quantum computer meets blackhole and quantum...
TRANSCRIPT
Quantum Computer Meets
Blackhole and Quantum Chaos
Workshop on OTO correlators
藤井 啓祐 東京大学 大学院 工学系研究科付属 光量子科学研究センター
JST さきがけ
量子情報とOTO
2008年4月~2011年3月 京都大学 工学研究科 原子核工学専攻 博士課程 日本学術振興会特別研究員 DC1
2011年4月~2013年3月 大阪大学 基礎工学研究科 物質創成専攻 井元研 特任研究員
2013年4月~2015年3月 京都大学 白眉センター (情報学研究科 通信情報システム専攻 岩間研 特定助教)
2015年4月~2016年3月 京都大学 白眉センター (理学研究科 物理学宇宙物理学専攻 高橋研 特定助教)
2016年4月~ 東京大学 工学系研究科 光量子科学研究センター 小芦研 助教 (物理工学専攻 兼担)
2016年10月~ JSTさきがけ「量子の状態制御と機能化」さきがけ研究員兼任
自己紹介:藤井 啓祐工学
基礎工学
情報
物理
物理工学
Outline• Quantum information perspective of random unitary:
blackhole as a mirror
• Random unitary and OTOC
• Quantum computational supremacy benchmark on a quantum computer
• How to characterize random unitary operation: quantum chaos
• Toward random unitary with a translation invariant system using universality (simulate random unitary on a quantum simulator)
All these are about complex unitary operations, which is now becoming experimentally accessible!
Blackhole as a mirrorHayden-Preskill ’07
Alice’s information
Initial stateof black hole
Scrambling dynamics
=random unitary
Bob (observer with an unlimited computational power)
Hawking radiation
Where is Alice’s information?
Classical toy modelHayden-Preskill ’07
Alice’s information
Initial stateof blackhole
random permutationover 2n n-bit
strings Hawking radiation
00101011
00000000(n-k bit string)
(k bit string)
1101…(s bit string)
Bob wants to decode Alice’s information from HR. How many bits in HR is required to do this task?
→Channel capacity
Classical toy modelHayden-Preskill ’07
Alice’s information
Initial stateof blackhole
random permutation
over 2n n-bit strings HR
00101011
00000000(n-k bit string)
(k bit string)
1101…(s bit string)
source of information
encoding to an error
correction code
channel output
erased
eraser channel
The random permutation is an encoding to achieve the optimal channel capacity! s=k is enough to decode!
Classical toy modelHayden-Preskill ’07
source of information
00000 000…0 00001 000…0 00010 000…0
11111 000…0
…
Alice
{2k
Bob has a table of the random permutation,and decode what’s Alice’s information from HR.
10110 110…1 01001 101…0 01010 100…0
11010 010…1…
HR erased
{s bits
pfail = 2k2�s
probability to have the same bit strings in HR for different Alice’s message.
Total number of Alice’s bit strings
If s=k+c, pfail is sufficiently small.
encoding to an error
correction code
random permutation
over 2n bit strings
Entanglement assisted quantum eraser channel capacity
Alice’s information
Initial stateof blackhole R
maximally entangled state
BobE
B’
reference
maximally entangled state
NScrambling dynamics
=random unitary
V B
maximally entangled state?
Alice
black hole R
Bob
MES
E
B’
N
VB
MES
⇢NB(VB) = (IN ⌦ VB)⇢NB(IN ⌦ V †B)
Entanglement assisted quantum eraser channel capacity
⇢NB0(VB) = TrR[⇢NB(VB)]
⇢N (VB) = TrB0 [⇢NB0(VB)]
F (VB) = h�|⇢M̃N |�i = 1� k⇢NB0(VB)� ⇢N (VB)⌦ I 0Bk1
s=k+c qubit is enough to decode Alice’s info.
Decoupling theorem [Hayden et al ‘07]:Z
dVBk⇢NB0(VB)� ⇢N (VB)⌦ IB0k1 |NB||R|2 Tr[⇢2NB ] = 22(k�s)
kAk1 = Tr[pAA†]
2
⇢NB = TrE [⇢NBE ]
Outline• Quantum information perspective of random unitary:
blackhole as a mirror
• Random unitary and OTOC
• Quantum computational supremacy benchmark on a quantum computer
• How to characterize random unitary operation: quantum chaos
• Toward random unitary with a translation invariant system using universality (simulate random unitary on a quantum simulator)
All these are about complex unitary operations, which is now becoming experimentally accessible!
