Computation at the Nanoscale:

Size: px

Start display at page:

Download "Computation at the Nanoscale:"

Jean Griffin
5 years ago
Views:

Computation at the Nanoscale: thoughts on information processing in

Sciences Center Physics Department University of California at Davis

Many Dimensions Telluride, Colorado 16 June 2013 Joint work with

1 Computation at the Nanoscale: thoughts on information processing in novel materials, molecules, & atoms Jim Crutchfield Complexity Sciences Center Physics Department University of California at Davis Telluride Summer workshop on The Complexity of Dynamics & Kinetics in Many Dimensions Telluride, Colorado 16 June 2013 Joint work with Dowman Varn (UCD) & Paul Riechers (UCD) Copyright James P. Crutchfield 1

2 Agenda History of Substrates Information Processing Intrinsic Computation Applications Looking Forward 2

History of Computing Substrates Mechanical: Gears Electron tube circuits Gates: Electron tubes, semiconductors Memory: Mercury delay lines, storage scopes,.

3 History of Computing Substrates Mechanical: Gears Electron tube circuits Gates: Electron tubes, semiconductors Memory: Mercury delay lines, storage scopes,... Molecular Computing (1970s) Physics and Computation (MIT Endicott House 1981) Quantum Computing (Feynman there) Josephson Junction Computers (IBM 1980s) Nanotech (Drexler/Merkle XEROX PARC 1990s) Design goal: Useful computing 3

4 History Of Intrinsic Computing Nature already computes Information: H(Pr(X)) (Shannon 1940s) In chaotic dynamics: (Kolmogorov 1950s) h µ Physics and Computation (MIT Endicott House 1981) Intrinsic computing there too! 4

5 Information Processing 5

6 Information-Theoretic Analysis of Complex Systems... Process Pr( X, X ) is a communication channel X X from the past to the future : Past Present Future Channel 6

7 Information-Theoretic Analysis of Complex Systems... Process Pr( X, X ) is a communication channel X X from the past to the future : Past Present Future Information Rate h µ Channel Capacity C 6

8 Information-Theoretic Analysis of Complex Systems... Process Pr( X, X ) is a communication channel X X from the past to the future : Past Present Future Information Rate h µ Channel Capacity C Channel Utilization: Excess Entropy E = I[ X ; X ] 6

9 Roadmap to Information(s) H(L) µ E + h L H(L) E Block Entropy H(L) H[! X L ] E T h L µ 0 0 L J. P. Crutchfield and D. P. Feldman, Regularities Unseen, Randomness Observed: Levels of Entropy Convergence, CHAOS 13:1 (2003)

10 Is Information Theory Sufficient? No! Measurements = process states? Wrong! Hidden processes No direct measure of structure 8

11 Intrinsic Computation (1) How much of past does process store? (2) In what architecture is that information stored? (3) How is stored information used to produce future behavior? 9

12 Computational Mechanics: What are the hidden states? Group all histories that give same prediction: ( x )={ x : Pr( X x ) = Pr( X x )} Equivalence relation: x x Equivalence classes are process s causal states: S = Pr( X, X )/!-Machine: Optimal, minimal, unique predictor. J. P. Crutchfield, K. Young, Inferring Statistical Complexity, Physical Review Letters 63 (1989)

13 Computational Mechanics!-Machine: M = S, {T (x) : x A} Dynamic: State State A B Transient States T (x), = Pr(,x) C , S Recurrent States D 11

14 Varieties of ε-machine Denumerable Causal States Fractal J. P. Crutchfield, Calculi of Emergence: Computation, Dynamics, and Induction, Physica D 75 (1994) Continuous 12

15 Kinds of Intrinsic Computing Directly from!-machine: Stored information (Statistical complexity): C µ = Pr( ) log 2 Pr( ) S Information production (Entropy rate): h µ = S Pr( ) S,s A Pr( s ) log 2 Pr( s ) 13

16 Computational Mechanics Theorem (Causal Shielding): Pr( X, X S) = Pr( X S)Pr( X S) Theorem (Optimal Prediction): Pr( X S) = Pr( X X ) Corollary (Capture All Shared Information): I[S; X ]=E (Prescient models) Theorem:!-Machine is smallest prescient model C µ H[S] H[R] 14

17 Prediction V. Modeling Hidden: State information via measurement. So, how accessible is state information? How do measurements reveal internal states? Quantitative version: Prediction ~ Modeling ~ E C µ 15

18 Information Accessibility How hidden is a hidden Process? Crypticity: = C µ E Stored Information Apparent Information 16

19 Summary Information stored in the present is not that shared between the past and the future. 17

20 Cautionary Cryptic Processes: Excess IC entropy can be arbitrarily small ( E 0). Even for very structured ( ) processes. C µ 1 Care when applying informational analyses to complex systems. Best to focus on causal architecture, then calculate what you need. 18

21 Intrinsic Computation (1) How much of past does process store? (2) In what architecture is that information stored? (3) How is stored information used to produce future behavior? 19

22 Intrinsic Computation (1) How much of past does process store? C µ (2) In what architecture is that information stored? (3) How is stored information used to produce future behavior? 19

23 Intrinsic Computation (1) How much of past does process store? (2) In what architecture is that information stored? n o S, {T (s),s2a} (3) How is stored information used to produce future behavior? C µ 19

24 Intrinsic Computation (1) How much of past does process store? (2) In what architecture is that information stored? n o S, {T (s),s2a} (3) How is stored information used to produce future behavior? C µ h µ 19

25 Applications Chaotic Crystallography Single Molecule Dynamics Atomic Computing 20

$diffraction studies", Acta Cryst.$

26 Chaotic Crystallography via ε-machine Spectral Reconstruction Close-packed structures: Polytypes (semiconductors) ε-msr D. P. Varn, G. S. Canright, and J. P. Crutchfield, "ε-machine spectral reconstruction theory: A direct method for inferring planar disorder and structure from X-ray diffraction studies", Acta Cryst. Sec. A 69:2 (2013)

27 Designer Semiconductors Hypothesis: Structure key to computational & physical properties.!msr: New theory of structure in disordered materials Infer intrinsic computation Calculate new physical properties (length scales, interaction energy,...) Exotic semiconductors = Rational design of polytypes: Identify!M with desired physical+informational properties Run!MSR backwards to assemble polytypic materials Desired properties in an ensemble of realizations, reduces complexity of assembly 22

28 Molecular Dynamics Spectroscopy C.-B. Li, H. Yang, & T. Komatsuzaki, Proc. Natl. Acad. Sci USA 105:2 (2008)

Atomic Computing Chaotic ionization mechanism: Nonlinear turnstiles Momentum pr Rydberg atoms: Isolated, highly excited electron states K. Burke, K.

29 Atomic Computing Chaotic ionization mechanism: Nonlinear turnstiles Momentum pr Rydberg atoms: Isolated, highly excited electron states K. Burke, K. Mitchell, B. Wyker, S.Ye, and F. B. Dunning, Demonstration of Turnstiles as a Chaotic Ionization Mechanism in Rydberg Atoms, Physical Review Letters 107 (2011)

30 Atomic Computing Measured ionization well predicted (classically!) Current work: Phase shift as a function of T. Small displacements in T are applied to 1D and experimental data to separate markers. Intrinsic computational analysis via!m Embed logic gates in turnstile dynamics 25

31 Atomic Computing Couple to build circuits... Rydberg Computers? Optical lattice of Rydberg atoms: Rydberg atoms in solid-state materials? 26

32 Looking forward Theory of Computational Mechanics: Complete, closed-form analysis of intrinsic computing. Experiment: Analyze intrinsic computation in dynamic, nonlinear nanosystems. Information Engine UC Davis: Workshops, visit Davis, collaborate,...! 27

33 Thanks! J. P. Crutchfield, C. J. Ellison, and J. R. Mahoney, Time's Barbed Arrow: Irreversibility, Crypticity, and Stored Information, Physical Review Letters 103:9 (2009) C. J. Ellison, J. R. Mahoney, and J. P. Crutchfield, Prediction, Retrodiction, and the Amount of Information Stored in the Present, Journal of Statistical Physics 137:6 (2009) J. P. Crutchfield, C. J. Ellison, J. R. Mahoney, and R. G. James, Synchronization and Control in Intrinsic and Designed Computation: An Information-Theoretic Analysis of Competing Models of Stochastic Computation, CHAOS 20:3 (2010) J. P. Crutchfield, Between Order and Chaos, Nature Physics 8 (January 2012) D. P. Varn, G. S. Canright, and J. P. Crutchfield, "!-Machine spectral reconstruction theory: A direct method for inferring planar disorder and structure from X-ray diffraction studies", Acta Cryst. Sec. A 69:2 (2013)

Demon Dynamics: Deterministic Chaos, the Szilard Map, & the Intelligence of Thermodynamic Systems.

Demon Dynamics: Deterministic Chaos, the Szilard Map, & the Intelligence of Thermodynamic Systems http://csc.ucdavis.edu/~cmg/ Jim Crutchfield Complexity Sciences Center Physics Department University of