# Homework Help: Partial differentiation question. Would all three methods work?

1. Feb 12, 2012

### s3a

I'm attaching the question and solution.

I'm talking about the first part since the second part is the same just with different variables and stuff. I get what the solution is saying but:

1) What if I computed a Jacobian, with
F = x^2 + xy + y^2 - z = 0
G = 2r + s - x = 0
H = r - 2s - y = 0
would it work? Whether it works or not, why does it work or not?

2) Why am I getting it wrong when I substitute x and y with the r and s equivalents and then attempt taking ∂z/∂r? Am I just making a mistake whenever I compute it or is this attempt just bound to fail for some theoretical reason?

I'm not necessarily concerned about the detailed algebra for (1) and (2) but rather just if the methods should work in theory and why or why they would or wouldn't work and if I was doing something wrong (sorry, the file which had my work was corrupted but I'm not concerned about the detailed algebra anyway apart from the part where you set things up prior to the tedious algebra/computation). Part of the reason why I am asking this is because, instinctively, I would have done method (2) and, for method (1), I'm just trying to see when Jacobians are better or worse to use. So, if Jacobians work for this problem, I'd be great if I can also be told how to recognize when Jacobians are the better or worse to use. Sorry, if I repeated myself a bit but I wanted to make my concerns very clear.

Any input would be greatly appreciated!

#### Attached Files:

• ###### P16.jpg
File size:
9.7 KB
Views:
152
2. Feb 12, 2012

### SammyS

Staff Emeritus

How do you propose using that Jacobian to find ∂z/∂r and ∂z/∂s ?​

Yes, that should work. Without you showing your work, it's hard for us to tell what the problem is.​

3. Feb 12, 2012

### s3a

For (2), I think the problem was that I was tired and sick ;). As for (1), is what I'm uploading now correct?

#### Attached Files:

• ###### MyWork.jpg
File size:
146.4 KB
Views:
132
4. Feb 12, 2012

### SammyS

Staff Emeritus
I'll look over #1 when I get a chance. Hopefully that won't be a whole lot later or perhaps someone else will come along & look it over.

For #2, you can even show more generally that both methods give the same result.
You have that $\displaystyle \frac{\partial z}{\partial r}=14r-3s\,.$

$\displaystyle \frac{\partial z}{\partial r}=5x+4y$
$\displaystyle \frac{\partial z}{\partial r}=5(2s+r)+4(r-2s)$
$\displaystyle =14r-3s\,.$