ml-levenberg-marquardt
    Preparing search index...

    Type Alias JacobianFunction

    JacobianFunction: (params: number[]) => (x: number) => number[]

    Optional analytical Jacobian. Given the parameter vector, it returns a function that, for an x value, yields the partial derivatives of the model with respect to every parameter (an array of length params.length, in the same order as params). When provided, it replaces the finite-difference approximation of the Jacobian, which is both faster (no extra model evaluation per parameter) and more accurate (so the optimizer usually needs fewer iterations).

    Type Declaration

      • (params: number[]): (x: number) => number[]
      • Parameters

        • params: number[]

        Returns (x: number) => number[]