Optimization Tools
Browse all available optimization tools.
Available tools
- AdaDelta Optimizer Step Calculator Calculate one scalar AdaDelta parameter update and its accumulated state.
- Adagrad Optimizer Step Calculator Calculate one scalar Adagrad parameter update and its accumulated gradient state.
- Adam Optimizer Step Calculator Calculate one scalar Adam parameter update with bias-corrected moments.
- Ant Colony TSP Solver Approximate a short traveling-salesperson tour with seeded Ant System pheromone updates.
- BFGS Optimizer Step Calculator Calculate a scalar BFGS inverse-Hessian update and the next parameter step.
- Conjugate Gradient Optimizer Step Compute a Fletcher–Reeves conjugate-gradient direction and one caller-sized step from current and previous gradients.
- Diagonal Trust Region Optimizer Step Solve a convex diagonal quadratic trust-region subproblem for a bounded Euclidean step. This is a diagonal model, not a full-Hessian solver.
- Gradient Descent Optimizer Step Apply one gradient-descent update x − η∇f; the caller supplies the gradient and learning rate.
- L-BFGS Optimizer Step Compute one limited-memory BFGS two-loop direction from valid curvature-pair histories and apply a caller-supplied step size.
- Log-Barrier Interior Point Step Apply one gradient step to f(x) − μ∑log(xᵢ) for strictly positive coordinates. Reject steps that leave the feasible region.
- Nadam Optimizer Step Apply a bias-corrected Nesterov Adam update from caller-supplied first and second moments and the one-based iteration.
- Nesterov Accelerated-Gradient Step Calculate one scalar Nesterov momentum update from a gradient evaluated at the lookahead point.
- Newton Optimizer Step Calculator Calculate one damped scalar Newton optimization step from a gradient and Hessian.
- Non-Convex Multistart Solver Run bounded multistart gradient descent on a selected one-dimensional non-convex benchmark.
- Particle Swarm Optimization Solver Minimize a selected benchmark function with a reproducible bounded particle swarm.
- RMSProp Optimizer Step Calculator Calculate one scalar RMSProp parameter update and its moving squared-gradient state.
- Robust Minimax Optimization Solver Find a one-dimensional decision that minimizes the worst squared loss across supplied scenarios using a bounded grid search.
- Steepest Descent Optimizer Step Move a specified Euclidean distance along the negative normalized gradient; this differs from learning-rate gradient descent.
- Stochastic Gradient Descent Step Calculator Apply one stochastic-gradient-descent parameter update from a supplied gradient.