Use this optimization step or heuristic to study an explicitly defined objective and constraints with reproducible parameters, initialization, and stopping rules.

Previous parameter

Required number input.

Current parameter

Required number input.

Previous gradient

Required number input.

Current gradient

Required number input.

Previous inverse Hessian

Required number input.

Learning rate

Required number input.

Calculate a scalar BFGS inverse-Hessian update and the next parameter step. The tool evaluates the supplied inputs together and returns the named outputs below; it does not infer omitted operating conditions or change the units shown.^[1]

BFGS step

The resulting bfgs step returned as an object.

• Convergence and solution quality depend on smoothness, convexity, scaling, gradients, conditioning, hyperparameters, randomness, constraints, and implementation details. Non-convex methods need not find a global optimum.
• Use finite inputs in the displayed units and preserve more precision than the final presentation requires. Independently verify safety-critical, financial, compliance, or production decisions.

Scale variables, monitor objective and constraint residuals, compare starts and algorithms, and verify small instances or local optimality with an independent solver.

• 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.