SlideShare a Scribd company logo
Consequences of the Physical CTT over
experimental setups in quantum physics
Antonio Acín1, Ariel Bendersky1, Gonzalo de La Torre1,
Santiago Figueira2 y Gabriel Senno2
1ICFO-Institut de Ciencies Fotoniques, Barcelona, España
2Departamento de Computación, FCEN, Universidad de Buenos Aires
Workshop on Information and Physics,
Paris, 2015
The Physical Church-Turing Thesis
Physical CTT
Every function calculable by a physical system is
Turing-computable.
2
Outline
We present two consequences of the use of pseudorandomness
instead of randomness in experimental quantum physics:
A local model for Bell-like experiments in which Alice and
Bob use computable (but unknown to an eavesdropper)
inputs.
3
Outline
We present two consequences of the use of pseudorandomness
instead of randomness in experimental quantum physics:
A local model for Bell-like experiments in which Alice and
Bob use computable (but unknown to an eavesdropper)
inputs.
When preparing mixed states by computably sampling
pure states, the final preparations retains information on
how it was mixed.
3
Bell inequality computability loophole
4
Bell’s experiments
A source (S) prepares and distributes two physical systems to
distant observers Alice and Bob.
Upon receiving their systems, each observer performs a
measurement on it.
The object of interest is
p(a, b|x, y)
the joint probability distribution of obtaining outcomes a and b
when Alice and Bob choose measurements x and y.
5
Locality
In general,
p(a, b|x, y) = p(a|x)p(b|y)
Local explanation for distant correlations: past common cause,
λ.
p(a, b|x, y, λ) = p(a|x, λ)p(b|y, λ)
λ may not be constant over all runs.
Hence, in general, we say that a probability distribution is local
if it can be written as:
p(a, b|x, y) =
Λ
dλq(λ)p(a|x, λ)p(b|y, λ)
Measurement independence assumption:
q(λ|x, y) = q(λ)
6
CHSH inequality
Suppose x, y ∈ {0, 1} and a, b ∈ {−1, +1}, and consider
axby =
a,b
ab p(a, b|x, y)
Let
S = a0b0 + a0b1 + a1b0 − a1b1
Theorem (Clauser, Horne, Shimony, Holt)
Any local probability distribution has to satisfy,
S ≤ 2
7
Quantum theory is non-local
Predictions for the outcomes of some distant measurements on
entangled systems violate the previous inequality.
8
Quantum theory is non-local
Predictions for the outcomes of some distant measurements on
entangled systems violate the previous inequality.
For example, if the systems are prepared in the singlet state and
Alice and Bob measure in the following spin directions,
8
Quantum theory is non-local
Predictions for the outcomes of some distant measurements on
entangled systems violate the previous inequality.
For example, if the systems are prepared in the singlet state and
Alice and Bob measure in the following spin directions,
We have that
S = a0b0 + a0b1 + a1b0 − a1b1
S =
√
2
2
+
√
2
2
+
√
2
2
−
−
√
2
2
= 2
√
2
8
Loopholes
A loophole, in this context, is an experimental situation
allowing for local devices to generate experimental data
violating a Bell’s inequality.
Examples:
Detection loophole.
Finite statistics loophole.
Locality loophole.
In this work we present, the computability loophole.
9
The computational loophole
It is convenient for what follows to rephrase the standard Bell
scenario in cryptographic terms. In this approach, Alice and
Bob get their systems from a non-trusted provider Eve.
10
The computational loophole
It is convenient for what follows to rephrase the standard Bell
scenario in cryptographic terms. In this approach, Alice and
Bob get their systems from a non-trusted provider Eve.
We will allow Eve access to the inputs and outputs of previous
rounds of the experiment.
10
The computational loophole
It is convenient for what follows to rephrase the standard Bell
scenario in cryptographic terms. In this approach, Alice and
Bob get their systems from a non-trusted provider Eve.
We will allow Eve access to the inputs and outputs of previous
rounds of the experiment. This memory scenario still allows to
see non-locality (Barret et al PRA 66:042111, Pironio et al
Nature 464(7291):1021-1024, Pironio et al PRA 87:012336).
10
The computational loophole
It is convenient for what follows to rephrase the standard Bell
scenario in cryptographic terms. In this approach, Alice and
Bob get their systems from a non-trusted provider Eve.
We will allow Eve access to the inputs and outputs of previous
rounds of the experiment. This memory scenario still allows to
see non-locality (Barret et al PRA 66:042111, Pironio et al
Nature 464(7291):1021-1024, Pironio et al PRA 87:012336).
We show that,
Theorem
If Alice and Bob choose their measurements following an algorithm,
Eve can prepare devices that locally violate CHSH inequality.
10
Predicting computable functions
Definition
A class of total computable functions C is identifiable by next
value (C ∈ NV) if there exists a computable function g (called a
predictor for C) such that for every f ∈ C,
(∃n0)(∀n ≥ n0) f(n) = g( f(0), . . . , f(n − 1) ) (1)
11
Predicting computable functions
Definition
A class of total computable functions C is identifiable by next
value (C ∈ NV) if there exists a computable function g (called a
predictor for C) such that for every f ∈ C,
(∃n0)(∀n ≥ n0) f(n) = g( f(0), . . . , f(n − 1) ) (1)
Proposition (Putnam)
The class of all total computable functions is not identifiable by next
value.
11
Predicting computable functions
Definition
A class of total computable functions C is identifiable by next
value (C ∈ NV) if there exists a computable function g (called a
predictor for C) such that for every f ∈ C,
(∃n0)(∀n ≥ n0) f(n) = g( f(0), . . . , f(n − 1) ) (1)
Proposition (Putnam)
The class of all total computable functions is not identifiable by next
value.
Theorem (Adleman)
A class of total computable functions is in NV if and only if it is a
