computational-optimization

The Pareto-Weighted-Sum-Tuning utilizes Learning-to-Rank Machine Learning to help solve Pareto (Multiobjective, Multicriteria) Optimization Problems.

This repository implements a Reinforcement Learning (RL) framework to solve the inverse bone remodeling problem. While the forward problem predicts bone density changes from known loads, this project utilizes Proximal Policy Optimization (PPO) to identify unknown mechanical loading conditions from observed bone morphology.

Opt4Deck is a unified optimization platform that integrates linear, nonlinear, evolutionary, and dynamic methods into a coherent and transparent environment. Designed for engineers and researchers, it provides clear methodology, computation transparency, and practical tools for reliably solving complex optimization problems.

Dissertation for MSc Operational Research with Computational Optimization

Solving the Multi‑Depot Vehicle Routing Problem with Time Windows (MDVRPTW). This program wires an enhanced memetic (genetic) algorithm, optional relocation local search and LNS repair, supports Cordeau-style and CSV dataset loading, and provides interactive visualizations (route canvas, convergence chart), logging, and CSV/image export.

Code for the paper "A Penalty-Based Guardrail Algorithm for Optimization Problems with Increasing Objectives and Non-decreasing Inequality Constraints"

An interactive command-line Java utility for generating custom vehicle routing benchmark problem instances and structured solution template files based on standard Cordeau specifications.

JavaFX Differential Evolution solver for Multi‑Depot Vehicle Routing with Time Windows (MDVRPTW). This program applies Differential Evolution, local search and LNS — visualizes routes and exports CSV/PNG.

High-performance Java/JavaFX MDVRPTW solver with ACO integration, Cordeau/CSV parsers, visualization, and automatic route export.

A Python utility for converting Cordeau unified format MDVRP (type 2) benchmark instances into time-window-constrained MDVRPTW (type 6) data files using Cordeau conventional standards.

A framework based on modular Java that allows for the generation of Asymmetric Generalized Traveling Salesman Problem (AGTSP) benchmark instances with elevation-driven cost asymmetry that can be used for conducting research into terrain-aware routing of EVs and UAVs when inquired to deliver packages during their last mile of service.

Optimization algorithms and heuristics