Interactive lab
The models of the course solved to optimality in the browser: change the data with the sliders or by dragging the points, and a solver written in JavaScript recomputes the solution, with the same numbers as the Python scripts and the lecture notes. Pick an app: it opens full page.
Models side by side
-
Three nonlinear models solved to optimality
The Markowitz portfolio, the optimal price of a ticket and the location of a service: pick the model, change the data with the sliders and the solver, in the browser, recomputes the optimal solution.
Deterministic models
-
Markowitz: optimal portfolio and frontier
The minimum-variance QP with a target return: move the required return, the limit per asset and the expected return of one asset, and watch the optimal portfolio slide along the efficient frontier.
-
Optimal price with capacity
The profit as a function of the ticket price with three demand functions: move the capacity and the unit cost and see when the capacity constraint sets the price.
-
Location: efficiency versus equity
Where to put the charging station: the point that minimizes the weighted average distance, with a cap on the distance of the farthest district. Lower the cap and see the trade-off.
Optimization and machine learning
-
SVM: margin and support vectors
The 90 customers of the chapter in the plane of the two indicators: drag the points and move \(C\), and see the separating line, the margin, the support vectors and the violations.
-
Least squares versus least absolute deviations
The 60 days of the bakery: the least-squares line (QP) and the least absolute deviations line (LP). Drag a point or move the hottest day and see which of the two gets pulled by the outliers.