numopt-js
    Preparing search index...

    Interface GradientDescentResult

    Result returned by gradient descent algorithm.

    interface GradientDescentResult {
        converged: boolean;
        finalCost: number;
        finalGradientNorm?: number;
        finalParameters: Float64Array;
        finalResidualNorm?: number;
        iterations: number;
        parameters: Float64Array;
        usedLineSearch: boolean;
    }

    Hierarchy (View Summary)

    Index

    Properties

    converged finalCost finalGradientNorm? finalParameters finalResidualNorm? iterations parameters usedLineSearch

    Properties

    converged: boolean

    Whether the algorithm converged successfully.

    finalCost: number

    Final cost (objective function value).

    finalGradientNorm?: number

    Final gradient norm (if applicable).

    finalParameters: Float64Array

    Final optimized parameter vector.

    finalResidualNorm?: number

    Final residual norm (for least squares problems). Available for algorithms that work with residual functions.

    iterations: number

    Number of iterations performed.

    parameters: Float64Array

    Final optimized parameter vector.

    Use finalParameters instead. This alias will be removed in a future release.

    usedLineSearch: boolean

    Whether line search was used.

    Settings

    Member Visibility

    On This Page

    Properties