Vous êtes sur la page 1sur 42

Trading resources in quantum Shannon theory

Mark M. Wilde

Hearne Institute for Theoretical Physics, Department of Physics and Astronomy, Center for Computation and Technology, Louisiana State University, Baton Rouge, Louisiana, USA

mwilde@lsu.edu

Based on arXiv:1709.01111, 1605.04922, 1206.4886, 1105.0119, 1004.0458, 1001.1732, 0901.3038, 0811.4227 (with Bradler, Guha, Hayden, Hsieh, Leditzky, Qi, Touchette) and Chapter 25 of arXiv:1106.1445

CQuIC Seminar, University of New Mexico, January 31, 2019

Question: What are the net rates at which a sender and receiver can generate classical communication, quantum communication, and entanglement by using a quantum channel many times?

Many special cases are known, such as the classical capacity theorem [Hol98, SW97], quantum capacity theorem [Sch96, SN96, BNS98, BKN00, Llo97, Sho02, Dev05], and the entanglement-assisted classical capacity theorem [BSST02]

A priori, this question might seem challenging, but there is a surprisingly simple answer for several channels of interest:

Just combine a single protocol with teleportation, super-dense coding, and entanglement distribution

Quantum channels

Quantum channels represent noisy physical evolutions of quantum systems. Mathematically, a quantum channel is a linear, completely positive, trace-preserving map, thus taking an input quantum state to an output quantum state. Quantum channels are usually denoted by N ,M, P, etc.

Isometric extensions of quantum channels

Every quantum channel has an isometric extension: There exists an isometry taking an input density operator to the tensor-product Hilbert space of output and environment. Channel is realized by applying isometry and discarding environment

Quantum measurements

A quantum measurement is a special type of quantum channel with quantum input and classical output

Resources [Ben04, DHW04, DHW08]

Let [c c] denote a noiseless classical bit channel from Alice

(sender) to Bob (receiver), which performs the following mapping on

a qubit density operator

ρ = ρ 00

ρ 10

ρ 11 ρ 00

ρ 01

0

11

0

ρ

Let [q q] denote a noiseless quantum bit channel from Alice to

Bob, which perfectly preserves a qubit density operator.

Let [qq] denote a noiseless ebit shared between Alice and Bob, which

is a maximally entangled state |Φ + AB = (|00 AB + |11 AB )/ 2.

Entanglement distribution, super-dense coding, and teleportation are

non-trivial protocols for combining these resources

|0〉 H A |0〉 id A’ A’→B B
|0〉
H
A
|0〉
id
A’
A’→B
B

Alice performs local operations (the Hadamard and CNOT) and

consumes one use of a noiseless qubit channel to generate one

noiseless ebit |Φ + AB shared with Bob.

Resource inequality: [q q] [qq]

Conditional Operations

x1 x2

+

AB

Channel Qubit X Z x1 x2
Channel Qubit
X
Z
x1 x2

Bell Measurement

Alice and Bob share an ebit. Alice would like to transmit two classical

bits x 1 x 2 to Bob. She performs a Pauli rotation conditioned on x 1 x 2

and sends her share of the ebit over a noiseless qubit channel. Bob

then performs a Bell measurement to get x 1 x 2 .

Resource inequality: [q q] + [qq] 2[c c]

Bell Measurement Two Channels Classical |ψ〉 A’ |Ф + 〉 AB |ψ〉 X Z B
Bell Measurement
Two Channels Classical
|ψ〉
A’
+
AB
|ψ〉
X
Z
B

Conditional Operations

Alice would like to transmit an arbitrary quantum state |ψ A to Bob.

Alice and Bob share an ebit before the protocol begins. Alice can

“teleport” her quantum state to Bob by consuming the entanglement

and two uses of a noiseless classical bit channel.

Resource inequality: 2[c c] + [qq] [q q]

Think of each protocol as a rate triple (C, Q, E)

Entanglement distribution is (0, 1, 1)

Super-dense coding is (2, 1, 1)

Teleportation is (2, 1, 1)

All achievable rate triples are then given by

{(C, Q, E) = α(2, 1, 1) + β(2, 1, 1) + γ(0, 1, 1) : α, β, γ 0}

Writing as a matrix equation, inverting, and applying constraints

α, β, γ 0 gives the following achievable rate region:

C + Q

+ E

0,

Q

+ E

0,

C + 2Q

0.

