Sharpening Occam’s razor with Quantum Mechanics SISSA Journal Club Matteo Marcuzzi 8th April, 2011
Niclas Koppernigck (Copernicus) Clausius Ptolemaeus (Ptolemy) Tyge Brahe (Tychonis) Describing Systems
Johannes Kepler
Describing Systems
Algorithmic Abstraction
Describing Systems Algorithmic Abstraction Same output
Describing Systems Same output Different intrinsic information! Solar system celestial objects Sun Flares Planet Orography Meteorology People behaviour Compton Scattering
Describing Systems Same output Different intrinsic information! Much more memory required! OCCAM’S RAZOR
Describing Systems N Spin Chain Up parity 1 spin-flip per second if even if odd 0
Describing Systems N Spin Chain Up parity if even if odd 0 1 spin-flip per second 1
Describing Systems N Spin Chain Up parity if even if odd 0 1 spin-flip per second 10
Describing Systems N Spin Chain Up parity if even if odd 0 1 spin-flip per second 101
Describing Systems N Spin Chain Up parity if even if odd 0 1 spin-flip per second 1010
Describing Systems N Spin Chain Up parity if even if odd 0 1 spin-flip per second 10101
Describing Systems N Spin Chain Up parity if even if odd 0 1 spin-flip per second
Describing Systems N Spin Chain Up parity if even if odd 0 1 spin-flip per second
Describing Systems N Spin Chain Up parity if even if odd 0 1 spin-flip per second N bits needed
Describing Systems Hidden System read x return (x+1) mod 2 1-bit only! Statistically equivalent output N bits
Computational Mechanics Statistical equivalence Measure of complexity Pattern identification
Computational Mechanics Statistical equivalence Measure of complexity Pattern identification
Computational Mechanics Statistical equivalence Measure of complexity Pattern identification
Computational Mechanics ? Statistical equivalence Measure of complexity Pattern identification
Computational Mechanics Stochastic Process Discrete Stationary Random Variables Alphabet
Computational Mechanics Stochastic Process Discrete Stationary Random Variables Alphabet Pasts Futures
Computational Mechanics Stochastic Process Discrete Stationary Random Variables Alphabet Set of histories Set of future strings
Computational Mechanics Stochastic Process Discrete Stationary Machine … Statistical Equivalence
Computational Mechanics Stochastic Process Discrete Stationary Machine … … … States Partition R
Computational Mechanics Stochastic Process Discrete Stationary Machine States Partition R
a Computational Mechanics Stochastic Process Discrete Stationary Machine Transition Rates
Computational Mechanics Stochastic Process Discrete Stationary OCCAM POOL
Computational Mechanics A little information theory Shannon entropy Conditional entropy Mutual information Excess entropy
Computational Mechanics MachineCannot distinguish between them Partition R We want to preserve information
Computational Mechanics Machine Partition R We want to preserve information with the least possible memory Log(# states)minimize
Computational Mechanics Machine Partition R We want to preserve information with the least possible memory minimize Statistical complexity
Computational Mechanics We want to preserve information with the least possible memory minimize Statistical complexity OCCAM POOLOptimal partition
Computational Mechanics Optimal partition We want to preserve information with the least possible memory minimize Statistical complexity ε-machine ε if Causal States (unique)
Computational Mechanics: Examples 2-periodic sequence 2-periodic, ends with A B I initial state
Computational Mechanics: Examples 2-periodic sequence A B I initial state recurrent transient
Computational Mechanics: Examples 1 D Ising model transfer matrix
Computational Mechanics: Examples 1 D Next-nearest-neighbours Ising 2
Computational Mechanics: Examples 1 D Next-nearest-neighbours Ising 2 3
Computational Mechanics: Examples 1 D Next-nearest-neighbours Ising 2 3 1
Computational Mechanics: Examples 1 D Next-nearest-neighbours Ising 2 3 1
Computational Mechanics: Examples 1 D Next-nearest-neighbours Ising negligible
Computational Mechanics: Examples 1 D Next-nearest-neighbours Ising period 3 period 1
Sharpening the razor with QM Statistical complexity Excess entropy Ideal system
Sharpening the razor with QM ε ε-machines are deterministic ε
ε Sharpening the razor with QM
ε fixed i,cunique j fixed j,cunique i ideal
Sharpening the razor with QM ε qεqε causal state Risystem state symbol “s”symbol state q-machine states
qεqε system state symbol state q-machine states Sharpening the razor with QM CLASSICAL QUANTUM Prepare Measure C.S. Probability t
qεqε system state symbol state q-machine states Sharpening the razor with QM CLASSICAL QUANTUM
qεqε system state symbol state q-machine states Sharpening the razor with QM CLASSICAL QUANTUM Ideal system
qεqε system state symbol state q-machine states Sharpening the razor with QM CLASSICAL QUANTUM Non-ideal systems Quantum mechanics improves efficiency
Sharpening the razor with QM single spin ?
References M. Gu, K. Wiesner, E. Rieper & V. Vedral - "Sharpening Occam's razor with Quantum Mechanics" - arXiv: quant-ph/ v2 (2011) C. R. Shalizi & J. P. Crutchfield - "Computational Mechanics: Pattern and Prediction, Structure and Simplicity" - arXiv: cond-mat/990717v2 (2008) D. P. Feldman & J. P. Crutchfield - "Discovering Noncritical Organization: Statistical Mechanical, Information Theoretic, and Computational Views of Patterns in One-Dimensional Spin Systems" - Santa Fe Institute Working Paper (1998)