Advanced Cross-Domain Tutorial

This tutorial is the Astro/Starlight home for the advanced cross-domain workflow. The method is domain-neutral: the domain model generates a sampled net-benefit matrix, while voiage computes the value of additional information.

Engineering design

For a reliability or design problem, represent uncertain material strength, loads, temperature, tolerance, and maintenance costs as sampled parameters. For every sample, calculate one net benefit column per design. A conservative, balanced, and aggressive design can then be compared with the same DecisionAnalysis API used for health-economics decisions.

import numpy as np

from voiage.analysis import DecisionAnalysis
from voiage.schema import ValueArray

rng = np.random.default_rng(42)
strength = rng.normal(400, 50, 1_000)
load = rng.normal(10_000, 2_000, 1_000)
temperature = rng.normal(70, 15, 1_000)

benefits = np.column_stack(
    [
        strength - np.maximum(load - 9_000, 0) - np.maximum(temperature - 85, 0) * 10,
        strength * 1.05 - np.maximum(load - 10_000, 0) - np.maximum(temperature - 90, 0) * 10,
        strength * 1.15 - np.maximum(load - 11_000, 0) - np.maximum(temperature - 95, 0) * 10,
    ]
)

analysis = DecisionAnalysis(
    nb_array=ValueArray.from_numpy(benefits, ["Conservative", "Balanced", "Aggressive"]),
    parameter_samples={"strength": strength, "load": load, "temperature": temperature},
)
print(f"EVPI: {analysis.evpi():.3f}")

Finance and portfolio decisions

For portfolio analysis, generate sampled returns and costs under the chosen market model. Each allocation is a strategy column. The VOI result quantifies the maximum value of resolving the selected uncertainty before committing to an allocation; it does not replace suitability, risk, or regulatory review.

Reproducibility checklist

• Seed or record the random-number generator used to create the uncertainty samples.
• Preserve the decision labels and parameter names alongside the result.
• Record the voiage version, Rust core version, and fixture or dataset provenance.
• Compare cross-language adapters against the shared conformance fixtures before publication.

Runnable notebooks for finance, environmental, and engineering examples are listed in the examples index.