subclass of a complexity class (w.r.t. some complexity measure).
11
Explanation of the loophole
If
Alice and Bob choose their measurements using
computable functions,
Eve doesn’t know the functions, but knows a time
complexity class C that contains them and
on every round, Eve receives the choices of Alice and Bob
of previous rounds..
Then, Eve can prepare devices that
1 on every round, using the choices of previous rounds,
execute a predictor for C and guess what the new choices
will be and
2 output the necessary values for the statistic to violate
CHSH inequality.
12
Importance of the loophole
Since every computable function belongs to some
complexity class, Alice and Bob can never rule out the
possibility of Eve predicting their functions.
13
Importance of the loophole
Since every computable function belongs to some
complexity class, Alice and Bob can never rule out the
possibility of Eve predicting their functions.
Therefore no computable pseudo randomness criterion
will suffice for a proper Bell inequality violation.
13
Importance of the loophole
Since every computable function belongs to some
complexity class, Alice and Bob can never rule out the
possibility of Eve predicting their functions.
Therefore no computable pseudo randomness criterion
will suffice for a proper Bell inequality violation.
Other sources of randomness:
1 Quantum coins. Not desirable to assume a non-local theory,
like quantum mechanics, in order to test non-locality.
2 Free will. Can humans generate non-computable
sequences?
13
Mixing states computably
14
Two kinds of mixed states
Proper mixed states
Describe ensembles of pure states of which we have classical
uncertainty.
Improper mixed states
Describe systems which form part of bigger quantum system in
a pure state.
15
Case 1
R2D2 chooses from each box. The observer only knows
that R2D2 will pick half times each state but not how he’ll
pick each time.
16
Case 1
R2D2 chooses from each box. The observer only knows
that R2D2 will pick half times each state but not how he’ll
pick each time.
The state, as described by the observer is ρ = I
2 .
16
Case 2
C3PO chooses from each box. The observer only knows
that C3PO will pick half times each state but not how he’ll
pick each time.
17
Case 2
C3PO chooses from each box. The observer only knows
that C3PO will pick half times each state but not how he’ll
pick each time.
The state, as described by the observer is ρ = I
2 .
17
Undistiguishable
Both situations seem to be undistinguishable.
18
However...
... they are robots, so they can only choose in a computable
manner.
Any classical system used to choose only yields computable
choices.
19
Therefore
We will be able to distinguish both situations.
20
Distinguishing computable preparations
Assumption
We have a black box containing one of the two previous
situations and we want to know which one it is.
21
Distinguishing computable preparations
Assumption
We have a black box containing one of the two previous
situations and we want to know which one it is.
Procedure
We measure every odd qubit on the basis of eigenstates of σX
and every even qubit on the basis of eigenstates of σZ
21
Distinguishing computable preparations
22
Distinguishing computable preparations
Now what?
We obtain two sequences.
When we measure in the same basis as the preparation, the
sequence obtained is computable.
When we measure in the other basis, the sequence
obtained is a fair coin.
23
Distinguishing computable preparations
Now what?
We obtain two sequences.
When we measure in the same basis as the preparation, the
sequence obtained is computable.
When we measure in the other basis, the sequence
obtained is a fair coin.
Let’s go classical
Can we distinguish a computable sequence from one arising
from a fair coin with high probability of success?
We proved this to be true.
23
The distinguishing protocol
Idea
Almost all sequences arising from independent tosses of fair
coins have incompressible prefixes.
24
The distinguishing protocol
Input: k ∈ N and X, Z ∈ {0, 1}ω, two bit sequences with the
promise that one of them is computable.
Output: ‘X’ or ‘Z’ as the candidate for being computable;
wrong answer with probability bounded by O(2−k).
for t = 0, 1, 2 . . . do
for p = 0, . . . , t do
if Ut(p) = X k|p| then
output ‘X’ and halt
if Ut(p) = Z k|p| then
output ‘Z’ and halt
25
Probability of misrecognition
Probability of error
Perror ≤
>0
2
2k
=
2−(k−1)
1 − 2−(k−1)
which goes to 0 as k goes to ∞.
We pick a k such that the error is lower than what we want, and
then we run the recognition algorithm.
26
Some subtelties
Our algorithm, although it runs in finite time, is infeasible.
27
Some subtelties
Our algorithm, although it runs in finite time, is infeasible.
Still, the state has the information on how it was mixed.
This is surprising from a fundamental point of view.
27
Some subtelties
Our algorithm, although it runs in finite time, is infeasible.
Still, the state has the information on how it was mixed.
This is surprising from a fundamental point of view.
A slight variation on the algorithm makes it noise tolerant
for noise rates up to 0,21.
27
Assuming the Physical CTT
Physical CTT
Every function calculable by a physical system is
Turing-computable.
So, deterministic physical processes, if we accept the
impossibility of preparing non-computable initial conditions,
won’t do it.
28
Assuming the Physical CTT
Physical CTT
Every function calculable by a physical system is
Turing-computable.
So, deterministic physical processes, if we accept the
impossibility of preparing non-computable initial conditions,
won’t do it.
Consequences
We are left with quantum randomness. This means that:
Only kind of mixed state is those being part of larger
system in a pure state.
Need quantum randomness to test quantum non-locality.
28

