Golden Ratio & Fibonacci Spiral

Build Fibonacci squares, compare consecutive ratios, and see how the recurrence approaches the golden ratio.

Loading interactive simulation...

why phi is the fixed point 🖖

The Fibonacci recurrence keeps replacing a ratio with 1 plus its reciprocal: F(n)/F(n-1) = 1 + F(n-2)/F(n-1). Once that ratio stabilizes at x, it must satisfy x = 1 + 1/x, so x² = x + 1 and the positive solution is phi. The spiral is a geometric picture of that same self-similarity: each larger square contains the previous rectangle plus one new square.

The ratio zigzags in toward phi 🖖

Watch the bar chart: consecutive Fibonacci ratios land alternately just above and just below φ, and the gap shrinks at every step. By 21/13 = 1.615 you are already within 0.003 of φ, and a few terms later the ratio matches φ to as many decimals as the display can show. Convergence is fast, and no single overshoot ever stays far off for long.

The most irrational number 🖖

Written as a continued fraction, φ = [1; 1, 1, 1, …] — nothing but ones, the smallest coefficients possible. That makes φ the hardest number to approximate well with any fraction, and its best approximations are exactly the Fibonacci ratios this tool plots. Nature exploits this: seeds placed 360°/φ² ≈ 137.5° apart never line up into gaps, giving sunflowers and pinecones their tight, even packing.

Problem solved in full

  1. Error measuring 144/89 = 1.617978 against φ = 1.618034 6 steps

    144/89 = 1.617978 against φ = 1.618034. Predict that error before measuring it, then explain why φ is the worst number in mathematics to approximate.

    1. The defining equation is not assumed, it is forced: if the ratio of consecutive terms settles anywhere, the recurrence says the limit satisfies this.

    2. Solve the quadratic and take the positive root.

    3. The twelfth convergent, and the error you are asked to predict. It is negative, and the previous one was positive — the convergents alternate around φ.

    4. Binet's formula gives the error in closed form. The conjugate root ψ is what the Fibonacci numbers throw away, and it is exactly what the approximation is missing.

    5. Compare against Hurwitz's theorem, which is the strongest statement possible about approximating any irrational by rationals. The error is not merely below the bound, it is on it.

    6. The continued fraction says why. All ones is the slowest convergence available, and the golden angle is that fact turned into a growth rule.

    Answer

    The error is ψ¹¹/F₁₁ = −5.6461 × 10⁻⁵, and it sits exactly on the floor no irrational can go under. Hurwitz proved that every irrational α has infinitely many rationals with |p/q − α| < 1/(√5 q²), and that the √5 cannot be improved. Here 1/(√5 × 89²) = 5.6459 × 10⁻⁵ — the actual error and the theoretical floor agree to one part in forty thousand. That is what "the most irrational number" means: not that φ is mysterious, but that its continued fraction [1;1,1,1,…] never offers a large digit, so no convergent ever gets a bargain and every one of them lands on the bound. Plants exploit it. Seeds placed 360°/φ² = 137.508° apart never fall into rows, because falling into rows means the angle is near a simple fraction of a turn — and φ is provably the number furthest from every simple fraction there is.

References (1)

Example problems

  • early wobble - First 5 Fibonacci terms: ratio 5/3 = 1.667 — visibly above φ = 1.618
  • near phi - 10 terms: ratio 89/55 = 1.618181… — settling within 0.004% of φ
  • locked in - 16 terms: ratio 1597/987 = 1.6180339… — matches φ to 7 significant figures
  • Binet check - Binet's formula: F(20) = 6765 from closed-form expression using φ and ψ