Big Powers by Hand

The theorem computes (98)5(98)^5 faster than repeated multiplication: write the base as a sum or difference of numbers with easy powers, then expand.

Computing ninety eight to the fifth via one hundred minus two

For example, 98=100298 = 100 - 2, so

(98)5=5C0(100)55C1(100)4(2)+5C2(100)3(22)5C3(100)2(23)+5C4(100)(24)5C5(25)(98)^5 = {^5C_0}(100)^5 - {^5C_1}(100)^4(2) + {^5C_2}(100)^3(2^2) - {^5C_3}(100)^2(2^3) + {^5C_4}(100)(2^4) - {^5C_5}(2^5)

=100000000001000000000+40000000800000+800032=9039207968= 10000000000 - 1000000000 + 40000000 - 800000 + 8000 - 32 = 9039207968

Same recipe: 96=100496 = 100 - 4, 102=100+2102 = 100 + 2, 101=100+1101 = 100 + 1, 99=100199 = 100 - 1, (1.1)=1+0.1(1.1) = 1 + 0.1, (0.99)=10.01(0.99) = 1 - 0.01.

Key Point: For a minus split the signs alternate — track them term by term. Powers of 100 shift the decimal, powers of the small number stay tiny; the terms are easy to add.

[Board Tip] For an approximation, keep only the first few terms: (0.99)5=(10.01)515(0.01)+10(0.01)2=0.951(0.99)^5 = (1 - 0.01)^5 \approx 1 - 5(0.01) + 10(0.01)^2 = 0.951 — the dropped terms are of order 10610^{-6}.

Comparisons and Divisibility

Which is larger? Expand the bigger-looking side and keep just enough terms:

Comparing one point zero one to a million power against ten thousand

For example: (1.01)1000000=(1+0.01)1000000=1+1000000(0.01)+positive terms=1+10000+>10000(1.01)^{1000000} = (1 + 0.01)^{1000000} = 1 + 1000000(0.01) + \text{positive terms} = 1 + 10000 + \cdots > 10000.

Key Point: For x>0x > 0, every term of (1+x)n(1+x)^n is positive, so (1+x)n>1+nx(1+x)^n > 1 + nx after just two terms. That single inequality settles all the standard comparisons.

Divisibility and remainders come from splitting the base and pulling out the modulus:

6n=(1+5)n=1+5n+52nC2+53nC3+=1+5n+25(nC2+5nC3+)6^n = (1+5)^n = 1 + 5n + 5^2\,{^nC_2} + 5^3\,{^nC_3} + \cdots = 1 + 5n + 25\left({^nC_2} + 5\,{^nC_3} + \cdots\right)

so 6n5n=25k+16^n - 5n = 25k + 1: dividing 6n5n6^n - 5n by 25 always leaves remainder 1. The same recipe shows more: 9n+1=(1+8)n+1=1+8(n+1)+64(n+1C2+)9^{n+1} = (1+8)^{n+1} = 1 + 8(n+1) + 64\left({^{n+1}C_2} + \cdots\right), so 9n+18n99^{n+1} - 8n - 9 is divisible by 64.

[JEE Tip] Choose the split so the modulus appears squared in the third term: to work mod 25 split off 5's; mod 64 split off 8's; mod 49 split off 7's. The first two terms then carry all the remainder information.

Conjugate Pairs: Surds That Cancel

Adding or subtracting the expansions of (a+b)n(a+b)^n and (ab)n(a-b)^n kills half the terms:

Conjugate pair sums keeping only even or odd terms

(a+b)n+(ab)n=2[nC0an+nC2an2b2+](even powers of b)(a+b)^n + (a-b)^n = 2\left[{^nC_0}a^n + {^nC_2}a^{n-2}b^2 + \cdots\right] \quad \text{(even powers of } b\text{)}

(a+b)n(ab)n=2[nC1an1b+nC3an3b3+](odd powers of b)(a+b)^n - (a-b)^n = 2\left[{^nC_1}a^{n-1}b + {^nC_3}a^{n-3}b^3 + \cdots\right] \quad \text{(odd powers of } b\text{)}

With surd values of bb, the surviving powers are even, so the surds vanish and the result is often an integer. For example: (x+1)6+(x1)6=2(x6+15x4+15x2+1)(x+1)^6 + (x-1)^6 = 2(x^6 + 15x^4 + 15x^2 + 1), and at x=2x = \sqrt{2}: 2(8+60+30+1)=1982(8 + 60 + 30 + 1) = 198.

[Board Tip] Work symbolically first — find the polynomial identity in xx, and substitute the surd only at the end. Substituting early buries you in surd arithmetic.

Solved Examples

Example 1: (98)5(98)^5

Compute (98)5(98)^5.

Computing ninety eight to the fifth via one hundred minus two

Solution:

Step 1 — Split the base. 98=100298 = 100 - 2: expand (1002)5(100-2)^5 with alternating signs.

Step 2 — Write the six terms. 10105(108)(2)+10(106)(4)10(104)(8)+5(100)(16)3210^{10} - 5(10^8)(2) + 10(10^6)(4) - 10(10^4)(8) + 5(100)(16) - 32.