More Related Content

Viewers also liked

Combined PAT Presentations
Combined PAT PresentationsCombined PAT Presentations
Combined PAT PresentationsJohn Osborne
 
ICT in School Education
ICT in School EducationICT in School Education
ICT in School Education
Vasudha Kamat
 
тварини і птахи нашої місцевості
тварини і птахи  нашої місцевостітварини і птахи  нашої місцевості
тварини і птахи нашої місцевості
maraphon
 
Pasos para realizar un blog (1)sissususus
Pasos para realizar un blog (1)sissusususPasos para realizar un blog (1)sissususus
Pasos para realizar un blog (1)sissususus
yosurileon13
 
Современные приёмы и средства повышения мотивации учащихся на
Современные приёмы и средства повышения мотивации учащихся наСовременные приёмы и средства повышения мотивации учащихся на
Современные приёмы и средства повышения мотивации учащихся на
Галина Свентуховская
 
Software testing tools
Software testing toolsSoftware testing tools
Software testing tools
Software Testing Books
 

Viewers also liked (6)

Combined PAT Presentations
Combined PAT PresentationsCombined PAT Presentations
Combined PAT Presentations
 
ICT in School Education
ICT in School EducationICT in School Education
ICT in School Education
 
тварини і птахи нашої місцевості
тварини і птахи  нашої місцевостітварини і птахи  нашої місцевості
тварини і птахи нашої місцевості
 
