Multivariable calculus
Partial Derivatives of sin^2(x) + cos^2(y)
For f(x, y) = sin squared of x plus cos squared of y, the partial derivative with respect to x is 2 sin(x)cos(x), which equals sin(2x), and the partial derivative with respect to y is negative 2 cos(y)sin(y), which equals negative sin(2y). Each term contains only one variable, so each partial sees just one term.
Separated variables make each partial a one term job
The function is a sum where the first term has only and the second only . Differentiating with respect to kills outright, since a term with no in it is a constant.
The power rule supplies the factor and the chain rule supplies . The same structure on the other term gives times the derivative of cosine, which is where the minus sign comes from.
At the partials are and , so the surface rises in and falls in at the same rate.
The mistake: reaching for sin^2 + cos^2 = 1
The Pythagorean identity needs the same angle in both terms. Here the arguments are and , so is not and the function is not constant. At , it equals , but at , it equals , and at , it equals .
- Check the arguments match before applying any trigonometric identity.
- The values of this function fill the range from to , so it is far from constant.
- If the identity did apply, would be constant and both partials would be zero everywhere, yet at .
Critical points and the mixed partial
Both partials vanish when and , that is when and are integer multiples of . The critical points form a lattice across the plane.
Because no term contains both variables, the mixed partial is zero everywhere: . That makes the second derivative test easy, since the discriminant is just with and .
At that gives , so the origin is a saddle. Minimum in one direction, maximum in the other, which is exactly what a sum of two independent oscillations produces.
Frequently asked questions
Is sin^2(x) the same as sin(x^2)?
No. means , while squares the input first. Their derivatives differ too: against .
Why write the answers as sin(2x) and negative sin(2y)?
The double angle identity compresses each answer into one term, which makes the zeros easy to read off and the second derivatives quicker to take.