Quasi-Newton method for truss optimization problem.

This code is designed for solving the truss optimization problem with continuous design variables via quasi-Newton gradient methods.

This code is designed for solving the truss optimization problem with continuous design variables via quasi-Newton gradient methods. The problem firstly make gradient check to adapt the quasi-Newton gradient method to truss optimization problem. Then solve it using MATLAB fmincon tool.

Changelog:

• Refactor (Objective Function): Stripped the penalty term from truss_objective.m. The function now returns the pure "Raw Weight," allowing fmincon to handle constraints natively via truss_constraints.m (nonlcon).
• Feature (History Logging): Implemented a "Best-So-Far" (monotonic) filter in the fmincon_capture_penalized_history function. This prevents the logging of high-penalty (infeasible) spikes during the line-search process, resulting in a cleaner convergence graph.
• Fix (3D FEM Analysis): Corrected the Degree of Freedom (DOF) mapping and the B-matrix size in truss_analysis_and_sensitivity.m to correctly support 3D truss structures (3 DOFs per node).
• Improvement (Reporting): The script now manually calculates the "Penalized Weight" (W⋅(1+P)^2 for the final output, ensuring that QN results are directly comparable to GA results in the FeasibilityPerformance.txt report.