Pasos para realizar un blog (1)sissususus
Pasos para realizar un blog (1)sissusususPasos para realizar un blog (1)sissususus
Pasos para realizar un blog (1)sissususus
 
Современные приёмы и средства повышения мотивации учащихся на
Современные приёмы и средства повышения мотивации учащихся наСовременные приёмы и средства повышения мотивации учащихся на
Современные приёмы и средства повышения мотивации учащихся на
 
Software testing tools
Software testing toolsSoftware testing tools
Software testing tools
 

Similar to Workshop

Experimental one-way
Experimental one-wayExperimental one-way
Experimental one-way
CAA Sudan
 
Quantum Cryptography - Seminar report
Quantum Cryptography - Seminar reportQuantum Cryptography - Seminar report
Quantum Cryptography - Seminar report
Shyam Mohan
 
1542 inner products
1542 inner products1542 inner products
1542 inner products
Dr Fereidoun Dejahang
 
Introduction to Quantum Computing & Quantum Information Theory
Introduction to Quantum Computing & Quantum Information TheoryIntroduction to Quantum Computing & Quantum Information Theory
Introduction to Quantum Computing & Quantum Information Theory
Rahul Mee
 
What is the point of Boson sampling?
What is the point of Boson sampling?What is the point of Boson sampling?
What is the point of Boson sampling?
University of Glasgow Optical Sciences Seminar
 
Basic Concepts of Entanglement Measures
Basic Concepts of Entanglement MeasuresBasic Concepts of Entanglement Measures
Basic Concepts of Entanglement Measures
Ryohei Suzuki
 
Root locus description in lucid way by ME IITB
Root locus description in lucid way by ME IITBRoot locus description in lucid way by ME IITB
Root locus description in lucid way by ME IITB
AmreshAman
 
What happens if measure the electron spin twice?
What happens if measure the electron spin twice?What happens if measure the electron spin twice?
What happens if measure the electron spin twice?
Fausto Intilla
 
Quantum computing
Quantum computingQuantum computing
Quantum computing
Amr Kamel Deklel
 
The Extraordinary World of Quantum Computing
The Extraordinary World of Quantum ComputingThe Extraordinary World of Quantum Computing
The Extraordinary World of Quantum Computing
Tim Ellison
 
Quantum Computing
Quantum ComputingQuantum Computing
Quantum Computing
Amr Mohamed
 
Quantum entanglement is one of the most intriguing and counterintuitive pheno...
Quantum entanglement is one of the most intriguing and counterintuitive pheno...Quantum entanglement is one of the most intriguing and counterintuitive pheno...
Quantum entanglement is one of the most intriguing and counterintuitive pheno...
NishaJaiswal34
 
osama-quantum-computing and its uses and applications
osama-quantum-computing and its uses and applicationsosama-quantum-computing and its uses and applications
osama-quantum-computing and its uses and applications
Rachitdas2
 
quantum-computing.ppt
quantum-computing.pptquantum-computing.ppt
quantum-computing.ppt
MainakChakraborty40
 
osama-quantum-computing.ppt
osama-quantum-computing.pptosama-quantum-computing.ppt
osama-quantum-computing.ppt
SainadhDuppalapudi
 

Similar to Workshop (20)

Experimental one-way
Experimental one-wayExperimental one-way
Experimental one-way
 
Quantum Cryptography - Seminar report
Quantum Cryptography - Seminar reportQuantum Cryptography - Seminar report
Quantum Cryptography - Seminar report
 
1542 inner products
1542 inner products1542 inner products
1542 inner products
 
Introduction to Quantum Computing & Quantum Information Theory
Introduction to Quantum Computing & Quantum Information TheoryIntroduction to Quantum Computing & Quantum Information Theory
Introduction to Quantum Computing & Quantum Information Theory
 
What is the point of Boson sampling?
What is the point of Boson sampling?What is the point of Boson sampling?
What is the point of Boson sampling?
 
Basic Concepts of Entanglement Measures
Basic Concepts of Entanglement MeasuresBasic Concepts of Entanglement Measures
Basic Concepts of Entanglement Measures
 
Fss
FssFss
Fss
 
Root locus description in lucid way by ME IITB
Root locus description in lucid way by ME IITBRoot locus description in lucid way by ME IITB
Root locus description in lucid way by ME IITB
 
What happens if measure the electron spin twice?
What happens if measure the electron spin twice?What happens if measure the electron spin twice?
What happens if measure the electron spin twice?
 
