Analogy Between Gambling and Measurement-Based Work Extraction Dror - - PowerPoint PPT Presentation

analogy between gambling and measurement based work
SMART_READER_LITE
LIVE PREVIEW

Analogy Between Gambling and Measurement-Based Work Extraction Dror - - PowerPoint PPT Presentation

Analogy Between Gambling and Measurement-Based Work Extraction Dror Vinkler 1 Haim Permuter 1 Neri Merhav 2 1 Ben Gurion University 2 Technion - Israel Institute of Technology ISIT 2014 1 / 17 Outline Past results and brief physics background


slide-1
SLIDE 1

Analogy Between Gambling and Measurement-Based Work Extraction

Dror Vinkler 1 Haim Permuter 1 Neri Merhav 2

1Ben Gurion University 2Technion - Israel Institute of Technology

ISIT 2014

1 / 17

slide-2
SLIDE 2

Outline

Past results and brief physics background The analogy Gambling on continuous random variables Consequences

Universal engine Memory

Summary and future work

2 / 17

slide-3
SLIDE 3

Past Results - Horse Race Gambling

Kelly 1956

A race of m horses.

3 / 17

slide-4
SLIDE 4

Past Results - Horse Race Gambling

Kelly 1956

A race of m horses. Xi - winning horse. Yi - side information.

3 / 17

slide-5
SLIDE 5

Past Results - Horse Race Gambling

Kelly 1956

A race of m horses. Xi - winning horse. Yi - side information. Xi, Yi are pairwise i.i.d.

3 / 17

slide-6
SLIDE 6

Past Results - Horse Race Gambling

Kelly 1956

A race of m horses. Xi - winning horse. Yi - side information. Xi, Yi are pairwise i.i.d. Sn - gambler’s capital after n rounds, S1 = bX|Y (X1|Y1)oX(X1)S0, bX|Y (X|Y ) - betting strategy.

  • X(X)
  • odds.

3 / 17

slide-7
SLIDE 7

Past Results - Horse Race Gambling

Kelly 1956

A race of m horses. Xi - winning horse. Yi - side information. Xi, Yi are pairwise i.i.d. Sn - gambler’s capital after n rounds, S1 = bX|Y (X1|Y1)oX(X1)S0, bX|Y (X|Y ) - betting strategy.

  • X(X)
  • odds.

Sn =

n

  • i=1

bX|Y (Xi|Yi)oX(Xi)S0,

3 / 17

slide-8
SLIDE 8

Past Results - Horse Race Gambling

Kelly 1956

The goal: finding the optimal betting strategy b∗

X|Y ,

b∗

X|Y = arg max bX|Y

E [log Sn] . log Sn - capital growth.

4 / 17

slide-9
SLIDE 9

Past Results - Horse Race Gambling

Kelly 1956

The goal: finding the optimal betting strategy b∗

X|Y ,

b∗

X|Y = arg max bX|Y

E [log Sn] . log Sn - capital growth. This optimal betting strategy is b∗

X|Y = PX|Y .

4 / 17

slide-10
SLIDE 10

Past Results - Horse Race Gambling

Kelly 1956

The goal: finding the optimal betting strategy b∗

X|Y ,

b∗

X|Y = arg max bX|Y

E [log Sn] . log Sn - capital growth. This optimal betting strategy is b∗

X|Y = PX|Y .

A bet is fair if oX(x) = 1/PX(x) ∀x.

4 / 17

slide-11
SLIDE 11

Past Results - Horse Race Gambling

Kelly 1956

The goal: finding the optimal betting strategy b∗

X|Y ,

b∗

X|Y = arg max bX|Y

E [log Sn] . log Sn - capital growth. This optimal betting strategy is b∗

X|Y = PX|Y .

A bet is fair if oX(x) = 1/PX(x) ∀x. For a fair bet, the maximal growth is nI(X; Y ).

4 / 17

slide-12
SLIDE 12

Brief Background on Statistical Mechanics

For large systems, classical analysis is not possible.

5 / 17

slide-13
SLIDE 13

Brief Background on Statistical Mechanics

For large systems, classical analysis is not possible. Principle 1:

5 / 17

slide-14
SLIDE 14

Brief Background on Statistical Mechanics

For large systems, classical analysis is not possible. Principle 1:

X - state of particle.

