Multivariable calculus
Second Partial Derivatives of ln(x^2 + y^2 + 1)
Write D = x^2 + y^2 + 1. For f(x,y) = ln(D) the second partials are f_xx = 2(y^2 - x^2 + 1)/D^2, f_yy = 2(x^2 - y^2 + 1)/D^2, and the mixed partial f_xy = -4xy/D^2. Both mixed orders agree. The pure ones add to 4/D^2, which is positive everywhere, so this surface is not harmonic.
Second and mixed partials
Chain rule out, quotient rule back in
Write , which is at least 1 everywhere, so the logarithm and every derivative below are defined on the whole plane. The first partials are chain rule applications of .
Now is a genuine quotient in , since contains . Apply the quotient rule with numerator and denominator .
For the mixed partial the numerator is constant in , so only the denominator moves and the derivative is much shorter.
Symmetry as a shortcut and a check
This function is unchanged when you swap and , so must be with the roles of and exchanged. You get it without doing the quotient rule a second time.
Clairaut symmetry gives an independent check on the mixed partial. Differentiating in gives , matching the other order exactly. Since never vanishes, all of these are continuous everywhere and the theorem applies at every point.
Adding the pure partials collapses nicely, which is the sort of simplification worth looking for after any messy quotient rule.
The mistake students make
The most common wrong answer is , from differentiating only the numerator and forgetting that the denominator also depends on . That version misses the sign change: the correct is negative once , because the surface flattens out far from the origin.
The second common error is a wrong chain rule at the very start, writing without the factor from the inside function. Everything downstream then fails.
Check any candidate formula at a point. At , you have , and the correct , clearly negative.
Frequently asked questions
Why keep the plus 1 inside the logarithm?
It keeps the argument positive at the origin. Without it, blows up at and no derivative exists there. With it, the function is smooth on the entire plane.
Is ln(x^2 + y^2 + 1) harmonic?
No. Its Laplacian is , which is strictly positive. The function is the harmonic one, and it is undefined at the origin.