Quantum computing
Quantum computingQuantum computing
Quantum computing
 
Lesson 29
Lesson 29Lesson 29
Lesson 29
 
AI Lesson 29
AI Lesson 29AI Lesson 29
AI Lesson 29
 
The Extraordinary World of Quantum Computing
The Extraordinary World of Quantum ComputingThe Extraordinary World of Quantum Computing
The Extraordinary World of Quantum Computing
 
Quantum Computing
Quantum ComputingQuantum Computing
Quantum Computing
 
Quantum entanglement is one of the most intriguing and counterintuitive pheno...
Quantum entanglement is one of the most intriguing and counterintuitive pheno...Quantum entanglement is one of the most intriguing and counterintuitive pheno...
Quantum entanglement is one of the most intriguing and counterintuitive pheno...
 
Ieee lecture
Ieee lectureIeee lecture
Ieee lecture
 
MZ2
MZ2MZ2
MZ2
 
osama-quantum-computing and its uses and applications
osama-quantum-computing and its uses and applicationsosama-quantum-computing and its uses and applications
osama-quantum-computing and its uses and applications
 
quantum-computing.ppt
quantum-computing.pptquantum-computing.ppt
quantum-computing.ppt
 
osama-quantum-computing.ppt
osama-quantum-computing.pptosama-quantum-computing.ppt
osama-quantum-computing.ppt
 

