Maths › Practise › Questions › Reduction formulae
Questions on reduction formulae
A reduction formula expresses an integral in terms of the same integral with a lower index, and is usually found by integrating by parts. Running it down from a base case, or from two when the index drops by 2, then gives each member of the family.
15 original questions · 69 marks · the reduction formulae notes · Further Pure 2
To print: the questions (PDF, 55 KB) · the worked answers (PDF, 74 KB)
These are original InkMaths questions. Write a complete answer before opening the worked solution and marking guidance.
Use integration by parts to evaluate ∫ xex dx between 0 and 1.
Worked solution and marking guidance
Take u = x and dv = ex dx, so the integral is [xex] − ∫ex dx = [xex − ex] from 0 to 1 = (e − e) − (0 − 1) = 1. M1 for parts with u = x, A1 for the integrated expression, A1 for 1. Differentiating the power rather than the exponential is the choice that makes progress, since it lowers the index by one.In = ∫ sinnx dx between 0 and π/2 satisfies nIn = (n − 1)In−2. Given I0 = π/2 and I1 = 1, find I4 and I5.
Worked solution and marking guidance
I4 = (3/4)I2 = (3/4)(1/2)(π/2) = 3π/16, about 0.5890. I5 = (4/5)I3 = (4/5)(2/3)(1) = 8/15, about 0.5333. M1 for applying the formula twice to reach I4, A1 for 3π/16, M1 for the odd chain, A1 for 8/15. Even n runs down to I0 and keeps a π; odd n runs down to I1 and does not.Write down the formula for integration by parts, and state which factor should be differentiated when a reduction formula for ∫xnex dx is being derived.
Worked solution and marking guidance
∫u (dv/dx) dx = uv − ∫v (du/dx) dx, taken between the same limits when the integral is definite. B1 for the formula. Take u = xn, so that the power is the factor being differentiated: each application lowers the index by one and leaves an integral of the same shape with n − 1 in place of n, which is what a reduction formula needs. B1 for differentiating the power. Differentiating the exponential instead raises the index of the power and leaves a harder integral than the one started with.In = ∫ (ln x)n dx between 1 and e. Show that In = e − nIn−1 for n ≥ 1.
Worked solution and marking guidance
There is only one factor in sight, so integrate by parts with u = (ln x)n and dv = 1 dx, giving v = x. Then du/dx = n(ln x)n−1 × (1/x), and the x in v cancels that 1/x: In = [x(ln x)n] − ∫x × n(ln x)n−1/x dx = [x(ln x)n] − n∫(ln x)n−1 dx. M1 for parts with v = x.
The boundary term is e × 1n = e at x = e, since ln e = 1, and 1 × 0n = 0 at x = 1, since ln 1 = 0 and n ≥ 1. A1 for the boundary term e.
What remains is n times the same integral with the index lowered by one, so In = e − nIn−1. A1 for the printed formula.
The base value is I0 = ∫1 dx = e − 1, and the formula then gives I1 = e − (e − 1) = 1, which one round of parts on ∫ln x dx confirms. The substitution x = et turns this integral into ∫tnet dt from 0 to 1, which is why the recurrence is the same one that integral satisfies.In = ∫ sinnx dx between 0 and π/2. Show that nIn = (n − 1)In−2 for n ≥ 2.
Worked solution and marking guidance
Write the integrand as sinn−1x × sin x and integrate by parts, differentiating the first factor. The boundary term is [−sinn−1x cos x], which vanishes at both limits, since cos(π/2) = 0 and sin 0 = 0. What is left is (n − 1)∫sinn−2x cos²x dx. Replacing cos²x by 1 − sin²x splits it into (n − 1)In−2 − (n − 1)In, so In + (n − 1)In = (n − 1)In−2, which is the result. M1 for splitting off sin x, M1 for integrating by parts, A1 for the boundary term vanishing, M1 for replacing cos²x, A1 for the printed result.In = ∫ xnex dx between 0 and 1. Show that In = e − nIn−1, and hence find I3.
Worked solution and marking guidance
Integrating by parts with u = xn gives [xnex] − n∫xn−1ex dx. The boundary term is e at x = 1 and 0 at x = 0, so In = e − nIn−1. Starting from I0 = e − 1, the chain runs I1 = e − (e − 1) = 1, I2 = e − 2, and I3 = e − 3(e − 2) = 6 − 2e, about 0.5634. M1 for parts with u = xn, A1 for the boundary term, A1 for the printed formula, M1 for I0 = e − 1, dM1 for running the chain, A1 for 6 − 2e.In = ∫ tannx dx between 0 and π/4. Show that In = 1/(n − 1) − In−2 for n ≥ 2, and find I4.