Step 3 — Evaluate. 100000000001000000000+40000000800000+80003210000000000 - 1000000000 + 40000000 - 800000 + 8000 - 32.

Step 4 — Add. =9039207968= 9039207968.

Takeaway: Powers of 100 shift the decimal; powers of 2 stay tiny — six easy terms beat four long multiplications.

Example 2: (96)3(96)^3

Evaluate (96)3(96)^3.

Solution:

Step 1 — Split. 96=100496 = 100 - 4: (1004)3=1063(104)(4)+3(100)(16)64(100-4)^3 = 10^6 - 3(10^4)(4) + 3(100)(16) - 64.

Step 2 — Evaluate. 1000000120000+4800641000000 - 120000 + 4800 - 64.

Step 3 — Add. =884736= 884736.

Takeaway: Row 3 (1,3,3,11, 3, 3, 1) and powers of 4 — sign alternation from the minus split.

Example 3: (102)5(102)^5

Evaluate (102)5(102)^5.

Solution:

Step 1 — Split. 102=100+2102 = 100 + 2 — a plus split, so ALL signs are positive.

Step 2 — Write the terms. 1010+5(108)(2)+10(106)(4)+10(104)(8)+5(100)(16)+3210^{10} + 5(10^8)(2) + 10(10^6)(4) + 10(10^4)(8) + 5(100)(16) + 32.

Step 3 — Add. 10000000000+1000000000+40000000+800000+8000+32=1104080803210000000000 + 1000000000 + 40000000 + 800000 + 8000 + 32 = 11040808032.

Takeaway: Same skeleton as (98)5(98)^5 — only the signs differ between 1002100-2 and 100+2100+2.

Example 4: (101)4(101)^4

Evaluate (101)4(101)^4.

Solution:

Step 1 — Split. (100+1)4(100+1)^4: powers of 1 vanish, leaving the row itself.

Step 2 — Write the terms. 108+4(106)+6(104)+4(100)+110^8 + 4(10^6) + 6(10^4) + 4(100) + 1.

Step 3 — Add. 100000000+4000000+60000+400+1=104060401100000000 + 4000000 + 60000 + 400 + 1 = 104060401.

Takeaway: The digits of row 4 — 1,4,6,4,11, 4, 6, 4, 1 — appear directly in the answer: 1 04 06 04 01.

Example 5: (99)5(99)^5

Evaluate (99)5(99)^5.

Solution:

Step 1 — Split. (1001)5(100-1)^5: powers of 1 vanish, signs alternate.

Step 2 — Write the terms. 10105(108)+10(106)10(104)+5(100)110^{10} - 5(10^8) + 10(10^6) - 10(10^4) + 5(100) - 1.

Step 3 — Add. 10000000000500000000+10000000100000+5001=950990049910000000000 - 500000000 + 10000000 - 100000 + 500 - 1 = 9509900499.

Takeaway: A ±1\pm 1 split is the cleanest case — the row entries are the whole computation.

Example 6: The million-power comparison

Which is larger: (1.01)1000000(1.01)^{1000000} or 1000010000?

Comparing one point zero one to a million power against ten thousand

Solution:

Step 1 — Split the base. (1.01)1000000=(1+0.01)1000000(1.01)^{1000000} = (1 + 0.01)^{1000000}.

Step 2 — Keep the first two terms. =1+1000000(0.01)+other positive terms=1+10000+= 1 + 1000000(0.01) + \text{other positive terms} = 1 + 10000 + \cdots

Step 3 — Compare. 1+10000+positive terms>100001 + 10000 + \text{positive terms} > 10000. So (1.01)1000000(1.01)^{1000000} is larger.

Takeaway: For x>0x > 0, (1+x)n>1+nx(1+x)^n > 1 + nx — two terms of the expansion settle the comparison.

Example 7: A second comparison

Which is larger: (1.1)10000(1.1)^{10000} or 10001000?

Solution:

Step 1 — Two-term bound. (1+0.1)10000>1+10000(0.1)=1001(1 + 0.1)^{10000} > 1 + 10000(0.1) = 1001.

Step 2 — Compare. 1001>10001001 > 1000, so (1.1)10000(1.1)^{10000} is larger.

Takeaway: The same inequality works whenever the truncated value already beats the target.

Example 8: Approximation

Find an approximation of (0.99)5(0.99)^5 using the first three terms of its expansion.

Solution:

Step 1 — Split. (0.99)5=(10.01)5(0.99)^5 = (1 - 0.01)^5.

Step 2 — Keep three terms. 15(0.01)+10(0.01)2=10.05+0.0011 - 5(0.01) + 10(0.01)^2 = 1 - 0.05 + 0.001.

Step 3 — Evaluate. 0.951\approx 0.951. The dropped terms are of order 10(0.01)3=10510(0.01)^3 = 10^{-5} — negligible at three decimals.

Takeaway: Approximations keep the first few terms and estimate the size of the first dropped one.

Example 9: Remainder mod 25

Prove that 6n5n6^n - 5n always leaves remainder 1 when divided by 25.

