AI summary by TwinMind
The Real Butterfly Effect
Sabine Hossenfelder
6m 11s
Historical Origins and Misconceptions
- Common butterfly effect interpretation: small disturbances in chaotic systems grow to have large consequences
- Name popularized by James Gleick in 1987 book "Chaos"
- Usually attributed to meteorologist Edward Lorenz
- Recent paper by Tim Palmer, Andreas Doering, Gregory Eyink revealed misunderstanding of Lorenz's original meaning
Lorenz's Original 1969 Paper
- Lorenz did not write about butterfly wings initially
- Referenced seagulls wings, attributed to unnamed meteorologist
- Butterfly reference came from 1972 talk titled "Does the flap of a butterfly's wings in Brazil set off a tornado in Texas"
- Talk title suggested by session chair Phil Merilees (meteorologist)
- 1972 talk was summary of 1969 paper findings
Standard vs Real Butterfly Effect
Standard Butterfly Effect
- Chaotic systems sensitive to initial conditions
- Small uncertainties in initial knowledge eventually blow up
- If initial state precisely known, precise predictions possible
- Deterministic system with unlimited prediction potential given perfect data
Real Butterfly Effect (Lorenz 1969)
- Much stronger claim than standard interpretation
- System divided into equal-sized pieces with detectors
- Each detector measures quantities needed for prediction (air pressure, temperature)
- Assumes arbitrarily accurate measurements
Grid Resolution and Prediction Limits
- Typical grid: ~1000 kilometers → 5-day prediction accuracy
- Errors from smaller scales corrupt predictions after 5 days
- Decreasing grid by factor of 2 → only increases prediction time by half original time
- 5 days + 2.5 days = 7.5 days total
- Further grid refinement follows same pattern:
- 5 + 2.5 + 1.25 + ... days
- Infinite series converges to finite limit: 10 days maximum
Core Implications
- Even with arbitrarily fine grid and precise initial conditions
- Predictions only possible for finite time period
- System remains deterministic yet fundamentally non-predictable beyond time limit
- Errors from sub-grid scales continuously corrupt larger-scale predictions
Mathematical and Physical Reality
- Differential equations exist with real butterfly effect behavior
- Unclear if Navier-Stokes equations (weather modeling) exhibit real butterfly effect
- Currently unsolved problem in mathematics
- Macroscopic equations are approximations
- Fundamental level: particle physics and quantum mechanics
- Quantum mechanics equations linear → no butterfly effects
- Quantum mechanics impractical for weather prediction
Summary
- Real butterfly effect: deterministic systems may have finite predictability limits
- Even perfect knowledge cannot extend predictions beyond certain time horizons
- Distinguishes between sensitivity to initial conditions vs fundamental predictability limits
