Interface DlmPriorSpec

Prior specification for dlmPrior.

Matches MATLAB/R DLM dlmGibbsDIG conventions: Inverse-Gamma priors on observation and system variance(s), with optional Normal priors on AR coefficients.

Example: Weakly informative IG on all variances

import { dlmMLE, dlmPrior } from 'dlm-js';

const prior = dlmPrior({
  obsVar:     { shape: 2, rate: 1 },
  processVar: { shape: 2, rate: 1 },
});
const result = await dlmMLE(y, { order: 1, loss: prior });

interface DlmPriorSpec {
    arCoef?: NormalPrior | NormalPrior[];
    obsVar?: InverseGammaPrior;
    processVar?: InverseGammaPrior | InverseGammaPrior[];
}

Properties

`Optional`arCoef

arCoef?: NormalPrior | NormalPrior[]

Normal(mean, std) prior on AR coefficient(s) φⱼ. Single spec → recycled for all p coefficients.

Array → per-coefficient. Only used when params.arCoefficients.fit is enabled.

`Optional`obsVar

obsVar?: InverseGammaPrior

IG(shape, rate) prior on observation variance s².

Ignored when params.obsStd.fixed is set in dlmMLE options.

`Optional`processVar

processVar?: InverseGammaPrior | InverseGammaPrior[]

IG(shape, rate) prior on process variance(s) wᵢ². Single spec → recycled for all m components (MATLAB DLM convention). Array → per-component (length should equal state dimension m).