Worked solution and marking guidance
Split off tan²x = sec²x − 1, giving ∫tann−2x sec²x dx − In−2. The first integral is [tann−1x/(n − 1)], which is 1/(n − 1) at π/4 and 0 at 0. So In = 1/(n − 1) − In−2. With I0 = π/4 this gives I2 = 1 − π/4, and I4 = 1/3 − (1 − π/4) = π/4 − 2/3, about 0.1187. M1 for tan²x = sec²x − 1, A1 for the split, M1 for integrating the first part, A1 for the printed result, M1 for I0 = π/4 and the chain, A1 for π/4 − 2/3.Explain why a reduction formula on its own is not an answer, and how you choose the starting value.
Worked solution and marking guidance
The formula only relates one integral to another, so it generates a chain rather than a number. You need one integral evaluated directly to anchor it. The index drops by a fixed step, so follow the chain down. A step of 2 lands on I0 for even n and I1 for odd n, and a step of 1 always lands on I0. B1 for the formula only linking one integral to another, B1 for needing one evaluated directly, B1 for choosing the starting value by the step size.In = ∫ xncos x dx between 0 and π/2. Show that In = (π/2)n − n(n − 1)In−2 for n ≥ 2, and write down I0 and I1.
Worked solution and marking guidance
Integrate by parts with u = xn and dv = cos x dx, so v = sin x: In = [xnsin x] − n∫xn−1sin x dx. The boundary term is (π/2)n × 1 at the top and 0 at the bottom. M1 for the first application of parts, A1 for the boundary term (π/2)n.
Now integrate the remaining integral by parts again, with u = xn−1 and v = −cos x: ∫xn−1sin x dx = [−xn−1cos x] + (n − 1)∫xn−2cos x dx. Its boundary term vanishes at both limits, since cos(π/2) = 0 and, for n ≥ 2, xn−1 = 0 at x = 0. What is left is (n − 1)In−2. M1 for the second application of parts, A1 for the printed formula.
The base values are I0 = ∫cos x dx = [sin x] = 1 and, by one round of parts, I1 = [x sin x + cos x] = π/2 − 1. B1 for both. Two applications of parts are needed because a power against a trigonometric function returns to its own shape only after the sine has become a cosine again, which is why this formula drops the index by two rather than one.In = ∫ sinnx dx between 0 and π/2 and Jn = ∫ cosnx dx between the same limits. Show, using the substitution u = π/2 − x, that In = Jn for every integer n ≥ 0.
Worked solution and marking guidance
Put u = π/2 − x, so x = π/2 − u and dx = −du. M1 for the substitution with its differential. When x = 0, u = π/2, and when x = π/2, u = 0, so the limits swap over; the minus sign in dx swaps them back. A1 for handling the limits.
The integrand becomes sinn(π/2 − u) = cosnu, so In = ∫ cosnu du between 0 and π/2, which is Jn with the variable renamed. A1 for the integrand, A1 for the conclusion. A definite integral does not depend on the letter used, so renaming u as x finishes it. The result says the two families obey the same reduction formula with the same base values, which is why I0 = J0 = π/2 and I1 = J1 = 1.In = ∫ xne−2x dx from 0 to infinity. Show that In = (n/2)In−1 for n ≥ 1, and hence find I3.
Worked solution and marking guidance
By parts with u = xn and v = −½e−2x: In = [−½xne−2x] + (n/2)∫xn−1e−2x dx. M1 for parts with u = xn. The boundary term is 0 at the origin and tends to 0 as x grows, since the exponential decays faster than any power grows, so In = (n/2)In−1. A1 for the boundary term vanishing at both ends, A1 for the printed formula.
The base value is I0 = ∫e−2x dx = [−½e−2x] = ½. Running the chain, I1 = (1/2)(1/2) = 1/4, I2 = 1 × 1/4 = 1/4, and I3 = (3/2)(1/4) = 3/8. B1 for I0 with the chain to 3/8. In general In = n!/2n+1, and 3!/24 = 6/16 = 3/8 agrees. The ½ from integrating e−2x is what turns the usual factor n into n/2, so it has to be carried through both the boundary term and the remaining integral.In = ∫ xne−x dx from 0 to infinity. Show that In = nIn−1, and hence evaluate I5.
