Big-M and convex hulls
Learn how nonconvex models are written as integer programs using big-M strategies, and why it should be called small-M.
Automatic dualization
Primal or dual arbitrary in primal-dual solver? No, but YALMIP can help you reformulate your model.
The CUTSDP solver explained
A little known solver
Nonlinear operators - integer models
Mixed-integer representations of nonlinear operators
Nonlinear operators - graphs and conic models
Epi- and hypograph conic representations of nonlinear operators
Nonlinear operators - callbacks
Callback representations of nonlinear operators
Nonlinear operators
Working with nonlinear operators in a structured and efficient fashion
Logics and integer-programming representations
Logic programming in YALMIP means programming with operators such as alldifferent, number of non-zeros, implications and similiar combinatorial objects.
Envelope approximations for global optimization
Outer approximations of function envelopes are the core of the global solver BMIBNB
Infeasible or unbounded
…or both?
Strict inequalities
Wanted but not needed
Debugging numerical problems
Crap in crap out
Equalities with uncertainty
= ≠ ==. Horse and sheep purchases and warehouse logistics
Why is my MILP not finishing
How bad is exponential complexity?
Debugging nonsymmetric square warning
Extremely common
Debugging model creation failed
…but I won’t do that.
Debugging unbounded models
Where to start?
Debugging infeasible models
Where to start?
Constraints evaluates to trivially false
Asking for the impossible
NaN in model
Where why how?
Constraints without any variables
Code works for almost all cases, but suddenly fails.
Debugging unknown error
All solver and YALMIP to crash for diagnostics
Sparse parameterizations in optimizer objects
Be careful with unnecessary symbolic overhead
Parameterizing the uncertainty set in robust optimization
Uncertainty descriptions can only involve uncertain variables, so how can they be parameterized?
Multi-objective problems in YALMIP
How do I create a cheap Ferrari?
Supplying initial guesses to warm-start solvers
Give your solver a hint
Extracting low-level data from solvers
Extracting inputs and outputs from solvers
Extensions on the optimizer
Slice’n dice your problems
Duality
Extract dual solutions from conic optimization problems.
Complex-valued problems
Complex data in optimization models. No problem in reality.
Tagging constraints
Name your constraints for easy reference
Computing multiple solutions in one shot
Avoid that for-loop by using vector objectives
Working with square roots
sqrt, sqrtm, power or cpower?