Optimization

optimize!(prob; kwargs...) -> (g_opt, p_opt)

Run topology optimization with beta-continuation.

Keyword arguments:

• max_iter – iterations per beta value (default 40)
• β_schedule – projection steepness values to sweep
• α_schedule – optional loss schedule (same length as beta_schedule)
• use_constraints – enable linewidth constraints on the final beta epoch only
• tol – relative tolerance for convergence
• backup – enable autosaving (p, g_history) checkpoints
• backup_every – autosave interval (iterations)
• backup_path – optional checkpoint path (default joinpath(prob.root, "results_backup.jld2"))
• resume_from – optional checkpoint path to resume from

optimize!(prob::EigenOptimizationProblem; kwargs...) -> (g_opt, p_opt)

Run eigenvalue-based optimization with beta-continuation. Note: eigen sensitivities are TODO and will error during gradient evaluation.

objective_and_gradient!(grad, p, prob) -> Float64

Forward + adjoint pass. Main entry point for optimization.

Single evaluation (no optimization loop).

Finite-difference gradient check.