2

Problem: Find a function $f(x,y)$ such that $ \nabla f = <y,x>$

My work:

$\dfrac {\partial f}{\partial x} = y$

$\dfrac {\partial f}{\partial y} = x$

$f(x,y) = \displaystyle\int_{ }^{ } \dfrac{\partial f}{\partial x} dx + \int_{ }^{ } \dfrac {\partial f}{\partial y} dy = yx + xy = 2xy$

$\dfrac {\partial (2xy)}{\partial x} = 2x$

$\dfrac {\partial (2xy)}{\partial y} = 2y$

Then shouldn't the function be $xy$?

Dmoreno
  • 7,517
Quaxton Hale
  • 1,258

2 Answers2

1

$$df={(df/dx)_y}dx+{(df/dy)_x}dy=ydx+xdy=d(xy)$$ so $$f=xy$$.

Urgje
  • 1,941
1

Your method of constructing the function isn't really right. You can proceed as follows:

Since $\partial f / \partial x = y$ then

$$f(x,y) = \int \frac{\partial f}{\partial x} \, dx = xy + h(y).$$

Differentiating this with respect to $y$ gives

$$\frac{\partial f}{\partial y} = x + h'(y) = x, \text{ therefore } h'(y) = 0 \text{ and } h(y) = C.$$

Hence the family of functions is $f(x,y) = xy +C$.

Mark Fantini
  • 5,523
  • Just wondering, what is wrong with the way I did it? It seemed intuitive that I would take the integral of df/dx and df/dy. – Quaxton Hale Feb 08 '14 at 18:26
  • @МикроПингвин You can integrate either way, my choice is irrelevant. The point is that the original function is not the sum of the integrals. If you had a three variable function, you proceed by integrating with respect to one variable, arrive at a result plus some function depending on the two other variables. Differentiate what you get with respect to one of those other two and check. Integrate again, and so on. – Mark Fantini Feb 08 '14 at 21:18
  • Okay, I found a PDF with an example: http://www.math.wisc.edu/~conrad/f07/potentials.pdf

    My textbook didn't have an example.

    – Quaxton Hale Feb 08 '14 at 23:22
  • @МикроПингвин Mind if I ask which book are you using? Is it russian? I heard Piskunov is pretty good. – Mark Fantini Feb 09 '14 at 00:46