AI summary by TwinMind
What if you just keep squaring?
Veritasium
33m 5s
Introduction to 10-adic Numbers
- Pattern demonstration: 5² = 25, 25² = 625, 625² = 390,625
- Pattern observation: 5 ends in 5, 25 ends in 25, 625 ends in 625
- 390,625² doesn't end in itself but last 5 digits match (90625)
- 90625² does end in itself, extending pattern to 10 digits
- Converging on infinite-digit number that equals its own square
- Introduction to 10-adic numbers: infinite digits left of decimal point
Properties of 10-adic Numbers
- Addition: digit-by-digit from right to left, works normally
- Multiplication: last digit depends only on last digits of multiplicands
- Example calculation: ...857142857143 × 7 = ...000000000001
- This number equals 1/7th (rational numbers exist without division symbol)
- Finding 1/3: infinite string of 6s followed by 7 (...666666667)
- Method: 3 × 7 = 21 (gives 1), 3 × 6 + 2 = 20 (gives 0 with carry 2)
Comparison with Decimal Expansions
- Traditional: 0.999... = 1 (proved by setting k = 0.999..., 10k - k = 9, k = 1)
- 10-adic: ...999999 = -1 (proved by setting m = ...999999, 10m - m = -9, m = -1)
- Verification: ...999999 + 1 = 0 (all carries create zeros)
- Negative numbers: ...999999993 = -7
- Finding negatives: take 9's complement and add 1
Problems with Composite Base Systems
- Issue: n(n-1) = 0 where n is the self-squaring number
- Violates zero product property (if ab = 0, then a = 0 or b = 0)
- Root cause: base 10 is composite (10 = 5 × 2)
- In units place: 5 × 4 = 20, gives 0 without either factor being 0
- Solution: use prime number bases instead
Introduction to Prime-adic (p-adic) Numbers
- Example: 3-adic numbers (base 3, digits 0, 1, 2)
- Multiplication table mod 3: only way to get 0 is if one factor is 0
- Restores zero product property for any prime base
- 3-adic representation: 1×3⁰ + 2×3¹ + 1×3² + 1×3³ + ...
- 3-adic -1: infinite string of 2s (...222222)
- Verification: ...222222 + 1 = 0
Historical Context: Fermat's Last Theorem
- 1637: Pierre de Fermat reading Diophantus' Arithmetica
- Fermat's claim: xⁿ + yⁿ = zⁿ has no integer solutions for n > 2
- Famous margin note: "truly marvelous proof...too long for margin"
- Remained unproven for 358 years
- Required invention of p-adic numbers for solution
Diophantine Problem Example
- Problem: find three squares whose areas add to create bigger square
- Area of first square = side length of second square
- Area of second square = side length of third square
- Equation: x² + x⁴ + x⁸ = y²
- Real solutions easy to find (e.g., x = 1, y = √3)
- Rational solutions much harder to find
Hensel's Approach and Modular Arithmetic
- Kurt Hensel (late 1800s): expand solutions as powers of primes
- x = x₀ + x₁×3 + x₂×3² + x₃×3³ + ...
- Coefficients: 0, 1, or 2 for base 3
- Modular arithmetic: numbers reset after reaching modulus
- Clock analogy: 12-hour clock, 10 + 4 = 2 (mod 12)
- Method: solve equation mod 3, then mod 9, then mod 27, etc.
Step-by-Step Solution Process
Solving mod 3:
- x₀ can be 0, 1, or 2
- If x₀ = 0: x² + x⁴ + x⁸ = 0
- If x₀ = 1 or 2: x² = x⁴ = x⁸ = 1, sum = 3 ≡ 0 (mod 3)
- Solutions: (0,0), (1,0), (2,0)
- Choose (1,0): x₀ = 1, y₀ = 0
Solving mod 9:
- Expand equation with x₀ = 1, find x₁
- After algebraic manipulation: 3 + 15x₁ ≡ 0 (mod 9)
- Solution: x₁ = 1 (since 3 + 15 = 18 ≡ 0 mod 9)
Solving mod 27:
- Continue process to find x₂ = 1
- Pattern emerges: all coefficients equal 1
- 3-adic solution: ...111111 (infinite string of 1s)
Geometric Series Connection
- 3-adic ...111111 = 1 + 3 + 3² + 3³ + ...
- Geometric series formula: 1/(1-λ) where λ = 3
- Result: 1/(1-3) = -1/2
- Verification: x = -1/2 satisfies original equation
- x² = 1/4, x⁴ = 1/16, x⁸ = 1/256
- Sum: 64/256 + 16/256 + 1/256 = 81/256 = (9/16)²
Rational Solution Found
- First square: side length 1/2
- Second square: side length 1/4
- Third square: side length 1/16
- Combined area: square with side length 9/16
- Solved using "absurd" p-adic numbers with infinite left digits
P-adic Geometry and Topology
- P-adics don't exist on number line
- Visualization: infinite triple-branching tree structure
- Viewed from above: Sierpinski gasket pattern
- Each p-adic number: stack of infinite cylinders (shorter/narrower going up)
- Distance measurement: determined by lowest disagreement level
- Counterintuitive: "big" numbers are actually "small"
P-adic Distance and Convergence
- Two numbers close if they agree up to large powers of prime
- Distance ∝ 1/p^n where n is first disagreement place
- Reversal of usual size intuition enables convergence
- Geometric series works because higher powers contribute less
- P-adic absolute value satisfies: non-negative, multiplicative, triangle inequality
Applications to Fermat's Last Theorem
- 1995: Andrew Wiles and Richard Taylor proved Fermat's Last Theorem
- Proof heavily used p-adic numbers (impossible to be Fermat's original proof)
- "3-5 trick": switched between 3-adic and 5-adic systems when stuck
- Each prime gives completely unrelated number system
Mathematical Philosophy
- Quote by Kazuya Kato: "Real numbers are like the sun, p-adics like stars"
- Sun blocks stars during day, humans sleep at night
- P-adics equally important but hidden from view
- Reminder of undiscovered mathematics remaining
- Applications in algebraic geometry, number theory, cutting-edge research