X = 0 X = 1 X = 2 X = 3 X = 4 E

5 / 17

slide-15
SLIDE 15

Brief Background on Statistical Mechanics

For large systems, classical analysis is not possible. Principle 1:

X - state of particle. For a fixed temperature, PX(x) ∼ e−E(x)/kBT .

X = 0 X = 1 X = 2 X = 3 X = 4 E

PX(x) - the Boltzmann distribution.

5 / 17

slide-16
SLIDE 16

Brief Background on Statistical Mechanics

For large systems, classical analysis is not possible. Principle 1:

X - state of particle. For a fixed temperature, PX(x) ∼ e−E(x)/kBT .

X = 0 X = 1 X = 2 X = 3 X = 4 E

PX(x) - the Boltzmann distribution.

5 / 17

slide-17
SLIDE 17

Brief Background on Statistical Mechanics

For large systems, classical analysis is not possible. Principle 1:

X - state of particle. For a fixed temperature, PX(x) ∼ e−E(x)/kBT .

X = 0 X = 1 X = 2 X = 3 X = 4 E

PX(x) - the Boltzmann distribution. Principle 2:

5 / 17

slide-18
SLIDE 18

Brief Background on Statistical Mechanics

For large systems, classical analysis is not possible. Principle 1:

X - state of particle. For a fixed temperature, PX(x) ∼ e−E(x)/kBT .

X = 0 X = 1 X = 2 X = 3 X = 4 E

PX(x) - the Boltzmann distribution. Principle 2:

The second law of thermodynamics: ∆S ≥ 0. S - the entropy of the system.

5 / 17

slide-19
SLIDE 19

Brief Background on Statistical Mechanics

For large systems, classical analysis is not possible. Principle 1:

X - state of particle. For a fixed temperature, PX(x) ∼ e−E(x)/kBT .

X = 0 X = 1 X = 2 X = 3 X = 4 E

PX(x) - the Boltzmann distribution. Principle 2:

The second law of thermodynamics: ∆S ≥ 0. S - the entropy of the system. In a complete cycle: E[W] ≤ 0.

5 / 17

slide-20
SLIDE 20

Past Results - Measurement-Based Work Extraction

Originated with Maxwell’s demon in the 19th century.

6 / 17

slide-21
SLIDE 21

Past Results - Measurement-Based Work Extraction

Originated with Maxwell’s demon in the 19th century. Using fluctuation theorems, it was shown that [Sagawa and Ueda 2010] E[W] ≤ kBTI(X; Y ).

6 / 17

slide-22
SLIDE 22

Past Results - Measurement-Based Work Extraction

Originated with Maxwell’s demon in the 19th century. Using fluctuation theorems, it was shown that [Sagawa and Ueda 2010] E[W] ≤ kBTI(X; Y ). Some systems achieve this, but no general achievability scheme.

6 / 17

slide-23
SLIDE 23

Past Results - The Szilard Engine

Presented by Sagawa and Ueda as an example:

b c d

V = 1

7 / 17

slide-24
SLIDE 24

Past Results - The Szilard Engine

Presented by Sagawa and Ueda as an example:

c d

V = 1 V0(X)

7 / 17

slide-25
SLIDE 25

Past Results - The Szilard Engine

Presented by Sagawa and Ueda as an example:

d

V = 1 V0(X) Y

7 / 17

slide-26
SLIDE 26

Past Results - The Szilard Engine

Presented by Sagawa and Ueda as an example: V = 1 V0(X) Y Vf(X)

7 / 17

slide-27
SLIDE 27

Past Results - The Szilard Engine

Presented by Sagawa and Ueda as an example: V = 1 V0(X) Y Vf(X)

7 / 17

slide-28
SLIDE 28

Past Results - The Szilard Engine

Presented by Sagawa and Ueda as an example: V = 1 V0(X) Y Vf(X) X - the particle’s location. Y - a noisy measurement.

7 / 17

slide-29
SLIDE 29

Past Results - The Szilard Engine

Presented by Sagawa and Ueda as an example: V = 1 V0(X) Y Vf(X) X - the particle’s location. Y - a noisy measurement. Extracted work is W = kBT ln Vf(X|Y ) V0(X) .

7 / 17

slide-30
SLIDE 30

Past Results - The Szilard Engine

Optimal value of Vf(X|Y ) is V ∗

f (X|Y ) = PX|Y (X|Y ).

8 / 17

slide-31
SLIDE 31

Past Results - The Szilard Engine

Optimal value of Vf(X|Y ) is V ∗

f (X|Y ) = PX|Y (X|Y ).

