Skip to main content

Ctrl+K

discopt

Tutorials — Problem Classes

Linear Programming
Quadratic Programming
Mixed-Integer Linear Programming (MILP)
0-1 Knapsack: Packing a Backpack
Mixed-Integer Quadratic Programming (MIQP)
Mixed-Integer Nonlinear Programming (MINLP)
Generalized Disjunctive Programming (GDP)
Dynamic Optimization with DAE Collocation
Robust Optimization
Adjustable Robust Optimization
Multi-Objective Optimization
Complementarity and MPEC Models
Bilevel Optimization (Stackelberg games)
Bilevel via KKT + Model.complementarity()
Stochastic Programming (two-stage recourse)
The pooling problem and the pq-formulation (discopt.pooling)
Geometric programming

Tutorials — Solver Backends

Outer Approximation Solver
MIP-NLP decomposition
Benders Decomposition
Generalized Benders Decomposition
Lagrangian Relaxation
The Automatic Decomposition Advisor
Certified-global MINLP with AMP
Black-box global search with DIRECT (solver="direct")
Expensive black boxes: the surrogate backend (solver="surrogate")
POUNCE: Pure-Rust Ipopt Port
POUNCE: Parametric Sensitivity Analysis (sIPOPT)
Differentiable optimization layers (POUNCE-backed)
Differentiable MILP / MIQP
cyipopt: The Ipopt Solver Backend
Solver Backend Selection Guide

Tutorials

Run discopt in your browser
discopt Quickstart Tutorial
Modeling Guide: Expressive vs. Fast Construction
Sets & Indexing for Sparse Models
Piecewise-Linear Functions
The discopt Optimization Course (discopt tutor)
discopt MINLP Examples
Problem Class Auto-Detection in discopt

Advanced Features

Disjunction semantics — the source contract
Complementarity semantics — the source contract
Global optimization: what discopt can and can’t certify
discopt Advanced Features
Solver Internals: BARON-Parity Features
Presolve
Bound tightening: FBBT, row activity, and OBBT
Debugging infeasible models with an IIS
Conflict analysis and no-good cuts
Primal Improvement Heuristics
Derivatives of models
Parametric Sensitivity Analysis (sIPOPT-style)
LLM Integration
LLM Integration
Warm-Starting with Initial Solutions
Convex NLP Fast Path
Convexity detection
Deriving envelopes & relaxations with SymPy
Reduced-Space Global Optimization of Hidden-Function Models
Head-to-head: discopt vs. SUSPECT
Nonlinear Branch & Bound (NLP-BB)
Implicit-Function Node: Recycle Streams and Algebraic Loops
External Functions: Grey-Box Models with Your Own Derivatives
Grey-Box Design: Sizing a Pipeline Around an Iterative Friction Solver
Exporting Models to MPS and LP Formats
Saving and loading models
Rich model representations
Using discopt as a Pyomo solver
Callbacks and User-Defined Cuts
Interactive Branch-and-Bound Debugger

Solver Backends

QP Solver Backend: POUNCE
NLP Backend Comparison: POUNCE vs Ipopt
The CUTEst Interface
Benchmarks by Problem Class
Benchmark Dashboard

Applications

Neural Network Embedding
Scaling ReLU Network Embeddings: discopt vs OMLT, and Depth vs Width
Hybrid physics + ML: training neural DAEs
Minimum-Energy Crossing Points (MECP / MECI)
AC optimal power flow (discopt.opf)
Decision-Focused Learning with discopt
Parameter Estimation
One model, four jobs: simulate, fit, design, optimize

API Reference

API Reference
- discopt

Appendix

Ask discopt: the in-browser docs assistant
Comparison of Optimization Frameworks
Using discopt as a GAMS solver
References

Repository
Open issue

Search

Ctrl+K

By John Kitchin

© Copyright 2026, John Kitchin.