Why does the Hessian determinant Δ_p = -1 imply that P(0, 0) is a saddle point?

Join the discussion
Registration is free. Start your own thread to ask a follow-up.
1 reply · 3K views
s3a
Messages
828
Reaction score
8

Homework Statement


For the function f(x, y) = xye^[-(x^2 + y^2)] find all the critical points and classify them each as a relative maximum, a relative minimum, or a saddle point.

Homework Equations


Partial differentiation and Hessian determinants.


The Attempt at a Solution


I get how to compute the derivatives. I also get how to compute the Hessian determinants. Basically, I get all the algebraic details but what I would like to ask about (at least for now) is why does the Hessian determinant Δ_p = -1 imply that P(0, 0) is a saddle point?

Any input would be greatly appreciated!
Thanks in advance!
 
Attachments
  • Solution.jpg
    Solution.jpg
    41.5 KB · Views: 487
Physics news on Phys.org
s3a said:

Homework Statement


For the function f(x, y) = xye^[-(x^2 + y^2)] find all the critical points and classify them each as a relative maximum, a relative minimum, or a saddle point.

Homework Equations


Partial differentiation and Hessian determinants.


The Attempt at a Solution


I get how to compute the derivatives. I also get how to compute the Hessian determinants. Basically, I get all the algebraic details but what I would like to ask about (at least for now) is why does the Hessian determinant Δ_p = -1 imply that P(0, 0) is a saddle point?

Any input would be greatly appreciated!
Thanks in advance!

The Hessian at (x0,y0) gives the quadratic terms in the Taylor expansion at that point. In other words, if we write
[tex]f(x,y) \approx f(x_0,y_0) + a_0 (x-x_0) + b_0 (y-y_0)<br /> + \frac{1}{2} r_0 (x-x_0)^2 + s_0 (x-x_0)(y-y_0) + \frac{1}{2} t_0 (y-y_0)^2 + \text{ higher order terms },[/tex] the constants [itex]r_0, s_0, t_0[/itex] are the elements of the Hessian at (x0,y0); that is, the Hessian is
[tex]H(x_0,y_0) = \left[ \matrix{r_0&s_0\\s_0&t_0} \right].[/tex] So, if the Hessian is [tex]H = \left[ \matrix{0&1\\1&0}\right],[/tex] that means that for small x and y we have [itex]f(x,y) \approx xy,[/itex] and if you plot the surface z = xy around (0,0), you will see that (0,0) is, indeed, a saddle point.

Doesn't your textbook or your course notes have this material?

RGV