Maximal extracted work is max E[W] = kBTI(X; Y ).

8 / 17

slide-32
SLIDE 32

Past Results - The Szilard Engine

Optimal value of Vf(X|Y ) is V ∗

f (X|Y ) = PX|Y (X|Y ).

Maximal extracted work is max E[W] = kBTI(X; Y ). Upper bound is achieved.

8 / 17

slide-33
SLIDE 33

Past Results - The Szilard Engine

Optimal value of Vf(X|Y ) is V ∗

f (X|Y ) = PX|Y (X|Y ).

Maximal extracted work is max E[W] = kBTI(X; Y ). Upper bound is achieved. Result is specific to one particle of ideal gas.

8 / 17

slide-34
SLIDE 34

Main Results

Gambling Maxwell’s Demon Side information Measurements results

9 / 17

slide-35
SLIDE 35

Main Results

Gambling Maxwell’s Demon Side information Measurements results

  • X - odds

1/V0 - initial vol.

9 / 17

slide-36
SLIDE 36

Main Results

Gambling Maxwell’s Demon Side information Measurements results

  • X - odds

1/V0 - initial vol. bX|Y - betting strategy Vf - final vol.

9 / 17

slide-37
SLIDE 37

Main Results

Gambling Maxwell’s Demon Side information Measurements results

  • X - odds

1/V0 - initial vol. bX|Y - betting strategy Vf - final vol. log Sn - log of capital Wn - extracted work

9 / 17

slide-38
SLIDE 38

Main Results

Gambling Maxwell’s Demon Side information Measurements results

  • X - odds

1/V0 - initial vol. bX|Y - betting strategy Vf - final vol. log Sn - log of capital Wn - extracted work log Sn =

n

  • i=1

log bX|Y (Xi|Yi)oX(Xi) , Wn =

n

  • i=1

kBT ln Vf(Xi|Yi) V0(Xi) .

9 / 17

slide-39
SLIDE 39

Some Immediate Consequences

  • X ⇔ 1/V0 = 1/PX

10/ 17

slide-40
SLIDE 40

Some Immediate Consequences

  • X ⇔ 1/V0 = 1/PX ⇒ In physics, the bet is always fair.

10/ 17

slide-41
SLIDE 41

Some Immediate Consequences

  • X ⇔ 1/V0 = 1/PX ⇒ In physics, the bet is always fair.

Second law of thermo. - without measurements E[W] ≤ 0.

10/ 17

slide-42
SLIDE 42

Some Immediate Consequences

  • X ⇔ 1/V0 = 1/PX ⇒ In physics, the bet is always fair.

Second law of thermo. - without measurements E[W] ≤ 0. Enables straightforward generalization to m dividers V ∗

f (X|Y ) = PX|Y (X|Y ).

10/ 17

slide-43
SLIDE 43

Recap

What we achieved so far:

11/ 17

slide-44
SLIDE 44

Recap

What we achieved so far: One-to-one mapping.

11/ 17

slide-45
SLIDE 45

Recap

What we achieved so far: One-to-one mapping. Specific to Szilard engine.

11/ 17

slide-46
SLIDE 46

Continuous Random Variables - Physics

Horowitz and Parrondo 2011, Esposito and Van den Broeck 2011

A single particle in a potential field:

E0 fX x

12/ 17

slide-47
SLIDE 47

Continuous Random Variables - Physics

Horowitz and Parrondo 2011, Esposito and Van den Broeck 2011

A single particle in a potential field:

E0 E0 fX x x fX|y Y

12/ 17

slide-48
SLIDE 48

Continuous Random Variables - Physics

Horowitz and Parrondo 2011, Esposito and Van den Broeck 2011

A single particle in a potential field:

E0 E0 fX x x x fX|y fX|y Ef Y E0 → Ef

12/ 17

slide-49
SLIDE 49

Continuous Random Variables - Physics

Horowitz and Parrondo 2011, Esposito and Van den Broeck 2011

A single particle in a potential field:

E0 E0 fX x x x fX|y fX|y Ef Y E0 → Ef

12/ 17

slide-50
SLIDE 50

Continuous Random Variables - Physics

Horowitz and Parrondo 2011, Esposito and Van den Broeck 2011

A single particle in a potential field:

E0 E0 fX x x x fX|y fX|y Ef Y E0 → Ef

X - particle’s location. Y - noisy measurement.

