Second-Order Derivatives via the Chain Rule
You'll learn how to find second partial derivatives of composite multivariable functions by applying the chain rule multiple times.
The Core Idea
When you have a composite function like z = f(g(x,y), h(x,y)), finding the first partial derivative ∂z/∂x requires the chain rule. But what if you need the second derivative ∂²z/∂x² or ∂²z/∂x∂y?
The key is to treat your first derivative as a new function, then differentiate again using the chain rule. Each term in your first derivative may itself depend on x or y through multiple paths, so the chain rule gets applied to each piece.
How It Works
Suppose you've found ∂z/∂x and it contains terms involving ∂f/∂u, ∂g/∂x, and the original variables. To find ∂²z/∂x²:
- Differentiate each term in
∂z/∂xwith respect toxagain - Apply product rule where necessary (because terms like
(∂f/∂u)·(∂g/∂x)require it) - Apply chain rule again when differentiating
∂f/∂uor∂g/∂x, since these may depend onxindirectly
For example, differentiating ∂f/∂u with respect to x gives:
∂/∂x(∂f/∂u) = (∂²f/∂u²)·(∂u/∂x) + (∂²f/∂u∂v)·(∂v/∂x)
This creates a cascade of derivatives. Tree diagrams become especially helpful here—trace every path from the final variable back through all intermediate functions.
Mixed Partials
For ∂²z/∂y∂x, first find ∂z/∂x, then differentiate that result with respect to y. Again, chain rule applies to every term that depends on y.
Key Takeaway: Second-order derivatives of composite functions require applying the chain rule repeatedly to each term, using both product and chain rules as first derivatives themselves get differentiated—track every dependency path carefully.