Register to reply

Is this a saddle?

by cambo86
Tags: saddle
Share this thread:
cambo86
#1
Dec28-12, 09:21 AM
P: 25
For [tex]f(x,y)=x^2+y^3[/tex]
Is there a saddle point at (0,0) or does the function have to have 2 or more sides going down like [tex]g(x,y)=x^2-y^2[/tex]
Phys.Org News Partner Science news on Phys.org
Climate change increases risk of crop slowdown in next 20 years
Researcher part of team studying ways to better predict intensity of hurricanes
New molecule puts scientists a step closer to understanding hydrogen storage
Zondrina
#2
Dec28-12, 09:25 AM
P: 1,462
Okay so have you taken the required derivative and found your critical points?
cambo86
#3
Dec28-12, 09:32 AM
P: 25
The critical point is (0,0). I'm just wondering if f(0,0) is a saddle point when the graph is shaped more like a chair than a saddle.

Dick
#4
Dec28-12, 09:39 AM
Sci Advisor
HW Helper
Thanks
P: 25,251
Is this a saddle?

Quote Quote by cambo86 View Post
The critical point is (0,0). I'm just wondering if f(0,0) is a saddle point when the graph is shaped more like a chair than a saddle.
The exact shape doesn't matter. A saddle point is a stationary point that isn't a local max or min. Is yours?
Zondrina
#5
Dec28-12, 09:39 AM
P: 1,462
Quote Quote by cambo86 View Post
The critical point is (0,0). I'm just wondering if f(0,0) is a saddle point when the graph is shaped more like a chair than a saddle.
Are you familiar with the second derivative test? As in where you check :

[itex] D = f_{xx}f_{yy} - f_{xy}^{2}[/itex]

This will allow you to see if your function has a saddle, min or max at a critical point ( Doesn't alwayyyyys work though ).
cambo86
#6
Dec28-12, 09:40 AM
P: 25
Quote Quote by Dick View Post
The exact shape doesn't matter. A saddle point is a stationary point that isn't a local max or min. Is yours?
Yes, thank you.
Ray Vickson
#7
Dec28-12, 01:46 PM
Sci Advisor
HW Helper
Thanks
P: 4,946
Quote Quote by Dick View Post
The exact shape doesn't matter. A saddle point is a stationary point that isn't a local max or min. Is yours?
The Hessian test fails in this case. If H is the Hessian evaluated at the stationary point, the exact second-order conditions are: (i) a necessary condition for a minimum is that H is positive semi-definite; and (ii) a sufficient condition for a strict local minimum is that H is positive definite. For testing a max, just test -H for a min of -f (or replace positive (semi) definite by negative (semi) definite).

Note that there is a bit of mismatch between the necessary and sufficient second-order conditions: there is no sufficient condition for a non-strict local min, so functions like that of the OP slip through the cracks. This happens as well in one dimension, where testing functions like f(x) = x^3 for an optimum at x = 0 cannot just rely on the second derivative.

For the OP's function f(x,y), the Hessian at (0,0) is positive semi-definite, so (0,0) cannot be a local *maximum *; it must either be a local minimum or a saddle point. In fact, f(x,0) = x^2, so x = 0 is a minimum along the line (x,0), and f(0,y) = y^3, so y = 0 is an inflection point along the line (0,y). Thus, there are points near (0,0) giving f(x,y) > f(0,0) and other points near (0,0) giving f(x,y) < f(0,0); that is more-or-less what we mean by a 'saddle point'.


Register to reply

Related Discussions
What's a Saddle Point? Calculus & Beyond Homework 3
Multivariable saddle point Calculus & Beyond Homework 5
Min, Max or Saddle Point Calculus & Beyond Homework 1
Lagrangian on a saddle advice? Classical Physics 1
Saddle points Introductory Physics Homework 7