Consider the equation $e^{xz}+y-z=e$. Using the implicit function theorem shows $z$ is a smooth function of $x,y$ about $(1,1,1)$. I needed to calculate a directional derivative of $z$ at $(1,1)$ and managed that using the implicit function theorem to recover the gradient of $z$.

Now I'm asked whether the partial derivatives of $z$ are symmetric about $(1,1)$ and furthermore, I need to calculate them.

I think the partial derivatives are symmetric because the original function $e^{xz}+y-z=e$ is smooth, which means so is $z=z(x,y)$. I don't understand however how to find second order derivatives. The "formula" $$\frac{\partial z}{\partial x}=-\frac{\frac{\partial f}{\partial x}}{\frac{\partial f}{\partial z}}$$ (read with matrix inverse instead of quotient in matrix case) does not really make sense before it's evaluated at a point, since the RHS has additional variables.

So how to find $\frac{\partial ^2z}{\partial x\partial y}(1,1)$?


Formulas for second partial derivatives of $z$ are given at this link: implicit surface derivatives. Some discussion of how to compute these formulas is here: curvature of implicit surface

  • $\begingroup$ Wait, am I right about the symmetry? $\endgroup$ – eleventeen Feb 9 '17 at 1:35
  • $\begingroup$ The mixed partials of $z$ are symmetric because the mixed partials of $f(x,y,z)=e^{xz}+y-z-e$ are symmetric. See the formula for $z_{xy}$ at the link. If you switch $x$ and $y$, you get the same thing except for an $f_{yx}$ term instead of $f_{xy}.$ For these formulas to make sense, $f_z$ must also be non-zero, which it is at $(1,1,1)$. $\endgroup$ – J. Heller Feb 13 '17 at 6:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.