← ~/visualizations

lagrange-multipliers

Shows a moving point constrained to g(x,y)=0 (a circle) while minimizing f(x,y)=x^2+2y^2. At the constrained optimum, the visualization highlights gradient parallelism by drawing ∇f, ∇g, and λ∇g from the same point and displaying the stationarity condition ∇f = λ∇g alongside feasibility g(x)=0.

canvasclick to interact
t=0s

practical uses

  • 01.Constrained optimization in engineering design (e.g., minimize weight subject to strength constraints)
  • 02.Economics/utility maximization under budget constraints
  • 03.Machine learning optimization with equality constraints (e.g., normalization constraints)

technical notes

Pure Canvas2D. The point evolves via tangential descent: v = -∇f + λ∇g with λ = (∇f·∇g)/(∇g·∇g) to make v orthogonal to ∇g, then re-projected to the circle to enforce g(x)=0. Contours are polyline approximations snapped to a pixel grid for a blocky aesthetic; animation is time-based with a 3–5s cycle and periodic restarts.