Currently, MultivariatePolynomialFunction uses nested for-loops to cycle through all possible coefficients for a given multivariate polynomial. Function evaluation time could be reduced only including terms with non-zero coefficients, and by implementing a faster algorithm for evaluating polynomials (e.g., Horner's scheme using greedy algorithms).
These improvements could substantially improve performance in FunctionBasedPath, which relies on MultivariatePolynomialFunction when fitting paths with PolynomialPathFitter.
Currently,
MultivariatePolynomialFunction
uses nested for-loops to cycle through all possible coefficients for a given multivariate polynomial. Function evaluation time could be reduced only including terms with non-zero coefficients, and by implementing a faster algorithm for evaluating polynomials (e.g., Horner's scheme using greedy algorithms).These improvements could substantially improve performance in
FunctionBasedPath
, which relies onMultivariatePolynomialFunction
when fitting paths withPolynomialPathFitter
.