1

0.5

E

0

-0.5

-1

(0,-1,1) ED (0,0,0) SD TP (-2,1,-1) (2,-1,-1) -1 -2 0 0 Q C 1 2
(0,-1,1)
ED
(0,0,0)
SD
TP
(-2,1,-1)
(2,-1,-1)
-1
-2
0
0
Q
C
1
2

The unit resource

capacity region is C + Q + E 0, Q + E 0,

C + 2Q 0 and is provably optimal.

Main question: What net rates of classical communication, quantum

communication, and entanglement generation can we achieve by

using a quantum channel N many times?

That is, what are the rates C out , Q out , E out , C in , Q in , E in 0

achievable in the following resource inequality?

N + C in [c c] + Q in [q q] + E in [qq]

C out [c c] + Q out [q q]

+ E out [qq]

The union of all achievable rate triples

(C out C in , Q out Q in , E out E in ) is called the quantum dynamic

capacity region.

Reference t t f R SA A1 N M E A’ A’ N B B Alice
Reference
t
t f
R
SA
A1
N
M E
A’ A’
N
B B
Alice
TA
N
B
A’ A2
L
TB
B B
Bob
D
S B1 B
B
M

Figure: The most general protocol for generating classical communication, quantum communication, and entanglement with the help of the same respective resources and many uses of a quantum channel.

What strategy should we use to communicate classical and quantum information simultaneously, along with the assistance of entanglement?

One idea is time-sharing, but it is possible to do better ...

The optimal rates are expressed in terms of entropies, which we

review briefly

Given density operator σ, quantum entropy defined as

H(σ) = Tr{σ log σ}.

Given a bipartite density operator ρ AB , the quantum mutual

information is defined as

I(A; B) ρ = H(A) ρ + H(B) ρ H(AB) ρ

The coherent information I(A B) ρ is defined as

I(A B) ρ = H(B) ρ H(AB) ρ

Given a tripartite density operator ρ ABC , the conditional mutual

information is defined as

I(A; B|C) ρ = H(AC) ρ + H(BC) ρ H(C) ρ H(ABC) ρ

(1)

Define the state-dependent region C CQE (N , σ) as the set of all rates

C, Q, and E, such that

C

+ 2Q

I(AX; B) σ ,

Q

+ E

I(A BX) σ ,

C + Q

+ E

I(X; B) σ + I(A BX) σ .

The above entropic quantities are with respect to a classical–quantum

state σ XAB , where

σ XAB

p X (x)|x x| X ⊗ N A B (φ x AA ),

x

and the states φ AA

x

are pure.

(1)

Define C CQE (N ) as the union of the state-dependent regions

(1)

C CQE (N , σ):

C CQE (N ) CQE (N , σ).

C

(1)

(1)

σ

Then the quantum dynamic capacity region C CQE (N ) of a channel N

is equal to the following expression:

C CQE (N ) =

k=1

1

k C

(1) CQE (N ⊗k ),
(1)
CQE (N ⊗k ),

where the overbar indicates the closure of a set.

It is implicit that one should consider states on A k instead of A

when taking the regularization.

Take the channel to be the qubit dephasing channel

N (ρ) = (1 p)ρ + pZρZ with dephasing parameter p = 0.2.

Take the input state as

σ XAA

1

2 (|0 0| X φ AA + |1 1| X φ

0

1

AA ),

where

φ 0 AA 1/4|00 AA + 3/4|11 AA ,

φ 1 AA 3/4|00 AA + 1/4|11 AA .

The state σ XAB resulting from the channel is N A B (σ XAA )

0.9 EAC 0.8 0.7 0.6 I 0.5 0.4 0.3 III 0.2 CEF CEF-TP 0.1 EAQ II
0.9
EAC
0.8
0.7
0.6
I
0.5
0.4
0.3
III
0.2
CEF
CEF-TP
0.1
EAQ
II
CEF-SD-ED
Quantum communication rate
0
0
LSD
0.1
CEQ
0.2
0.3
0.4
0
0.5
0.6
0.5
0.7
1
0.8
1.5
Classical communication rate
Entanglement consumption rate

(1)

Figure: An example of the state-dependent achievable region C CQE (N , σ) corresponding to a state σ XABE that arises from a qubit dephasing channel with dephasing parameter p = 0.2. The figure depicts the octant corresponding to the consumption of entanglement and the generation of classical and quantum communication.

