Crate pdp_lns 

pdp_lns

Adaptive Large Neighbourhood Search (ALNS) solver for the Pickup and Delivery Problem with Time Windows (PDPTW).

The solver combines several metaheuristic techniques:

• Multi-start ensemble initialization — four constructive heuristics run in parallel, seeding elite and route pools with diverse solutions.
• LNS with adaptive operator selection — destroy/repair operators weighted by recent performance (ALNS).
• Guided Ejection Search (GES) — vehicle-count minimization via ejection chains with squeeze and perturbation.
• Set Covering (SC) — MIP-based route recombination from a route pool.
• Crossover operators — CREX and SREX for elite pool exploitation.

Quick start

use pdp_lns::algorithm;
use pdp_lns::instance::Instance;
use std::path::Path;
use std::time::Duration;

let inst = Instance::from_file(Path::new("instance.txt"));
let solution = algorithm::solve(&inst, Duration::from_secs(60), 42);
println!("Vehicles: {}, Distance: {:.2}",
    solution.num_vehicles(), solution.total_distance(&inst));

Modules

algorithm

Top-level solver entry points and configuration.

instance

Problem instance representation and I/O.

solution

Solution representation, feasibility checks, and route utilities.

Functions

solve_data_core

Solve a PDPTW instance from raw arrays. Returns the instance and best solution found.

solve_instance_core

Solve a PDPTW instance from a file path. Returns the instance and best solution found.