Save and Reload Sobol Samples

SobolSamples stores both the unique model-evaluation rows and the metadata needed to reconstruct the internal Saltelli design.

import jaxgsa
from jaxgsa.benchmarks.ishigami import PROBLEM

samples = jaxgsa.sobol.sample(PROBLEM, n_samples=4096, seed=42)
samples.save("runs/ishigami")

restored = jaxgsa.sobol.SobolSamples.load("runs/ishigami")
Y = my_model(restored.samples)
result = jaxgsa.sobol.analyze(restored, Y)

The .npz suffix is optional. Both calls above address runs/ishigami.npz.

One compressed file contains:

• the unique sample matrix;
• stable sample identifiers;
• the expanded-to-unique Saltelli mapping;
• the input problem and output names;
• sampling settings such as base_n and second-order mode.

Use NumPy directly when a separate CSV or table is needed:

import numpy as np

np.savetxt("runs/ishigami.csv", samples.samples, delimiter=",")

The CSV is only a model-input table; keep the NPZ file for later jaxgsa analysis.