The Project

About IVRP

Guilherme Almeida Zeni

Senior Software Engineer · MSc Technology · ⬤ ORCID

10+ years in software engineering with a master's degree in technology and expertise in combinatorial optimization and Vehicle Routing Problems. This project explores whether IVRP constitutes a genuinely new VRP variant with applications in cislunar logistics and deep-space supply chains.

Motivation

"If you had to deliver supplies across the solar system, how would you optimize the routes?"

Traditional VRP operates on road networks with static arc costs. The interplanetary equivalent must account for the fact that the cost of traveling between two planets depends entirely on when you depart — driven by the relative positions of the planets and the available launch windows. A transfer that costs 5 km/s Δv at one departure date can cost several times that a few months later.

This temporal, physics-constrained cost function is what separates IVRP from a re-labeled graph problem. It also makes it computationally harder — the search space is the Cartesian product of route assignments and continuous departure dates.

Repository Structure

vrp-algorithms

Python solver: exact time-expanded MILP, greedy and ALNS heuristics, porkchop plots, JSON export.

vrp-simulation

This app — Astro PWA with Three.js 3D simulator and research pages.

vrp-formalization

LaTeX formal proofs: NP-hardness reduction, complexity bounds.

vrp-thesis

Academic thesis chapters in Markdown/LaTeX.

vrp-benchmarks

Benchmark instances and comparison scripts.

vrp-research-context

Literature notes, gap analysis, and positioning.

Technology Stack

Python

Solver & plots

Three.js

3D simulation

Astro

Web framework

React

UI islands

TypeScript

Type safety

Tailwind

Styling

HiGHS

MILP solver

Lambert

Transfer solver

PWA

Offline support