Solution:

Step 1 — Split the base. 6n=(1+5)n=1+5n+52nC2+53nC3++5n6^n = (1+5)^n = 1 + 5n + 5^2\,{^nC_2} + 5^3\,{^nC_3} + \cdots + 5^n.

Step 2 — Group the 25's. From the third term onward every term has factor 52=255^2 = 25:

6n5n=1+25(nC2+5nC3++5n2)=25k+16^n - 5n = 1 + 25\left({^nC_2} + 5\,{^nC_3} + \cdots + 5^{n-2}\right) = 25k + 1

Step 3 — Conclude. The remainder on division by 25 is always 1. ∎

Takeaway: Split so the modulus appears squared in the third term — the first two terms carry the remainder.

Example 10: Divisible by 64

Show that 9n+18n99^{n+1} - 8n - 9 is divisible by 64 for every positive integer nn.

Solution:

Step 1 — Split. 9n+1=(1+8)n+1=1+8(n+1)+82n+1C2+83n+1C3+9^{n+1} = (1+8)^{n+1} = 1 + 8(n+1) + 8^2\,{^{n+1}C_2} + 8^3\,{^{n+1}C_3} + \cdots

Step 2 — Subtract the linear part. 1+8(n+1)=8n+91 + 8(n+1) = 8n + 9, so

9n+18n9=64(n+1C2+8n+1C3+)9^{n+1} - 8n - 9 = 64\left({^{n+1}C_2} + 8\,{^{n+1}C_3} + \cdots\right)

Step 3 — Conclude. The bracket is an integer, so the expression is a multiple of 64. ∎

Takeaway: Same recipe as mod 25 — this time the 8's supply 82=648^2 = 64.

Example 11: A conjugate difference

Find (a+b)4(ab)4(a+b)^4 - (a-b)^4, and evaluate (3+2)4(32)4(\sqrt{3} + \sqrt{2})^4 - (\sqrt{3} - \sqrt{2})^4.

Solution:

Step 1 — Subtract the twin expansions. Even powers of bb cancel; odd powers double:

(a+b)4(ab)4=2[4C1a3b+4C3ab3]=2[4a3b+4ab3]=8ab(a2+b2)(a+b)^4 - (a-b)^4 = 2\left[{^4C_1}a^3b + {^4C_3}ab^3\right] = 2\left[4a^3b + 4ab^3\right] = 8ab(a^2 + b^2)

Step 2 — Substitute the surds. a=3,b=2a = \sqrt{3}, b = \sqrt{2}: ab=6ab = \sqrt{6} and a2+b2=3+2=5a^2 + b^2 = 3 + 2 = 5.

Step 3 — Evaluate. 86×5=4068\sqrt{6} \times 5 = 40\sqrt{6}.

Takeaway: Find the polynomial identity first; substitute surds last.

Example 12: A conjugate sum

Find (x+1)6+(x1)6(x+1)^6 + (x-1)^6, and evaluate (2+1)6+(21)6(\sqrt{2}+1)^6 + (\sqrt{2}-1)^6.

Conjugate pair sums keeping only even or odd terms

Solution:

Step 1 — Add the twin expansions. Odd powers cancel; even powers double:

(x+1)6+(x1)6=2[x6+15x4+15x2+1](x+1)^6 + (x-1)^6 = 2\left[x^6 + 15x^4 + 15x^2 + 1\right]

Step 2 — Substitute x=2x = \sqrt{2}. x2=2x^2 = 2, x4=4x^4 = 4, x6=8x^6 = 8: 2(8+60+30+1)2(8 + 60 + 30 + 1).

Step 3 — Evaluate. 2×99=1982 \times 99 = 198 — an integer; the surds vanished with the odd powers.

Takeaway: Conjugate sums keep even powers, where (2)even(\sqrt{2})^{\text{even}} is rational.

Example 13: A factor theorem via expansion

If aa and bb are distinct integers, prove aba - b is a factor of anbna^n - b^n.

Solution:

Step 1 — Rewrite the base. a=(ab)+ba = (a - b) + b, so an=[(ab)+b]na^n = \left[(a-b) + b\right]^n.

Step 2 — Expand with first quantity (ab)(a-b).

an=(ab)n+nC1(ab)n1b++nCn1(ab)bn1+bna^n = (a-b)^n + {^nC_1}(a-b)^{n-1}b + \cdots + {^nC_{n-1}}(a-b)b^{n-1} + b^n

Step 3 — Subtract bnb^n and factor. Every remaining term contains (ab)(a-b):

anbn=(ab)[(ab)n1+nC1(ab)n2b++nCn1bn1]a^n - b^n = (a-b)\left[(a-b)^{n-1} + {^nC_1}(a-b)^{n-2}b + \cdots + {^nC_{n-1}}b^{n-1}\right]

The bracket is an integer, so (ab)anbn(a-b) \mid a^n - b^n. ∎

Takeaway: Adding-and-subtracting inside the base (a=(ab)+ba = (a-b) + b) is the theorem's sneakiest, most reusable move.