scipyoptimize

scipy.optimize is a module within the SciPy library that provides a range of numerical optimization and root-finding algorithms. It supports unconstrained and constrained optimization, nonlinear least squares, and curve fitting, and is designed to work with NumPy arrays and vector-valued objectives.

Key entry points include minimize, minimize_scalar, root, least_squares, and curve_fit. For global optimization, the module provides differential_evolution and basinhopping. In addition, several helper utilities support line searches, constraints, and gradient checking, making it a versatile toolkit for scientific computing tasks.

Minimize solves general nonlinear optimization problems by supplying an objective function, an initial guess, and a method. Available methods include BFGS, Newton-CG, L-BFGS-B, TNC, SLSQP, Powell, Nelder-Mead, and COBYLA. minimize_scalar focuses on univariate problems, while root (and fsolve) finds zeros of a function. least_squares minimizes the sum of squares of residuals and supports robust loss functions and multiple solvers for nonlinear least squares.

curve_fit performs nonlinear regression to fit a user-specified model to data, typically via Levenberg-Marquardt. For exploration beyond local optima, differential_evolution and basinhopping offer population-based and stochastic search strategies. The package also allows passing analytic gradients or enabling numerical differentiation via jac or hess options, and it integrates with NumPy arrays and broadcasting.