Worked solution and marking guidance
By parts with u = xn, the boundary term [−xne−x] is 0 at the origin and tends to 0 at infinity, since the exponential beats any power. What remains is n∫xn−1e−x dx = nIn−1. Since I0 = 1, repeated use gives In = n!, so I5 = 120. M1 for parts with u = xn, M1 for the boundary term at infinity, A1 for it vanishing, A1 for the printed formula, B1 for I0 = 1, M1 for the chain, A1 for 120.In = ∫ secnx dx between 0 and π/4. Show that (n − 1)In = (√2)n−2 + (n − 2)In−2 for n ≥ 2, and hence find I4.
Worked solution and marking guidance
Split the integrand as secn−2x × sec2x and integrate by parts with u = secn−2x and dv = sec2x dx, so v = tan x. M1 for the split, M1 for parts with v = tan x.
The boundary term is [secn−2x tan x], which is (√2)n−2 × 1 at π/4 and 0 at 0, since sec(π/4) = √2 and tan 0 = 0. A1 for the boundary term. The derivative of secn−2x is (n − 2)secn−2x tan x, so what remains is −(n − 2)∫secn−2x tan2x dx. Replacing tan2x by sec2x − 1 splits that into −(n − 2)In + (n − 2)In−2. M1 for the identity tan2x = sec2x − 1.
So In = (√2)n−2 − (n − 2)In + (n − 2)In−2, and collecting the In terms on the left gives the printed result. A1.
The base value is I2 = ∫sec2x dx = [tan x] = 1. At n = 4: 3I4 = (√2)2 + 2I2 = 2 + 2 = 4, so I4 = 4/3. A1.In = ∫ sinnx dx between 0 and π/2. State the values of n for which the reduction formula nIn = (n − 1)In−2 holds, and explain what goes wrong if it is applied at n = 1.
Worked solution and marking guidance
The derivation writes the integrand as sinn−1x × sin x and integrates by parts, producing the boundary term [−sinn−1x cos x] between 0 and π/2. M1 for identifying the boundary term. At the upper limit cos(π/2) = 0, so that end contributes nothing whatever n is. At the lower limit sin 0 = 0, and sinn−10 is zero only when n − 1 > 0. So the formula holds for n ≥ 2. B1 for the range of n.
At n = 1 the boundary term is [−cos x] between 0 and π/2, which is 0 − (−1) = 1, not 0, so the derivation breaks there. A1 for the value 1. The formula itself would give 1 × I1 = 0 × I−1 = 0, whereas I1 = ∫sin x dx = 1. B1 for the contradiction.
The quantity it would reduce to does not exist either: I−1 = ∫cosec x dx over the same interval does not converge, because cosec x behaves like 1/x near the origin. B1 for that.In = ∫ sinnx dx between 0 and π/2, with I0 = π/2 and I1 = 1, and nIn = (n − 1)In−2 for n ≥ 2. Write I2m as a product of fractions times π/2, and use that product together with the odd chain to evaluate I6 and I7.
Worked solution and marking guidance
Rearranged, In = ((n − 1)/n)In−2, so running down from an even index 2m multiplies I0 by one fraction at each step: I2m = ((2m − 1)/2m)((2m − 3)/(2m − 2))…(1/2) × π/2, the odd numbers up to 2m − 1 over the even numbers up to 2m, all times π/2. M1 for running the recurrence down, A1 for the even chain landing on I0, A1 for the product form.
At m = 3: I6 = (5/6)(3/4)(1/2)(π/2) = (15/48)(π/2) = 5π/32, about 0.4909. A1.
An odd index runs down to I1 = 1 instead, so no π appears: I7 = (6/7)(4/5)(2/3)(1) = 48/105 = 16/35, about 0.4571. M1 for the odd chain, A1 for 16/35.
The two values sit close together because In decreases steadily: sinnx ≤ sinn−1x across the whole interval, so the sequence falls, and the ratio of consecutive terms tends to 1.
All questions, worked solutions and marking guidance are original InkMaths material. They are not copied from awarding-body papers or mark schemes. The codes show how InkMaths allocates the stated total; an awarding body may credit a valid alternative method differently.
The same practice on paper: the printable workbook for this topic, with the questions and a worked answer book.
Practise reduction formulae one question at a time
In the player, mark your own response after viewing the worked solution. Your score decides when the question is offered again.