Multivariable calculus
Partials of ln(3 + sin(x)cos(y))
For f(x, y) = the natural log of 3 plus sin(x)cos(y), the partial derivative with respect to x is cos(x)cos(y) divided by 3 plus sin(x)cos(y), and the partial derivative with respect to y is negative sin(x)sin(y) divided by the same quantity. The inside stays between 2 and 4, so the logarithm is always defined.
Logarithm outside, product inside
Write . The logarithm rule gives , so the whole job reduces to differentiating the inside, which you already know how to do.
Divide each by and you have both answers. The constant contributes nothing to the numerators, since the derivative of a constant is zero, but it stays in the denominator because that denominator is the original inside function.
At the origin and the gradient is .
The mistake: dropping the 3 from the denominator
The single most common error is writing the denominator as , on the reasoning that the differentiates away. It does differentiate away in the numerator, but the denominator is itself, untouched by differentiation.
- The chain rule for is , and in the denominator is the full inside expression.
- A second slip is splitting the logarithm as . There is no such rule for a sum inside a logarithm, and goes negative anyway.
- A third is losing the minus sign in the partial. It comes from .
Why the function is defined everywhere
Both and live between and , so their product lives between and and the inside lives between and . A logarithm needs a positive argument, and is comfortably positive, so there is no domain restriction to worry about and no denominator that can vanish.
Critical points need both numerators to be zero at once. From and you cannot have and together, so the solutions pair with , or with . Those are the points where hits or , that is the peaks, valleys, and saddles of the surface.
Frequently asked questions
What are the largest and smallest values of the function?
The inside ranges over , so ranges over , roughly to . The extremes occur where equals or .
Does the logarithm change the location of the critical points?
No. Since is strictly increasing, and its inside function have their maxima and minima at exactly the same points. That is why the numerators of the partials are just the partials of the inside.