Documentation

IntroductoryLecturesOnConvexOptimization_Nesterov_2004.Chap01.Example_1_6_11

noncomputable def saddlePointConvergenceObjective (x : EuclideanSpace (Fin 2)) :

The quartic example objective f(x₁, x₂) = (1 / 2) x₁² + (1 / 4) x₂⁴ - (1 / 2) x₂².

Instances For
    theorem saddlePointConvergenceObjective_unitStepGradientMethod_tendsto_origin :
    Filter.Tendsto (gradientMethod (fun (x : ) => 1) saddlePointConvergenceObjective (EuclideanSpace.single 0 1)) Filter.atTop (nhds 0)

    Example 1.6.11 (1): for f(x₁, x₂) = (1 / 2) x₁² + (1 / 4) x₂⁴ - (1 / 2) x₂², the unit-step gradient method started at (1, 0) converges to (0, 0).

    Example 1.6.11 (2): the limit point (0, 0) is a stationary point of saddlePointConvergenceObjective.

    Example 1.6.11 (3): the stationary point (0, 0) is not a local minimum of saddlePointConvergenceObjective.

    theorem unitStepGradientMethod_converges_to_nonminimizing_stationary_point :
    Filter.Tendsto (gradientMethod (fun (x : ) => 1) saddlePointConvergenceObjective (EuclideanSpace.single 0 1)) Filter.atTop (nhds 0) HasGradientAt saddlePointConvergenceObjective 0 0 ¬IsLocalMin saddlePointConvergenceObjective 0

    The unit-step quartic-example trajectory converges to a stationary point that is not a local minimum.