Understanding Hessian for multidimensional function

In summary, the Hessian matrix is a square matrix of second-order partial derivatives of a multidimensional function, providing essential information about the function's curvature. It plays a crucial role in optimization problems by indicating whether a critical point is a local minimum, local maximum, or saddle point. The eigenvalues of the Hessian help determine the nature of these points, with positive eigenvalues indicating a local minimum, negative eigenvalues indicating a local maximum, and mixed signs suggesting a saddle point. Understanding the Hessian is vital for analyzing the behavior of complex functions in various fields, including machine learning and economics.
  • #1
SaschaSIGI
3
0
Hello everybody,

I have a question regarding this visualization of a multidimensional function. Given f(u, v) = e^{−cu} sin(u) sin(v). Im confused why the maximas/minimas have half positive Trace and half negative Trace. I thought because its maxima it only has to be negative. 3D vis

2D visualization
 

Attachments

  • hesse2.PNG
    hesse2.PNG
    37.7 KB · Views: 81
  • hesse1.PNG
    hesse1.PNG
    39.1 KB · Views: 91
Physics news on Phys.org
  • #2
Hi,

You have me wondering what I am looking at. Is the Hessian projected as a color code on a plot of the function ?
Did it occur to you to write down the Hessian for this function ? So: what's the expression for the trace of the Hessian ? (*)

What do you mean with
SaschaSIGI said:
because its maxima it only has to be negative

Aren't there minima between the maxima ?

(by the way: single: minimum, maximum. Plural: minima, maxima)

(*) Notice the similarity with the Laplacian :smile: ?

##\ ##
 

Similar threads

Back
Top