AI summary by TwinMind

The Yellowstone Permutation - Numberphile

Numberphile

20m 59s

Introduction and Sequence Definition

  • Speaker returns from Iceland, saw geysers, connects to mathematical sequence with "geysers"
  • Yellowstone permutation sequence begins: 1, 2, 3
  • Rules:
    • No term repeated
    • Always pick smallest legal number
    • New term must be relatively prime to previous term (gcd = 1)
    • Must have common factor with term two positions back

Building the Sequence

  • Fourth term: needs to be relatively prime to 3, have common factor with 2
    • 4 works (smallest available, relatively prime to 3, shares factor 2 with term 2)
  • Fifth term: 5 can't work (relatively prime to 4 but no common factor with 3)
    • Need multiple of 3: 6 won't work (shares factor 2 with 4)
    • 9 works (shares factor 3 with term 3, relatively prime to 4)
  • Continuing sequence: 1, 2, 3, 4, 9, 8, 15, 14, 5, 6, 25, 12, 35, 16, 7...

Geyser Phenomenon

  • Periodic large jumps in sequence values
  • Example: 35 is three times position number (geyser eruption)
  • Between terms 100-200: geysers occur when hitting primes
  • Prime 61 causes sequence: 61, 2×61, 7×61 (geyser pattern)
  • All primes cause geysers
  • Geysers unpredictable like primes but follow pattern

Proof That Every Number Appears (7 Steps)

Step 1: Sequence is Infinite

  • Always candidates available: pick giant prime q, use 15×q
  • 15×q relatively prime to previous term, has common factor with term two back

Step 2: Define Key Functions

  • w(M) = when M appears in sequence (position number)
  • w(M) = -1 if M missing
  • L(M) = max{w(1), w(2), ..., w(M)} (last chance to see M)
  • For n > L(M): a(n) > M

Step 3: Infinitely Many Primes Divide Terms

  • Proof by contradiction: suppose prime p and all primes >p missing
  • Beyond L(p²), all terms >p² and products of primes <p
  • Any term a(n) has common factor q<p with previous term
  • Could use q×p < p² instead of current term (contradiction with "smallest" rule)

Step 4: Every Prime Divides Some Term

  • If prime 17 never divides any term, then 19 can't either
  • When 19 would appear, could use 17 instead
  • By induction: if 17 missing, all primes >17 missing
  • Contradicts Step 3 (infinitely many primes appear)

Step 5: Every Prime Divides Infinitely Many Terms

  • Suppose prime p divides only finitely many terms
  • Pick prime q > p×10⁶, q appears in sequence
  • When q appears, could use p^million instead (contradiction)

Step 6: Every Prime Appears "Naked" (Not as Multiple)

  • For huge multiple h×p in sequence
  • Next valid term can be p itself (same argument as building sequence)

Step 7: Every Number Appears

  • Suppose number k missing, k not prime
  • Let p be prime factor of k
  • Go far enough to find huge multiple h×p
  • Could place k two steps later unless blocked by common factors
  • To block k permanently, every subsequent term needs common factor with k
  • Impossible: infinitely many naked primes appear (Step 6)
  • Some naked prime won't share factor with k
  • Therefore k must appear

Visual Pattern Analysis

Graph Structure

  • Plot shows dots for each term
  • Reference line y = x (diagonal)
  • Most non-primes lie close to diagonal
  • Primes appear on line with slope 1/2

Line Classifications

  • Even numbers: line y ≈ x(1 - 1/(2 log x))
  • Odd composite numbers: line y ≈ x(1 + 1/log x)
  • Geysers (3×primes): slope 3/2 line
  • Geysers (5×primes): slope 5/2 line
  • Geysers (7×primes): slope 7/2 line
  • Pattern continues: (odd number)×prime geysers on slope (odd number)/2

Parity Pattern Theory

  • After position 212: sequence alternates even-odd-even-odd
  • Disruptions occur in groups of 5 terms when hitting primes
  • Example at position 213: prime 101 causes disruption
    • Pattern: 2×101, random odd, 101, random even, 5×101
  • Theory explains even numbers on line y = x(1 - 1/(2 log x))

Properties

  • Geysers become increasingly rare but can be arbitrarily large
  • No formal proof of visual pattern theory yet
  • Collaborative 6-author paper published on sequence
  • Proof uses only elementary mathematics (no calculus, measure theory, probability)

TwinMind

Get summaries like this for your own meetings

TwinMind takes notes in the background during meetings, lectures, and calls, then writes the summary for you.

Free plan to get started

No bot joins your calls

Phone, desktop & Chrome apps

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome