Developer Hub
Optimization from Python, Julia, R and Ruby
Build the model in the language you already use and solve LP, MILP, NLP and QUBO problems through one API. No solver to configure, no license file.
import numpy as np
import pyomo.environ as pyo
from quicopt import Client
# N binary variables, QUBO matrix Q — given
m = pyo.ConcreteModel()
m.x = pyo.Var(range(N), domain=pyo.Binary)
x = np.array([m.x[i] for i in range(N)])
m.obj = pyo.Objective(expr=x @ Q @ x, sense=pyo.minimize)
client = Client() # key from QUICOPT_API_KEY
result = client.solve(m)
result.objective # minimized xᵀQxusing QuicoptClient, JuMP
# N binary variables, QUBO matrix Q — given
m = Model()
@variable(m, x[1:N], Bin)
@objective(m, Min, x' * Q * x)
result = solve(m) # key from QUICOPT_API_KEY
result.objective # minimized xᵀQxrequire "quicopt"
# N binary variables, QUBO matrix Q — given
m = Quicopt::Model.new
x = Array.new(N) { |i| m.bin_var("x#{i}") }
pairs = (0...N).to_a.product((0...N).to_a)
m.minimize(pairs.sum { |i, j| Q[i][j] * x[i] * x[j] })
client = Quicopt::Client.new # key from QUICOPT_API_KEY
result = client.solve(m)
result.objective # minimized xᵀQxlibrary(quicopt)
# N binary variables, QUBO matrix Q — given
m <- model()
x <- bin_var(m, "x", N)
ij <- expand.grid(i = 1:N, j = 1:N)
minimize(m, sum(Q[cbind(ij$i, ij$j)] * x[ij$i] * x[ij$j]))
result <- solve(m) # key from QUICOPT_API_KEY
result$objective # minimized xᵀQxFrom zero to a solved model
- 1
Sign up
Create a free account. Your API key arrives by email.
Sign up free → - 2
Install
One package for Python, Julia, R or Ruby. Set QUICOPT_API_KEY.
Client libraries → - 3
Solve
Build a model in the tools you know and hand it to one solve call.
Quickstart →
Four official clients
All open source under Apache 2.0, all reading your key from QUICOPT_API_KEY.
One API for every class
Each class has a runnable model with its real output. Black-box objectives are coming soon.
Examples by problem class →Worked examples
Job scheduling (makespan)
Split jobs across machines to finish as early as possible — as a MILP, not M^n assignments.
Traveling salesman (TSP)
Shortest tour visiting every city once — as a MILP (MTZ), not (n-1)! permutations.
Vehicle routing (VRP)
Route a capacity-limited fleet from a depot at min distance — as a MILP (CVRP).
Portfolio optimization
Minimum-variance asset weights for a target return — as a QP, not Monte-Carlo sampling.
Max-cut
Partition a graph to maximize edges cut — as a QUBO, classically, no quantum computer.
Knapsack problem
Pick the highest-value items within a weight budget — as a MILP, not 2^n subsets.
Benchmarks →
What Quicopt reaches on public problem sets (LABS, Gset max-cut, maximum independent set) against the published reference results.