βˆ‡ProvenDecisions
Ctrl K

Problem classes

Every plan solves the same kinds of problems. Plans differ in how much you can do (word, photo and voice problems use a monthly allowance), how large a model can be (Pro adds vector syntax), and which tutorials you can open. Each answer says exactly what is proven: 🟒 proven optimal, 🟑 the best found and labelled as such, πŸ”΄ not solved yet, and the app says so.

What ProvenDecisions solves

Problem classHow it is solved
🟒Linear programs
Linear objective and constraints: production, blending, diet, staffing
Proven optimal (exact certificate and dual check)
🟒Networks
Transport, assignment, shortest path, flows
As linear programs
🟒Plans over several periods
Several periods solved together, with known data
As linear or integer programs
🟒Absolute values, max, min
Least absolute deviations, minimax, Chebyshev fits
Rewritten exactly as a linear program (shown)
🟒Ratios of linear functions
Return per unit, efficiency ratios
Exact change of variables
🟒Integer and yes/no decisions
Knapsack, assignment, facility location, scheduling, small routing
Optimal with bound and gap reported
🟒Dynamic programs (known data)
Stage-by-stage decisions such as the stagecoach problem
Written as a linear or integer program; no separate DP solver
🟒Convex quadratic programs
Portfolios, closest points, least squares, ridge and lasso
Optimal, with a dual or optimality check
🟒Norms and cones
Distances, enclosing circles, risk limits, robust lines already written out
Second-order cone solvers
🟒Logs and exponentials
Log utility, entropy, logistic regression
Exponential cone solvers
🟒Geometric programs
Posynomial design problems
Log change of variables (a few labelled limited)
🟒Small semidefinite programs
Eigenvalue bounds, matrix inequalities, nearest correlation matrix
Semidefinite solvers
🟒Mixed-integer convex
Integers with a convex quadratic or cone
Branch and bound (SCIP)
🟒🟑Classic shapes
Smallest enclosing or largest inscribed ellipse; geodesics on a sphere; the brachistochrone; k-means
Built-in methods, each checked
🟑Nonconvex problems
Indefinite quadratics, general nonlinear, products of variables
Best point from many starts, labelled β€œlocal optimum, not certified as global”
🟑Nonconvex with integers
Integers plus nonconvex terms
Solved when an exact rewrite exists; otherwise no answer yet
🟑Two objectives
Cost against emissions, return against risk
Solved as the one weighted objective you give
πŸ”΄Stochastic and robust
Random data, chance constraints, worst cases over an uncertainty set
Not yet: the app says so; scenario data typed as a fixed model is solved
πŸ”΄Multistage and dynamic with uncertainty
Decisions revised as outcomes appear, Markov decision processes
Not yet
πŸ”΄Very large or black-box
Over 5,000 variables, or an objective only a simulator can compute
Not yet
πŸ”΄Bilevel and equilibrium
One optimization inside another
Not yet

What differs by plan

No accountFreeStudentStandardPro
Problem classesthe samethe samethe samethe samethe same
Typed math5 tries a day on the front pageβœ…βœ…βœ…βœ…
Word problemsβ€”βœ… a few a monthβœ…βœ…βœ… the most
Photos and voice a monthβ€”5 and 5250 and 100200 and 200500 and 500
Vector syntax for large models (var x[100])β€”β€”β€”β€”βœ…
Duals, proofs, walkthrough depthβœ…βœ…βœ…βœ…βœ…
API and MCP access (scripts, AI assistants)β€”β€”β€”βœ…βœ…
Tutorialsβ€”Free onesMoreMoreAll

Details: Plans and limits Β· Pricing Β· API and MCP Β· Primal and dual

Try ProvenDecisions free: type, photograph or say an optimization problem; every answer is checked by a solver.

What do you want to optimize?

Type the math, describe it in words, snap a photo or say it. ProvenDecisions builds the model, solves it, checks it and walks you through it.

One constraint per line, or separated by commas. Add integer: x, y or binary: z for whole-number decisions. LaTeX works too.

Try: