About the course
Extrema of functions of one and two variables: gradient, Hessian, quadratic functions, gradient descent and Newton method. Constrained optimization: Lagrange multipliers, KKT conditions. Convex sets and functions, convex optimization. Linear programming: the graphical method, slack variables, the simplex tableau, the dual problem and shadow prices. The transportation problem: north-west corner and potentials. Functionals, the Euler–Lagrange equation and classic variational problems. Basics of operations research: the assignment problem, the Wilson formula, the M/M/1 queue. A continuation of linear algebra and multivariable calculus.
Course program: 33 lessons
Unconstrained optimization
- Extremum of a function of one variable
- Gradient and stationary points
- Hessian and the type of a point
- Quadratic function
- Gradient descent
- Newton's method
- Checkpoint: unconstrained optimization
Constrained optimization and convexity
- Equality constraint and the Lagrange multiplier
- The Lagrange function
- Inequality constraints and KKT conditions
- Convex sets and functions
- Testing convexity
- Convex optimization
- Checkpoint: constrained optimization and convexity
Linear programming
- Linear programming: the graphical method
- Standard form and basic solutions
- Simplex tableau: one step
- The simplex method to the end
- Checkpoint: linear programming
Duality and the transportation problem
- The dual problem
- Shadow prices
- Transportation problem: the first plan
- Potentials and optimality of a plan
- Checkpoint: duality and the transportation problem
Calculus of variations
- Functional
- The Euler–Lagrange equation
- Classic variational problems
- Checkpoint: calculus of variations
Basics of operations research
- The assignment problem
- Inventory: the Wilson formula
- The M/M/1 queue
- Checkpoint: operations research
Final test
- Final test in optimization