Entanglement-assisted classical and quantum communication There is a protocol that implements the following resource inequality: 1
Entanglement-assisted classical and quantum communication
There is a protocol that implements the following resource inequality:
1
1
N +
2 I(A; E|X) ρ [qq] ≥
2 I(A; B|X) ρ [q → q] + I(X; B) ρ [c →
c]
where ρ XABE is a state of the following form:
N
p X (x)|x x| X
⊗ U
ρ XABE ≡
A →BE (ϕ x
AA ),
x
N
the states ϕ x AA are pure, and U →BE is an isometric extension of the
A
channel N A →B .
Combine this with the unit protocols of teleportation, super-dense
coding, and entanglement distribution

Combining the protocols gives the following set of achievable rates:

Q = 1

C

E

0

1

2

1

1

2

1

1

β +

α

γ

I(X; B) σ

1

2 I(A; B|X) σ

1 2 I(A; E|X) σ

,

where α, β, γ 0.

Inverting the matrix equation, applying the constraints α, β, γ 0,

and using entropy identities gives the following region:

 

C

+ 2Q

I(AX; B) σ ,

Q

+ E

I(A BX) σ ,

C

+ Q

+ E

I(X; B) σ + I(A BX) σ ,

which establishes the achievability part.

How to achieve the following resource inequality?

1

N + 2 I(A; E|X) ρ [qq]

1

2 I(A; B|X) ρ [q q] + I(X; B) ρ [c c]

Tools for achievability part [Wil15, Chapter 25]

HSW classical capacity theorem [Hol98, SW97]

Entanglement-assisted classical capacity theorem [BSST02] (see also

[HDW08])

Modification of a classical trick called “superposition coding” [Sho04]

Another trick called coherent communication [Har04, DHW08]

Consider the most general protocol:

Reference t t f R A1 SA A’ N B M E A’ Alice N B
Reference
t
t f
R
A1
SA
A’
N
B
M E
A’
Alice
N
B
TA
N
B
A’ A2
L
TB
B B
Bob
D
S B1 B
B
M

Make use of uniform continuity bounds for entropy, quantum data

processing, and dimension bounds for information quantities

Let w (w C ,w Q ,w E ) R 3 be a weight vector,

R (C, Q, E) a rate

vector, and E ≡ {p X (x), φ AA x } an ensemble.

Can phrase the task of computing the boundary of the single-copy

capacity region as an optimization problem:

 

 

P (w

) sup w · R

 

R,E

subject to

C + 2Q

I(AX; B) σ ,

 

Q

+ E

I(A BX) σ ,

C + Q

+ E

I(X; B) σ + I(A BX) σ ,

where the optimization is with respect to all rate vectors

R and

ensembles E, with σ XAB a state of the previously given form.

By linear programming duality, if P (w ) < , then the optimization

problem is equivalent to computing the quantum dynamic capacity

formula, defined as

D

λ

σ

λ 1 , λ 2 , λ 3 0.

(N ) max λ 1 I(AX; B) σ +λ 2 I(A BX) σ +λ 3 [I(X; B) σ + I(A BX) σ ] ,

where σ XAB is a state of the previously given form and

λ (λ 1 , λ 2 , λ 3 ) is a vector of Lagrange multipliers such that

Suppose for a given channel N that D

λ

(N n ) = nD

λ

(N )

n 1

and

λ 0. Then the computation of the boundary simplifies

significantly. This happens for a number of important channels.

Erasure channel is defined as follows:

N ε (ρ) = (1 ε) ρ + ε|e e|,

where ρ is a d-dimensional input state, |e is an erasure flag state

orthogonal to all inputs (so that the output space has dimension

d + 1), and ε [0, 1] is the erasure probability.

Let N ε be a quantum erasure channel with ε [0, 1/2].

Then the

quantum dynamic capacity region C CQE (N ε ) is equal to the union of

the following regions, obtained by varying λ [0, 1]:

C

+ 2Q

(1

ε) (1 +

λ) log d,

Q

+ E

(1

2ε) λ log d,

C + Q

+ E

(1 ε ελ) log d.

