Minimization problem using partial derivatives

5 replies · 3K views · last reply
Leo Liu
Messages
353
Reaction score
156
Homework Statement
.
Relevant Equations
.
Homework problem 2F-1: find the point(s) on each surface closest to the origin, with the hint to minimize the squared distance. Surfaces: a) xyz^2 = 1; b) x^2 - yz = 1.

a) ONLY
The common way to solve this problem is minimizing the two-variable equation after using the substitution ##z^2=1/(xy)##. Yet I wondered if it is possible to optimize the distance equation with three varibles. So I wrote the following equations:
Distance:
$$f(x,y,z)=s^2=x^2+y^2+z^2$$
$$\begin{cases}f_x=2x\\ f_y=2y\\ f_z=2z\end{cases}\implies (0,0,0) \text{ is a critical point}$$
But the graph of ##xyz^2## says otherwise. Why?
3D surface plot of sqrt(1/(xy)) = 1 over x and y from -2 to 2, showing two curved sheets in opposite quadrants of the xy-plane, each rising steeply toward the z-axis.

Thank you.
 
  • Like
Likes   Reactions: PhDeezNutz and docnet
[tex]f(x,y,z)=f(x,y)=x^2+y^2+\frac{1}{xy}[/tex]

Another approach is using polar coordinates. As for a)
[tex]r^3\ \sin^2\theta \cos^2\theta\cos\phi \sin\phi=1[/tex]
 
Last edited:
Leo Liu said:
Homework Statement:: .
Relevant Equations:: .

View attachment 278155
a) ONLY
The common way to solve this problem is minimizing the two-variable equation after using the substitution ##z^2=1/(xy)##. Yet I wondered if it is possible to optimize the distance equation with three varibles. So I wrote the following equations:
Distance:
$$f(x,y,z)=s^2=x^2+y^2+z^2$$
$$\begin{cases}f_x=2x\\ f_y=2y\\ f_z=2z\end{cases}\implies (0,0,0) \text{ is a critical point}$$
But the graph of ##xyz^2## says otherwise. Why?
View attachment 278157
Thank you.

Realise it's just wrong thinking - you got your fx etc. from the s2 equation with no reference at all to what the surface you are asked about is. By that criterion the distance from the origin to every surface would be 0 !

So you need to go back to your textbook to recall what the method is.

For surface a) I get this distance to be 2√2 but I can have got it wrong too.
 
  • Like
Likes   Reactions: Leo Liu and docnet
epenguin said:
Realise it's just wrong thinking - you got your fx etc. from the s2 equation with no reference at all to what the surface you are asked about is. By that criterion the distance from the origin to every surface would be 0 !

So you need to go back to your textbook to recall what the method is.

For surface a) I get this distance to be 2√2 but I can have got it wrong too.
Wolfram says ##2\sqrt{2}## is correct.

https://www.wolframalpha.com/widgets/view.jsp?id=895957d708a52242400f57757f81e627
 
This looks like a problem that can be done using Lagrange multipliers: Minimize ## x^2+y^2+z^2## subject to the constraint of belonging to the respective surface.