VRP Solver — Genetic Algorithm & Differential Evolution
Side-by-Side Benchmarking of Genetic Algorithms & Differential Evolution with Interactive GUI
Executive Overview
A scientific desktop computational optimization application built in Python to evaluate evolutionary heuristics on NP-hard Capacity-Constrained Vehicle Routing Problems (CVRP). The software features an interactive Tkinter graphical canvas that visualizes fleet vehicle paths in real time and plots algorithmic convergence metrics comparing Genetic Algorithms (GA) against Differential Evolution (DE).
The Challenge & Solution
The Core Problem: Logistics and supply-chain fleets require optimal route planning to minimize total distance traveled while strictly satisfying vehicle payload capacity limits and customer delivery deadlines. Exact mathematical solvers scale exponentially, making evolutionary heuristics indispensable.
The Engineering Solution: Implemented specialized chromosome encoding, partially matched crossover (PMX), swap mutations, and adaptive differential mutation vectors to prevent premature stagnation in local minima. Built a custom Tkinter canvas that visually animates delivery routes and depot clusters.
System Architecture & Data Flow
Key Engineering Highlights
Empirical Heuristic Benchmarking
Direct comparative analysis of GA vs DE convergence rates across 20 to 100 customer nodes.
Real-Time Interactive Route Canvas
Live visual playback of fleet routes optimizing across successive generations.
Dynamic Parameter Tuning
Interactive GUI sliders for mutation probability, crossover rate, population size, and vehicle capacity.
Capacity Constraint Enforcement
Strict penalty fitness functions guaranteeing that vehicle capacities are never exceeded.
