Maths WorkshopMaths problems with full solutions
Exercise 723Calculus › Single-variable calculus › Integration › Integration techniquesL1Added 2 Nov 2026

A reduction formula

#reduction formula#integration by parts#recurrence

Statement

Let In=∫01xnex dxI_n=\displaystyle\int_0^1x^ne^x\,dx for n≥0n\ge0.

  1. Show that In=e−nIn−1I_n=e-nI_{n-1} for n≥1n\ge1.
  2. Compute I0,I1,I2,I3I_0,I_1,I_2,I_3.
  3. Show that 0<In≤en+10<I_n\le\dfrac e{n+1}, hence In→0I_n\to0.
Hint

Integrate by parts, differentiating xnx^n and integrating exe^x.

Show solutionHide solution

Solution

1. With u=xnu=x^n, v′=exv'=e^x:
In=[xnex]01−n∫01xn−1ex dx=e−nIn−1.I_n=\bigl[x^ne^x\bigr]_0^1-n\int_0^1x^{n-1}e^x\,dx=e-nI_{n-1}.

2. I0=∫01ex dx=e−1I_0=\int_0^1e^x\,dx=e-1. Then
- I1=e−I0=1I_1=e-I_0=1,
- I2=e−2I1=e−2I_2=e-2I_1=e-2,
- I3=e−3I2=e−3e+6=6−2e≈0.5634I_3=e-3I_2=e-3e+6=6-2e\approx0.5634.

3. On [0,1][0,1], 0<ex≤e0<e^x\le e, so 0<In≤e∫01xn dx=en+1→00<I_n\le e\int_0^1x^n\,dx=\frac e{n+1}\to0.

Remark. The recurrence In=e−nIn−1I_n=e-nI_{n-1} is *numerically unstable* when run forward: any small error is multiplied by nn at each step. To compute I20I_{20} one should run it backward, from a good estimate of I30I_{30} for instance.