Mathwords logoMathwords

Faà di Bruno's Formula — Definition, Formula & Example

Faà di Bruno's formula is a rule for taking the 2nd, 3rd, or any nth derivative of a composite function f(g(x))f(g(x)) in one step. It is the chain rule extended to higher-order derivatives.

If gg is nn times differentiable at xx and ff is nn times differentiable at g(x)g(x), then (f∘g)(n)(x)(f\circ g)^{(n)}(x) equals a sum over all nn-tuples of nonnegative integers (m1,…,mn)(m_1,\dots,m_n) with m1+2m2+⋯+nmn=nm_1+2m_2+\cdots+nm_n=n. Each term is a multinomial-type coefficient times a derivative of ff times a product of powers of the derivatives of gg. Each tuple corresponds to one way of splitting nn into parts, where mjm_j counts the parts of size jj.

Key Formula

dndxnf(g(x))=∑m1+2m2+⋯+nmn=nn!m1! m2!⋯mn! f(k)(g(x))∏j=1n(g(j)(x)j!)mj\frac{d^n}{dx^n} f(g(x)) = \sum_{m_1+2m_2+\cdots+nm_n=n} \frac{n!}{m_1!\,m_2!\cdots m_n!}\, f^{(k)}(g(x)) \prod_{j=1}^{n} \left(\frac{g^{(j)}(x)}{j!}\right)^{m_j}
Where:
  • nn = the order of the derivative you want
  • mjm_j = a nonnegative integer counting how many times g(j)g^{(j)} appears in a term
  • kk = the order of the derivative of ff in that term, k=m1+m2+⋯+mnk = m_1+m_2+\cdots+m_n
  • f(k)(g(x))f^{(k)}(g(x)) = the kth derivative of the outer function, evaluated at g(x)g(x)
  • g(j)(x)g^{(j)}(x) = the jth derivative of the inner function

Worked Example

Problem: Use Faà di Bruno's formula to find the third derivative of h(x)=ex2h(x)=e^{x^2}, then evaluate it at x=1x=1.
Step 1: Identify the outer and inner functions. Every derivative of f(u)=euf(u)=e^u is eue^u. The inner function is g(x)=x2g(x)=x^2, with derivatives g′=2xg'=2x, g′′=2g''=2, g′′′=0g'''=0.
f(k)(g(x))=ex2,g′=2x,  g′′=2,  g′′′=0f^{(k)}(g(x)) = e^{x^2}, \quad g'=2x,\; g''=2,\; g'''=0
Step 2: For n=3n=3, the tuples with m1+2m2+3m3=3m_1+2m_2+3m_3=3 are (3,0,0)(3,0,0), (1,1,0)(1,1,0), and (0,0,1)(0,0,1). Their coefficients are 3!3!=1\frac{3!}{3!}=1, then 3!1! 1!⋅12!=3\frac{3!}{1!\,1!}\cdot\frac{1}{2!}=3, then 3!1!⋅13!=1\frac{3!}{1!}\cdot\frac{1}{3!}=1. This gives the general third-order rule:
(f∘g)′′′=f′′′(g) (g′)3+3f′′(g) g′g′′+f′(g) g′′′(f\circ g)''' = f'''(g)\,(g')^3 + 3f''(g)\,g'g'' + f'(g)\,g'''
Step 3: Substitute the derivatives from Step 1, factor out ex2e^{x^2}, and evaluate at x=1x=1.
h′′′(x)=ex2(2x)3+3ex2(2x)(2)+ex2(0)=ex2(8x3+12x),h′′′(1)=20eh'''(x) = e^{x^2}(2x)^3 + 3e^{x^2}(2x)(2) + e^{x^2}(0) = e^{x^2}(8x^3+12x), \qquad h'''(1) = 20e
Answer: h′′′(x)=ex2(8x3+12x)h'''(x) = e^{x^2}(8x^3+12x), so h′′′(1)=20e≈54.37h'''(1) = 20e \approx 54.37.

Why It Matters

It gives the Taylor coefficients of a composite function, such as esin⁡xe^{\sin x}, without differentiating by hand over and over. In probability, it links moments to cumulants. In combinatorics, its terms count set partitions through Bell polynomials.

Common Mistakes

Mistake: Forgetting the product rule on the inner derivatives: differentiating f′(g(x)) g′(x)f'(g(x))\,g'(x) as if it were f′′(g(x)) (g′(x))2f''(g(x))\,(g'(x))^2 alone.
Correction: f′(g(x)) g′(x)f'(g(x))\,g'(x) is a product, so its derivative needs the product rule as well as the chain rule: (f∘g)′′=f′′(g) (g′)2+f′(g) g′′(f\circ g)'' = f''(g)\,(g')^2 + f'(g)\,g''. At higher orders every partition of nn contributes its own term, which is what the formula counts.

Related Terms