Workshop

  • 1. Consequences of the Physical CTT over experimental setups in quantum physics Antonio Acín1, Ariel Bendersky1, Gonzalo de La Torre1, Santiago Figueira2 y Gabriel Senno2 1ICFO-Institut de Ciencies Fotoniques, Barcelona, España 2Departamento de Computación, FCEN, Universidad de Buenos Aires Workshop on Information and Physics, Paris, 2015
  • 2. The Physical Church-Turing Thesis Physical CTT Every function calculable by a physical system is Turing-computable. 2
  • 3. Outline We present two consequences of the use of pseudorandomness instead of randomness in experimental quantum physics: A local model for Bell-like experiments in which Alice and Bob use computable (but unknown to an eavesdropper) inputs. 3
  • 4. Outline We present two consequences of the use of pseudorandomness instead of randomness in experimental quantum physics: A local model for Bell-like experiments in which Alice and Bob use computable (but unknown to an eavesdropper) inputs. When preparing mixed states by computably sampling pure states, the final preparations retains information on how it was mixed. 3
  • 6. Bell’s experiments A source (S) prepares and distributes two physical systems to distant observers Alice and Bob. Upon receiving their systems, each observer performs a measurement on it. The object of interest is p(a, b|x, y) the joint probability distribution of obtaining outcomes a and b when Alice and Bob choose measurements x and y. 5
  • 7. Locality In general, p(a, b|x, y) = p(a|x)p(b|y) Local explanation for distant correlations: past common cause, λ. p(a, b|x, y, λ) = p(a|x, λ)p(b|y, λ) λ may not be constant over all runs. Hence, in general, we say that a probability distribution is local if it can be written as: p(a, b|x, y) = Λ dλq(λ)p(a|x, λ)p(b|y, λ) Measurement independence assumption: q(λ|x, y) = q(λ) 6
  • 8. CHSH inequality Suppose x, y ∈ {0, 1} and a, b ∈ {−1, +1}, and consider axby = a,b ab p(a, b|x, y) Let S = a0b0 + a0b1 + a1b0 − a1b1 Theorem (Clauser, Horne, Shimony, Holt) Any local probability distribution has to satisfy, S ≤ 2 7
  • 9. Quantum theory is non-local Predictions for the outcomes of some distant measurements on entangled systems violate the previous inequality. 8
  • 10. Quantum theory is non-local Predictions for the outcomes of some distant measurements on entangled systems violate the previous inequality. For example, if the systems are prepared in the singlet state and Alice and Bob measure in the following spin directions, 8
  • 11. Quantum theory is non-local Predictions for the outcomes of some distant measurements on entangled systems violate the previous inequality. For example, if the systems are prepared in the singlet state and Alice and Bob measure in the following spin directions, We have that S = a0b0 + a0b1 + a1b0 − a1b1 S = √ 2 2 + √ 2 2 + √ 2 2 − − √ 2 2 = 2 √ 2 8
  • 12. Loopholes A loophole, in this context, is an experimental situation allowing for local devices to generate experimental data violating a Bell’s inequality. Examples: Detection loophole. Finite statistics loophole. Locality loophole. In this work we present, the computability loophole. 9
  • 13. The computational loophole It is convenient for what follows to rephrase the standard Bell scenario in cryptographic terms. In this approach, Alice and Bob get their systems from a non-trusted provider Eve. 10
  • 14. The computational loophole It is convenient for what follows to rephrase the standard Bell scenario in cryptographic terms. In this approach, Alice and Bob get their systems from a non-trusted provider Eve. We will allow Eve access to the inputs and outputs of previous rounds of the experiment. 10
  • 15. The computational loophole It is convenient for what follows to rephrase the standard Bell scenario in cryptographic terms. In this approach, Alice and Bob get their systems from a non-trusted provider Eve. We will allow Eve access to the inputs and outputs of previous rounds of the experiment. This memory scenario still allows to see non-locality (Barret et al PRA 66:042111, Pironio et al Nature 464(7291):1021-1024, Pironio et al PRA 87:012336). 10
  • 16. The computational loophole It is convenient for what follows to rephrase the standard Bell scenario in cryptographic terms. In this approach, Alice and Bob get their systems from a non-trusted provider Eve. We will allow Eve access to the inputs and outputs of previous rounds of the experiment. This memory scenario still allows to see non-locality (Barret et al PRA 66:042111, Pironio et al Nature 464(7291):1021-1024, Pironio et al PRA 87:012336). We show that, Theorem If Alice and Bob choose their measurements following an algorithm, Eve can prepare devices that locally violate CHSH inequality. 10
  • 17. Predicting computable functions Definition A class of total computable functions C is identifiable by next value (C ∈ NV) if there exists a computable function g (called a predictor for C) such that for every f ∈ C, (∃n0)(∀n ≥ n0) f(n) = g( f(0), . . . , f(n − 1) ) (1) 11
  • 18. Predicting computable functions Definition A class of total computable functions C is identifiable by next value (C ∈ NV) if there exists a computable function g (called a predictor for C) such that for every f ∈ C, (∃n0)(∀n ≥ n0) f(n) = g( f(0), . . . , f(n − 1) ) (1) Proposition (Putnam) The class of all total computable functions is not identifiable by next value. 11
  • 19. Predicting computable functions Definition A class of total computable functions C is identifiable by next value (C ∈ NV) if there exists a computable function g (called a predictor for C) such that for every f ∈ C, (∃n0)(∀n ≥ n0) f(n) = g( f(0), . . . , f(n − 1) ) (1) Proposition (Putnam) The class of all total computable functions is not identifiable by next value. Theorem (Adleman) A class of total computable functions is in NV if and only if it is a subclass of a complexity class (w.r.t. some complexity measure). 11
