Skip to content
voiage

Frequently Asked Questions

What is voiage?

Section titled “What is voiage?”

voiage is a cross-domain Value of Information library. It quantifies the expected value of reducing uncertainty before a decision, including EVPI, EVPPI, EVSI, and ENBS workflows.

Which runtimes are supported?

Section titled “Which runtimes are supported?”

The supported repository-owned surfaces are the Rust execution core, the Python facade, and thin R and Julia bindings. Mojo is an explicit upstream integration boundary. Go, TypeScript, .NET, and WASM are not supported v1.0 surfaces.

How do I install it?

Section titled “How do I install it?”

For the stable Python facade:

Terminal window
python -m pip install voiage

Optional capabilities are installed through named extras. See the installation guide for the supported dependency boundaries.

How do I calculate EVPI?

Section titled “How do I calculate EVPI?”

Create a ValueArray whose rows are uncertainty samples and whose columns are strategies, then pass it to DecisionAnalysis:

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


analysis = DecisionAnalysis(ValueArray.from_numpy(net_benefits))
evpi = analysis.evpi()

The public Python API preserves the stable result contract while delegating the supported numerical kernels to Rust.

Why is a method or accelerator marked experimental?

Section titled “Why is a method or accelerator marked experimental?”

Experimental methods and accelerator lanes require an explicit contract, parity evidence, and reproducible performance evidence before promotion. CPU fallback behavior remains authoritative until those gates pass.

Where should I report a problem?

Section titled “Where should I report a problem?”

Use the repository’s GitHub issue tracker with a minimal reproduction, package version, runtime version, operating system, and relevant diagnostics. Do not include confidential data in public issues.