1 SD TP 0. 8 SD 0. 6 0. 4 0. 2 0 −0.2 −0.4 −0.6
1
SD
TP
0.
8
SD
0.
6
0.
4
0.
2
0
−0.2
−0.4
−0.6
ED
CEF curve
ED
−0.8 −1
−0.5
0
−1
−1.5
0. 5
0
−0.5
1
1. 5
1
0. 5
2
Quantum communication rate
1. 5
2. 5
Classical communication rate
Entanglement consumption rate

Figure: The quantum dynamic capacity region for the (qubit) quantum erasure channel with ε = 1/4. The plot demonstrates that time-sharing is optimal.

The dynamic capacity region C CQE (∆ p ) of a dephasing channel with

dephasing parameter p [0, 1] is the set of all C, Q, and E such that

C + 2Q

1 + h 2 (ν) h 2 (γ(ν, p)),

Q

+ E

h 2 (ν) h 2 (γ(ν, p)),

C + Q

+ E

1 h 2 (γ(ν, p)),

where ν [0, 1/2], h 2 is the binary entropy function, and

γ(ν, p)

1

2

+

1 2 1 − 16 ·
1
2 1 − 16 ·

p

2

1

p

2

ν(1 ν).

1.5 1 SD TP SD 0.5 0 −0.5 ED ED CEF curve −1 −1 0 −2
1.5
1
SD
TP
SD
0.5
0
−0.5
ED
ED
CEF curve
−1 −1
0
−2
−1
1
0
1
Quantum communication rate
2
2
3
Classical communication rate
Entanglement consumption rate

Figure: A plot of the dynamic capacity region for a qubit dephasing channel with dephasing parameter p = 0.2. Slight improvement over time-sharing.

Pure-loss channel is defined from the following input-output relation:

aˆ

b = η aˆ+ 1 η eˆ,

ˆ

where aˆ is the input annihilation operator for the sender, eˆ is the

input annihilation operator for the environment, and η [0, 1] is the

transmissivity of the channel.

Pure-loss channel is defined from the following input-output relation: a ˆ → b = η a

Place a photon number constraint on the input mode to the channel,

such that the mean number of photons at the input cannot be greater

than N S [0, ).

Build trade-off codes from an ensemble of the following form:

p (1λ)N S (α), D A (α)|ψ TMS (λ) AA ,

where α C,

p (1λ)N S (α)

  • 1 λ) N S exp − |α| 2 / [(1 λ) N S ] ,

π (1

λ [0, 1] is a photon-number-sharing parameter, D A (α) is a

“displacement” unitary operator acting on system A , and |ψ TMS (λ) AA is

a “two-mode squeezed” (TMS) state:

|ψ TMS (λ) AA

n=0

[λN S ] n

1] n+1 |n A |n A ,

[λN S +

The quantum dynamic capacity region for a pure-loss bosonic channel with

transmissivity η 1/2 is the union of regions of the form:

C + 2Q

g(λN S ) + g(ηN S ) g((1 η) λN S ),

Q

+ E

g(ηλN S ) g((1 η) λN S ),

C + Q

+ E

g(ηN S ) g((1 η) λN S ),

where λ [0, 1] is a photon-number-sharing parameter and g(N) is the

entropy of a thermal state with mean photon number N.

This holds provided that an unsolved minimum-output entropy conjecture

is true. The region is still achievable if η < 1/2.

Q (qubits / channel use)

E (ebits / channel use)

1.6

1.2

0.8

0.4

0

η = 3/4 N = 200 S
η = 3/4
N
= 200
S

0

2

4

6

8

C

(cbits / channel use)

(a)

8

6

4

2

0

Time−sharing Trade−off coding
Time−sharing
Trade−off coding

9

10

11

C

(cbits / channel use)

(b)

Figure: Suppose channel transmits on average 3/4 of the photons to the receiver, while losing the other 1/4 en route. Take mean photon budget of about 200 photons per channel use at the transmitter. (a) classical–quantum trade-off, (b) classical comm. with rate-limited entanglement consumption. Big gains over time-sharing.

Amplifier channel is defined from the following input-output relation:

aˆ

b = G

ˆ

aˆ+ G 1 eˆ ,

where aˆ is the input annihilation operator for the sender, eˆ is the

input annihilation operator for the environment, and G [1, ) is the

gain of the channel. The channel is quantum-limited if the

environment is prepared in a vacuum state.

