3': array([-0.89800793])}]}, mu_star={'y': [{'x1': array([0.69887482]), 'x2': array([0.65136343]), 'x3': array([0.89805157])}]}, sigma={'y': [{'x1': array([0.96395158]), 'x2': array([0.6549141]), 'x3': array([0.79878356])}]}, relative_sigma={'y': [{'x1': array([1.37929075]), 'x2': array([1.00545113]), 'x3': array([0.88946291])}]}, min={'y': [{'x1': array([0.0338188]), 'x2': array([0.11821721]), 'x3': array([8.72820113e-05])}]}, max={'y': [{'x1': array([2.2360336]), 'x2': array([1.25769934]), 'x3': array([2.12052546])}]})

The samples can be saved on the disk using the to_pickle() function, e.g. to_pickle(sample, "my_samples.p"), in order to use them later to compute sensitivity indices.

The other way is to perform a sensitivity analysis from samples computed from another sensitivity analysis:

analysis = create_sensitivity_analysis("MorrisAnalysis", samples=samples)
indices = analysis.compute_indices()
indices

MorrisAnalysis.SensitivityIndices(mu={'y': [{'x1': array([-0.60047199]), 'x2': array([0.51230435]), 'x3': array([-0.89800793])}]}, mu_star={'y': [{'x1': array([0.69887482]), 'x2': array([0.65136343]), 'x3': array([0.89805157])}]}, sigma={'y': [{'x1': array([0.96395158]), 'x2': array([0.6549141]), 'x3': array([0.79878356])}]}, relative_sigma={'y': [{'x1': array([1.37929075]), 'x2': array([1.00545113]), 'x3': array([0.88946291])}]}, min={'y': [{'x1': array([0.0338188]), 'x2': array([0.11821721]), 'x3': array([8.72820113e-05])}]}, max={'y': [{'x1': array([2.2360336]), 'x2': array([1.25769934]), 'x3': array([2.12052546])}]})

The argument samples of create_sensitivity_analysis() can be either an IODataset as above or a pickle file path, e.g. create_sensitivity_analysis("MorrisAnalysis", samples="my_samples.p").