  • 20. Explanation of the loophole If Alice and Bob choose their measurements using computable functions, Eve doesn’t know the functions, but knows a time complexity class C that contains them and on every round, Eve receives the choices of Alice and Bob of previous rounds.. Then, Eve can prepare devices that 1 on every round, using the choices of previous rounds, execute a predictor for C and guess what the new choices will be and 2 output the necessary values for the statistic to violate CHSH inequality. 12
  • 21. Importance of the loophole Since every computable function belongs to some complexity class, Alice and Bob can never rule out the possibility of Eve predicting their functions. 13
  • 22. Importance of the loophole Since every computable function belongs to some complexity class, Alice and Bob can never rule out the possibility of Eve predicting their functions. Therefore no computable pseudo randomness criterion will suffice for a proper Bell inequality violation. 13
  • 23. Importance of the loophole Since every computable function belongs to some complexity class, Alice and Bob can never rule out the possibility of Eve predicting their functions. Therefore no computable pseudo randomness criterion will suffice for a proper Bell inequality violation. Other sources of randomness: 1 Quantum coins. Not desirable to assume a non-local theory, like quantum mechanics, in order to test non-locality. 2 Free will. Can humans generate non-computable sequences? 13
  • 25. Two kinds of mixed states Proper mixed states Describe ensembles of pure states of which we have classical uncertainty. Improper mixed states Describe systems which form part of bigger quantum system in a pure state. 15
  • 26. Case 1 R2D2 chooses from each box. The observer only knows that R2D2 will pick half times each state but not how he’ll pick each time. 16
  • 27. Case 1 R2D2 chooses from each box. The observer only knows that R2D2 will pick half times each state but not how he’ll pick each time. The state, as described by the observer is ρ = I 2 . 16
  • 28. Case 2 C3PO chooses from each box. The observer only knows that C3PO will pick half times each state but not how he’ll pick each time. 17
  • 29. Case 2 C3PO chooses from each box. The observer only knows that C3PO will pick half times each state but not how he’ll pick each time. The state, as described by the observer is ρ = I 2 . 17
  • 30. Undistiguishable Both situations seem to be undistinguishable. 18
  • 31. However... ... they are robots, so they can only choose in a computable manner. Any classical system used to choose only yields computable choices. 19
  • 32. Therefore We will be able to distinguish both situations. 20
  • 33. Distinguishing computable preparations Assumption We have a black box containing one of the two previous situations and we want to know which one it is. 21
  • 34. Distinguishing computable preparations Assumption We have a black box containing one of the two previous situations and we want to know which one it is. Procedure We measure every odd qubit on the basis of eigenstates of σX and every even qubit on the basis of eigenstates of σZ 21
  • 36. Distinguishing computable preparations Now what? We obtain two sequences. When we measure in the same basis as the preparation, the sequence obtained is computable. When we measure in the other basis, the sequence obtained is a fair coin. 23
  • 37. Distinguishing computable preparations Now what? We obtain two sequences. When we measure in the same basis as the preparation, the sequence obtained is computable. When we measure in the other basis, the sequence obtained is a fair coin. Let’s go classical Can we distinguish a computable sequence from one arising from a fair coin with high probability of success? We proved this to be true. 23
  • 38. The distinguishing protocol Idea Almost all sequences arising from independent tosses of fair coins have incompressible prefixes. 24
  • 39. The distinguishing protocol Input: k ∈ N and X, Z ∈ {0, 1}ω, two bit sequences with the promise that one of them is computable. Output: ‘X’ or ‘Z’ as the candidate for being computable; wrong answer with probability bounded by O(2−k). for t = 0, 1, 2 . . . do for p = 0, . . . , t do if Ut(p) = X k|p| then output ‘X’ and halt if Ut(p) = Z k|p| then output ‘Z’ and halt 25
  • 40. Probability of misrecognition Probability of error Perror ≤ >0 2 2k = 2−(k−1) 1 − 2−(k−1) which goes to 0 as k goes to ∞. We pick a k such that the error is lower than what we want, and then we run the recognition algorithm. 26
  • 41. Some subtelties Our algorithm, although it runs in finite time, is infeasible. 27
  • 42. Some subtelties Our algorithm, although it runs in finite time, is infeasible. Still, the state has the information on how it was mixed. This is surprising from a fundamental point of view. 27
  • 43. Some subtelties Our algorithm, although it runs in finite time, is infeasible. Still, the state has the information on how it was mixed. This is surprising from a fundamental point of view. A slight variation on the algorithm makes it noise tolerant for noise rates up to 0,21. 27
  • 44. Assuming the Physical CTT Physical CTT Every function calculable by a physical system is Turing-computable. So, deterministic physical processes, if we accept the impossibility of preparing non-computable initial conditions, won’t do it. 28
  • 45. Assuming the Physical CTT Physical CTT Every function calculable by a physical system is Turing-computable. So, deterministic physical processes, if we accept the impossibility of preparing non-computable initial conditions, won’t do it. Consequences We are left with quantum randomness. This means that: Only kind of mixed state is those being part of larger system in a pure state. Need quantum randomness to test quantum non-locality. 28