Place a photon number constraint on the input mode to the channel,

such that the mean number of photons at the input cannot be greater

than N S [0, ).

The quantum dynamic capacity region for a quantum-limited amplifier

channel with gain G 1 is the union of regions of the form:

C + 2Q

¯

¯

g(λN S ) + g(GN S + G) g( G[λN S + 1]),

 

¯

¯

Q

+ E

g(GλN S + G) g( G[λN S + 1]),

 

¯

¯

C + Q

+ E

g(GN S + G) g( G[λN S + 1]),

where

¯

G = G 1 and λ [0, 1] is a photon-number-sharing parameter

and g(N) is the entropy of a thermal state with mean photon number N.

Bosonic channel with a complete triple trade-off solution!!!!

Figure: Suppose channel amplifies with gain G = 2 the photons being transmitted to the receiver.

Figure: Suppose channel amplifies with gain G = 2 the photons being transmitted to the receiver. Take mean photon budget of about 200 photons per channel use at the transmitter. (a) classical–quantum trade-off, (b) classical comm. with rate-limited entanglement consumption. Big gains over time-sharing.

Trade-off between secret key, private classical communication, and

public classical communication [WH12a]

Bounds for triple trade-off capacities using operator space and

complex interpolation theory [GJL18]

Many examples of superadditive effects for the trade-off capacities

[ZZS17, ZZHS17]. Just because one capacity is single-letter, it

doesn’t mean that double- or triple-resource trade-off is single-letter.

If a channel is approximately Hadamard, then the triple-resource

capacity region is approximately single-letter [LKDW18].

Summary

The quantum dynamic capacity theorem characterizes the net rates at

which a sender and a receiver can generate classical communication,

quantum communication, and entanglement by using a quantum

channel many times

The region simplifies for several channels of interest

Open questions

Is there a simple characterization for distillation tasks? For progress,

see [HW10]

Can we sharpen the theorem? Strong converse bounds, error

exponents, finite-length, second-order, etc.

[BBC + 93]

Charles H. Bennett, Gilles Brassard, Claude Cr´epeau, Richard Jozsa, Asher Peres, and William K. Wootters. Teleporting an unknown quantum state via dual classical and Einstein-Podolsky-Rosen channels. Physical Review Letters, 70(13):1895–1899, March 1993.

[BDH + 14]

Charles H. Bennett, Igor Devetak, Aram W. Harrow, Peter W. Shor, and Andreas Winter. The quantum reverse Shannon theorem and resource tradeoffs for simulating quantum channels. IEEE Transactions on Information Theory, 60(5):2926–2959, May 2014. arXiv:0912.5537.

[Ben04]

Charles H. Bennett. A resource-based view of quantum information. Quantum Information and Computation, 4:460–466, December 2004.

[BKN00]

Howard Barnum, Emanuel Knill, and Michael A. Nielsen. On quantum fidelities and channel capacities. IEEE Transactions on Information Theory, 46(4):1317–1329, July 2000. arXiv:quant-ph/9809010.

[BNS98]

Howard Barnum, M. A. Nielsen, and Benjamin Schumacher. Information transmission through a noisy quantum channel. Physical Review A, 57(6):4153–4175, June 1998.

[BSST02]

Charles H. Bennett, Peter W. Shor, John A. Smolin, and Ashish V. Thapliyal. Entanglement-assisted capacity of a quantum channel and the reverse Shannon theorem. IEEE Transactions on Information Theory, 48(10):2637–2655, October 2002. arXiv:quant-ph/0106052.

[BW92]

Charles H. Bennett and Stephen J. Wiesner. Communication via one- and two-particle operators on Einstein-Podolsky-Rosen states. Physical Review Letters, 69(20):2881–2884, November 1992.

[Dev05]

Igor Devetak. The private classical capacity and quantum capacity of a quantum channel. IEEE Transactions on Information Theory, 51(1):44–55, January 2005. arXiv:quant-ph/0304127.

[DHW04]

Igor Devetak, Aram W. Harrow, and Andreas Winter. A family of quantum protocols. Physical Review Letters, 93(23):239503, December 2004.

arXiv:quant-ph/0308044.