12/ 17

slide-51
SLIDE 51

Continuous Random Variables - Physics

Horowitz and Parrondo 2011, Esposito and Van den Broeck 2011

A single particle in a potential field:

E0 E0 fX x x x fX|y fX|y Ef Y E0 → Ef

X - particle’s location. Y - noisy measurement. QX|y - the Boltzmann distribution that stems from Ef.

12/ 17

slide-52
SLIDE 52

Continuous Random Variables - Physics

Horowitz and Parrondo 2011, Esposito and Van den Broeck 2011

A single particle in a potential field:

E0 E0 fX x x x fX|y fX|y Ef Y E0 → Ef

X - particle’s location. Y - noisy measurement. QX|y - the Boltzmann distribution that stems from Ef. Theorem Q∗

X|y = arg min D(fX|y||QX|y) ∀y ∈ Y.

E[W] = kBT

  • I(X; Y ) − D(fX|Y ||Q∗

X|Y |fY )

  • .

12/ 17

slide-53
SLIDE 53

Continuous Random Variables - Gambling

What’s the analogous case in gambling?

13/ 17

slide-54
SLIDE 54

Continuous Random Variables - Gambling

What’s the analogous case in gambling? X is some continuous r.v., i.e., the price of a stock.

13/ 17

slide-55
SLIDE 55

Continuous Random Variables - Gambling

What’s the analogous case in gambling? X is some continuous r.v., i.e., the price of a stock. Betting strategy:

1 2 3 4

x

Horse Race

bX|y

13/ 17

slide-56
SLIDE 56

Continuous Random Variables - Gambling

What’s the analogous case in gambling? X is some continuous r.v., i.e., the price of a stock. Betting strategy:

1 2 3 4

x x

Horse Race Stocks

bX|y bX|y

13/ 17

slide-57
SLIDE 57

Continuous Random Variables - Gambling

What’s the analogous case in gambling? X is some continuous r.v., i.e., the price of a stock. Betting strategy:

1 2 3 4

x x

Horse Race Stocks

bX|y bX|y Theorem b∗

X|y = arg min D(fX|y||bX|y) ∀y ∈ Y.

E[log S∗

n] = n

  • I(X; Y ) − D(fX|Y ||b∗

X|Y |fY )

  • .

13/ 17

slide-58
SLIDE 58

Consequence 1 - Universal Engine

Gambling: X, Y discrete r.v. PX,Y is unknown.

14/ 17

slide-59
SLIDE 59

Consequence 1 - Universal Engine

Gambling: X, Y discrete r.v. PX,Y is unknown.

  • b - the universal portfolio [Cover and Ordentlich 1996].

14/ 17

slide-60
SLIDE 60

Consequence 1 - Universal Engine

Gambling: X, Y discrete r.v. PX,Y is unknown.

  • b - the universal portfolio [Cover and Ordentlich 1996].

Theorem lim

n→∞

1 nE[log Sn − log S∗

n] = 0.

  • Sn - capital using
  • b. S∗

n - capital using b∗ X|Y .

14/ 17

slide-61
SLIDE 61

Consequence 1 - Universal Engine

Gambling: X, Y discrete r.v. PX,Y is unknown.

  • b - the universal portfolio [Cover and Ordentlich 1996].

Theorem lim

n→∞

1 nE[log Sn − log S∗

n] = 0.

  • Sn - capital using
  • b. S∗

n - capital using b∗ X|Y .

Physics: X, Y discrete r.v. Unknown measurement error PY |X.

14/ 17

slide-62
SLIDE 62

Consequence 1 - Universal Engine

Gambling: X, Y discrete r.v. PX,Y is unknown.

  • b - the universal portfolio [Cover and Ordentlich 1996].

Theorem lim

n→∞

1 nE[log Sn − log S∗

n] = 0.

  • Sn - capital using
  • b. S∗

n - capital using b∗ X|Y .

Physics: X, Y discrete r.v. Unknown measurement error PY |X. We developed the analogous universal control protocol.

14/ 17

slide-63
SLIDE 63

Consequence 1 - Universal Engine

Gambling: X, Y discrete r.v. PX,Y is unknown.

  • b - the universal portfolio [Cover and Ordentlich 1996].

Theorem lim

n→∞

1 nE[log Sn − log S∗

n] = 0.

  • Sn - capital using
  • b. S∗

n - capital using b∗ X|Y .

