Home

Solvor all your optimization needs.

solvOR is a pure Python optimization library designed for learning and experimentation. No numpy, no scipy, no compiled extensions. Each algorithm fits in one readable file you can study, modify, and extend.

Learn optimization by reading the code.

---

Why solvOR?

• Readable Code

  ---

  Every algorithm is implemented in clear, documented Python. No magic, no hidden complexity. Perfect for understanding how solvers actually work.

• Educational First

  ---

  Designed for students and learners. Each solver includes Wikipedia links, complexity analysis, and practical tips. Learn the theory by reading working code.

• Easy to Customize

  ---

  Want to add a custom heuristic? Modify the neighborhood function? Everything is accessible Python you can copy, tweak, and experiment with.

• Practical Performance

  ---

  While not as fast as OR-Tools on large instances, solvOR handles real problems effectively. Solves Sudoku instantly, TSP-100 in seconds, and scales to thousands of variables for many problem types.

---

Quick Start

uv add solvor

Linear ProgrammingPathfindingConstraint ProgrammingMetaheuristics

from solvor import solve_lp

# Minimize 2x + 3y subject to x + y >= 4
result = solve_lp(c=[2, 3], A=[[-1, -1]], b=[-4])
print(result.solution)  # Optimal x, y
print(result.objective) # Minimum cost

from solvor import dijkstra

graph = {'A': [('B', 1), ('C', 4)], 'B': [('C', 2)], 'C': []}
result = dijkstra('A', 'C', lambda n: graph.get(n, []))
print(result.solution)  # ['A', 'B', 'C']
print(result.objective) # 3 (total distance)

from solvor import Model

m = Model()
x, y, z = [m.int_var(1, 9, name) for name in 'xyz']
m.add(m.all_different([x, y, z]))
m.add(x + y + z == 15)
result = m.solve()
print(result.solution)  # {'x': 3, 'y': 5, 'z': 7}

from solvor import solve_tsp

distances = [
    [0, 29, 20, 21],
    [29, 0, 15, 17],
    [20, 15, 0, 28],
    [21, 17, 28, 0]
]
result = solve_tsp(distances)
print(result.solution)  # Best tour found
print(result.objective) # Tour length

---

What's in the Box?

Category	Algorithms	Learn About
Linear Programming	solve_lp, solve_milp	Simplex method, branch and bound
Constraint Programming	solve_sat, Model, solve_exact_cover	SAT solving, Dancing Links
Metaheuristics	anneal, tabu_search, lns, evolve	Local search, genetic algorithms
Continuous Optimization	adam, bfgs, nelder_mead, bayesian_opt	Gradient descent, quasi-Newton methods
Graph Algorithms	dijkstra, astar, bellman_ford, floyd_warshall	Shortest paths, A* search
Graph Analysis	topological_sort, scc, pagerank, louvain, kcore	Dependency order, centrality, community detection
Network Flow	max_flow, min_cost_flow, kruskal, prim	Ford-Fulkerson, MST algorithms
Combinatorial	solve_knapsack, solve_bin_pack, solve_job_shop	Dynamic programming, scheduling
Assignment	solve_hungarian, network_simplex	Hungarian algorithm

---

Choosing an Algorithm

I need a provably optimal solution:

• Linear constraints, continuous variables → solve_lp
• Integer or binary variables → solve_milp
• Boolean satisfiability → solve_sat
• Complex constraints (all-different, etc.) → Model

A good solution quickly is fine:

• Have a starting point → tabu_search or anneal
• Continuous, no gradients → nelder_mead
• Expensive evaluations → bayesian_opt

Graph or network problem:

• Shortest path → dijkstra, astar, bellman_ford
• Maximum flow → max_flow
• Spanning tree → kruskal, prim
• Dependencies → topological_sort, strongly_connected_components
• Node importance → pagerank, kcore
• Communities → louvain

Full decision guide

---

Popular Examples

• Sudoku Solver

  ---

  Constraint programming with all_different. Classic puzzle, elegant solution.

• Traveling Salesman

  ---

  Tabu search for TSP. See how local search escapes local optima.

• Shortest Path Grid

  ---

  A* pathfinding on a 2D grid. Understand heuristics in action.

• N-Queens

  ---

  Place N queens with no conflicts. A constraint satisfaction classic.

All cookbook examples · More examples

---

Philosophy

Design principles

1. Working > perfect — Ship code that solves problems
2. Readable > clever — Tomorrow-you needs to understand today-you's code
3. Simple > general — Solve the problem at hand, not all possible problems

solvOR is not trying to compete with production solvers like OR-Tools, Gurobi, or CPLEX. It's designed to help you understand how these algorithms work, so you can make informed decisions about which tools to use and how to apply them.

---

Get Started Browse Examples API Reference