[DHW08]

Igor Devetak, Aram W. Harrow, and Andreas Winter. A resource framework for quantum Shannon theory. IEEE Transactions on Information Theory, 54(10):4587–4618, October 2008. arXiv:quant-ph/0512015.

[GJL18]

Li Gao, Marius Junge, and Nicholas LaRacuente. Capacity bounds via operator space methods. Journal of Mathematical Physics, 59(12):122202,

  • 2018. arXiv:1509.07294.

[Har04]

Aram Harrow. Coherent communication of classical messages. Physical Review Letters, 92(9):097902, March 2004. arXiv:quant-ph/0307091.

[HDW08]

Min-Hsiu Hsieh, Igor Devetak, and Andreas Winter. Entanglement-assisted capacity of quantum multiple-access channels. IEEE Transactions on Information Theory, 54(7):3078–3090, July 2008. arXiv:quant-ph/0511228.

[Hol98]

Alexander S. Holevo. The capacity of the quantum channel with general signal states. IEEE Transactions on Information Theory, 44(1):269–273, January 1998. arXiv:quant-ph/9611023.

[HW10]

Min-Hsiu Hsieh and Mark M. Wilde. Trading classical communication, quantum communication, and entanglement in quantum Shannon theory. IEEE Transactions on Information Theory, 56(9):4705–4730, September

  • 2010. arXiv:0901.3038.

[LKDW18] Felix Leditzky, Eneet Kaur, Nilanjana Datta, and Mark M. Wilde.

Approaches for approximate additivity of the Holevo information of quantum channels. Physical Review A, 97(1):012332, January 2018.

arXiv:1709.01111.

[Llo97]

Seth Lloyd. Capacity of the noisy quantum channel. Physical Review A, 55(3):1613–1622, March 1997. arXiv:quant-ph/9604015.

[QW17]

Haoyu Qi and Mark M. Wilde. Capacities of quantum amplifier channels. Physical Review A, 95(1):012339, January 2017. arXiv:1605.04922.

[Sch96]

Benjamin Schumacher. Sending entanglement through noisy quantum channels. Physical Review A, 54(4):2614–2628, October 1996.

[Sho02]

Peter W. Shor. The quantum channel capacity and coherent information. In Lecture Notes, MSRI Workshop on Quantum Computation, 2002.

[Sho04]

Peter W. Shor. Quantum Information, Statistics, Probability (Dedicated to A. S. Holevo on the occasion of his 60th Birthday): The classical capacity achievable by a quantum channel assisted by limited entanglement. Rinton Press, Inc., 2004. arXiv:quant-ph/0402129.

[SN96]

Benjamin Schumacher and Michael A. Nielsen. Quantum data processing and error correction. Physical Review A, 54(4):2629–2635, October 1996.

arXiv:quant-ph/9604022.

[SW97]

Benjamin Schumacher and Michael D. Westmoreland. Sending classical information via noisy quantum channels. Physical Review A, 56(1):131–138, July 1997.

[WH12a]

Mark M. Wilde and Min-Hsiu Hsieh. Public and private resource trade-offs for a quantum channel. Quantum Information Processing, 11(6):1465–1501, December 2012. arXiv:1005.3818.

[WH12b]

Mark M. Wilde and Min-Hsiu Hsieh. The quantum dynamic capacity formula of a quantum channel. Quantum Information Processing, 11(6):1431–1463, December 2012. arXiv:1004.0458.

[WHG12]

Mark M. Wilde, Patrick Hayden, and Saikat Guha. Information trade-offs for optical quantum communication. Physical Review Letters, 108(14):140501, April 2012. arXiv:1105.0119.

[Wil15]

Mark M. Wilde. From Classical to Quantum Shannon Theory. 2015. arXiv:1106.1445 (preprint for a 2nd edition).

[ZZHS17]

Elton Yechao Zhu, Quntao Zhuang, Min-Hsiu Hsieh, and Peter W. Shor. Superadditivity in trade-off capacities of quantum channels. August 2017.

arXiv:1708.04314.

[ZZS17]

Elton Yechao Zhu, Quntao Zhuang, and Peter W. Shor. Superadditivity of the classical capacity with limited entanglement assistance. Physical Review Letters, 119(4):040503, July 2017. arXiv:1704.06955.