The binomial theorem is more than a formula for expanding (a+b)n. It is a tool that produces clean proofs of divisibility statements, quick numerical estimates that would be impossibly tedious by hand, and combinatorial identities that resist direct attack. This subtopic showcases the three most common application patterns.
1. Divisibility proofs
The pattern: to show An is divisible by some integer d, write A=d⋅k+r (or some other splitting), expand using the binomial theorem, and identify the divisibility.
Template. To prove an−1 is divisible by a−1: write a=1+(a−1) and expand. Or simpler: an−1=(a−1)(an−1+an−2+⋯+1).
To prove an−bn divisible by a−b, expand a=b+(a−b):
an=(b+(a−b))n=bn+nbn−1(a−b)+(2n)bn−2(a−b)2+…
Subtract bn: every remaining term has (a−b) as a factor.
Worked Example A. Show 9n−8n−1 is divisible by 64 for all n∈N.
Write 9=1+8. Then
9n=(1+8)n=∑r=0n(rn)8r=1+8n+(2n)64+(3n)512+…
So 9n−8n−1=(2n)64+(3n)512+⋯=64[(2n)+8(3n)+…], divisible by 64.
Worked Example B. Show 4n+15n−1 is divisible by 9.
4=1+3. 4n=(1+3)n=1+3n+(2n)9+(3n)27+…
So 4n+15n−1=18n+(2n)9+⋯=9(2n+(2n)+…), divisible by 9.
2. Numerical estimates
When n is moderate and the binomial (a+b)n has b small compared to a, the early terms of the expansion dominate. We can compute approximations to high accuracy.
Worked Example C. Compute (1.01)10 to four decimals.
Many identities involving binomial coefficients fall out of expanding (1+x)n and either substituting a clever value of x, or comparing coefficients.
Worked Example E. Prove ∑r=0nr(r−1)(rn)=n(n−1)2n−2 for n≥2.
Differentiate (1+x)n=∑(rn)xr twice with respect to x:
n(n−1)(1+x)n−2=∑r=0nr(r−1)(rn)xr−2.
Set x=1:
n(n−1)2n−2=∑r=0nr(r−1)(rn).\qed
This differentiation trick is enormously powerful , it converts a sum involving r, r2, etc. into a sum we already know.
Worked examples (mixed)
Example 1. Show 5n−1 is divisible by 4 for all n∈N.
5=1+4. 5n=1+4n+(2n)16+… So 5n−1=4n+(2n)16+⋯=4(n+4(2n)+…), divisible by 4.
Example 2. Find the last two digits of 71000. (Hint: work mod 100.)
74=2401≡1(mod100). So 71000=(74)250≡1(mod100). Last two digits: 01.
Example 3. Compute ∑r=010(r10)2r.
This is (1+2)10=310=59049.
Example 4. Show (1.001)1000>2.
By the binomial theorem, (1+0.001)1000=1+1000⋅0.001+(21000)(0.001)2+⋯>1+1=2.
Example 5. Find the integral part of (3+1)6.
(3+1)6+(3−1)6=2[(06)33+(26)32+(46)3+(66)]=2[27+135+45+1]=416. Since 0<(3−1)6<1, we get (3+1)6=416−(3−1)6, so the integral part is 415.
Try it yourself
Show 32n−1 is divisible by 8 for all n∈N.
Show 6n−5n+4 is divisible by 25 for all n≥2. (Adjust as needed.)
Compute (1.02)7 to four decimals.
Find (101)5 exactly using 101=100+1.
Show 72n−1 is divisible by 48.
Find the last digit of 72026.
Compute ∑r=0n2r(rn).
Show (2+1)10+(2−1)10 is rational.
Prove (2)n+1>2n+1 for n≥2. (Hint: binomial expansion of 2n+1=(1+1)n+1.)
Show 10n−1 is divisible by 9.
Estimate (0.99)100 correct to three decimals.
Prove ∑r=0n(rn)3r=4n.
Pitfalls / Tricks
When proving divisibility, expand using the form a=(multiple of d)+1 so that early terms give the part you want to subtract.
For numerical estimates with (1+h)n where h is small, the first three or four terms are usually sufficient.
For irrational expressions like (a+b)n, pair with (a−b)n to extract the integer part.
Insight. A modest binomial expansion replaces a tedious calculation. Always try the binomial theorem first when you see an with a slightly off from a round number.