Physics: X, Y discrete r.v. Unknown measurement error PY |X. We developed the analogous universal control protocol. Theorem lim

n→∞

1 nE[ Wn − W ∗

n] = 0.

14/ 17

slide-64
SLIDE 64

Consequence 2 - Memory

Gambling: Dependent races. At round i, Xi−1, Y i are known.

15/ 17

slide-65
SLIDE 65

Consequence 2 - Memory

Gambling: Dependent races. At round i, Xi−1, Y i are known. Theorem [Permuter, Kim and Weissman 2011] b∗

Xi|Xi−1,Y i = PXi|Xi−1,Y i.

E[log Sn(Xn||Y n)] − E[log Sn(Xn)] = I(Y n → Xn). Sn(Xn||Y n) - capital with causal side information. Sn(Xn)

  • capital without side information.

15/ 17

slide-66
SLIDE 66

Consequence 2 - Memory

Gambling: Dependent races. At round i, Xi−1, Y i are known. Theorem [Permuter, Kim and Weissman 2011] b∗

Xi|Xi−1,Y i = PXi|Xi−1,Y i.

E[log Sn(Xn||Y n)] − E[log Sn(Xn)] = I(Y n → Xn). Sn(Xn||Y n) - capital with causal side information. Sn(Xn)

  • capital without side information.

Physics: Dependence through fast cycles, hysteresis, etc.

15/ 17

slide-67
SLIDE 67

Consequence 2 - Memory

Gambling: Dependent races. At round i, Xi−1, Y i are known. Theorem [Permuter, Kim and Weissman 2011] b∗

Xi|Xi−1,Y i = PXi|Xi−1,Y i.

E[log Sn(Xn||Y n)] − E[log Sn(Xn)] = I(Y n → Xn). Sn(Xn||Y n) - capital with causal side information. Sn(Xn)

  • capital without side information.

Physics: Dependence through fast cycles, hysteresis, etc. Theorem Q∗

Xi|Xi−1,Y i = PXi|Xi−1,Y i.

E[Wn(Xn||Y n) − Wn(Xn)] = kBTI(Y n → Xn).

15/ 17

slide-68
SLIDE 68

Summary

One-to-one mapping between gambling and measurement-based work extraction.

16/ 17

slide-69
SLIDE 69

Summary

One-to-one mapping between gambling and measurement-based work extraction. Extension of gambling to continuous random variables.

16/ 17

slide-70
SLIDE 70

Summary

One-to-one mapping between gambling and measurement-based work extraction. Extension of gambling to continuous random variables. Universal work extraction.

16/ 17

slide-71
SLIDE 71

Summary

One-to-one mapping between gambling and measurement-based work extraction. Extension of gambling to continuous random variables. Universal work extraction. Analysis of work extraction with cycle dependence.

16/ 17

slide-72
SLIDE 72

Summary

One-to-one mapping between gambling and measurement-based work extraction. Extension of gambling to continuous random variables. Universal work extraction. Analysis of work extraction with cycle dependence. Full version available on arXiv: http://arxiv.org/abs/1404.6788

16/ 17

slide-73
SLIDE 73

Future Work

Multiple particles.

17/ 17

slide-74
SLIDE 74

Future Work

Multiple particles. Generalize universal portfolios for continuous r.v.

17/ 17

slide-75
SLIDE 75

Future Work

Multiple particles. Generalize universal portfolios for continuous r.v.

Thank you !

17/ 17

slide-76
SLIDE 76

Future Work

Multiple particles. Generalize universal portfolios for continuous r.v.

Thank you !

17/ 17

slide-77
SLIDE 77

Future Work

Multiple particles. Generalize universal portfolios for continuous r.v.

Thank you !

17/ 17

slide-78
SLIDE 78

Future Work

Multiple particles. Generalize universal portfolios for continuous r.v.

Thank you !

17/ 17

slide-79
SLIDE 79

Maximization Over PX

In the Szilard engine: Control over PX through V0. Maximal extracted work is max E[W] = kBT max

PX I(X; Y ).

17/ 17

slide-80
SLIDE 80

Maximization Over PX

In the Szilard engine: Control over PX through V0. Maximal extracted work is max E[W] = kBT max

PX I(X; Y ).

In gambling: Control over PX through choice of race track. Maximal capital growth is max E[log Sn] = n max

PX∈P I(X; Y ).

17/ 17