Random unitaryExact Haar random unitary: not efficient, exponential time is required
Approximated t-design (Dankert et al., ‘06):X
i
piU⌦ti ⇢(U †
i )⌦t '
Z
Haar
U⌦t⇢(U †)⌦tdU
(using operator 1-norm or diamondnorm as a map)
decoupling theorem & approximated 2-design:X
i
pik⇢NB0(Vi)� ⇢N (Vi)⌦ I 0Bk21 22(k�s) + ✏
approximated 2-design is efficiently achievable(Dankert et al., ’06):
O(log(n) log(1/✏))time O(
pn log(n) log(1/✏))
localitycircuit depth circuit depth
Random unitary & OTOCRandom unitary → delocalization of quantum information:
2-norm of commutator
k[A(0), B(t)]k22kWk2 = (Tr[WW †])1/2
OTOC:
[A,B] = 0 $ hABABi = 1
{A,B} = 0 $ hABABi = �1
hABABi = 1
dTr[ABAB]
OTOC & 2-norm of commutator:
( A2 = B2 = I is assumed, e.g. Pauli operators)
k[A,B]k22 = 2Tr[I �ABAB]= 2d(1� hABABi)
2-design & 4-point OTOCScrambling (unitary 2-design) implies decay of OTOC
Unitary 2-design and Haar random unitary both result in the same 4-point OTOC!
4-point (2k-point) OTOC is a good witness (related to frame potential) of random unitary of 2-design (k-design).(Roberts-Yoshida ’16)
Wswap ⌘X
ij
|ii|jihj|hi|
Tr[AB(t)AB(t)⇢]
= Tr[(A⌦A)(Ut ⌦ Ut)(B ⌦B)(Ut ⌦ Ut)†Wswap]
Tr[AB(t)AB(t)]
Outline• Quantum information perspective of random unitary:
blackhole as a mirror
• Random unitary and OTOC
• Quantum computational supremacy benchmark on a quantum computer
• How to characterize random unitary operation: quantum chaos
• Toward random unitary with a translation invariant system using universality (simulate random unitary on a quantum simulator)
All these are about complex unitary operations, which is now becoming experimentally accessible!
Quantum “computational” supremacy
Google’s Milestone Chip Can Achieve Quantum Supremacy By The End Of 2017https://fossbytes.com/google-quantum-chip-50-qubit/
IBM’s First Commercial “Universal Quantum Computer” Can Be The New Future Of AIhttps://fossbytes.com/ibm-q-quantum-computing-50-qubit-computer/
22qubits-> 50qubits 16,17qubits
50qubits, d=250=1015 → O(100 peta bit)
What is quantum computer?“Nature isn't classical, dammit, and if you want to make a simulation of Nature, you'd better make it quantum mechanical, and by golly it's a wonderful problem, because it doesn't look so easy.” R. Feynman, Simulating Physics with Computers, Int. J. Theor. Phys. 21, 467 (1982).
→quantum computer, quantum simulator
W. S. Bakr et al., Nature 462, 74 (2009)
Kelly et al., Nature 519, 66 (2015)
Is there a quantum machine that can simulate physics universally?
What is quantum computer?
・・・
time evolution
unitary operator (dim = 2n)
measurement・・・ single-qubit
unitary operator
2qubit unitary operator
quantum circuit
|0i|0i|0i
|0i
initial state
・・・{N
qubi
t
qubit
Universal quantum computer|0i|0i|0i
|0i
initial state time evolution 測定
・・・
・・・
unitary operator
・・・{N qu
bit
Is there a quantum machine that can simulate physics universally?
H =1p2
✓1 11 �1
◆
Hadamard
e�i(⇡/8)Z =
✓e�i⇡/8 0
0 ei⇡/8
◆
π/8 (T gate)
UCNOT = |0ih0|⌦ I + |1ih1|⌦X
=
0
BB@
1 0 0 00 1 0 00 0 0 10 0 1 0
1
CCA
CNOT
Universal gate set
universal quantum computer=can implement an arbitrary unitary operator
R. P. Feynman, Quantum Mechanical Computers, Optics News (1985)
Outline• Quantum information perspective of random unitary:
blackhole as a mirror
• Random unitary and OTOC
• Quantum computational supremacy benchmark on a quantum computer
• How to characterize random unitary operation: quantum chaos
• Toward random unitary with a translation invariant system using universality (simulate random unitary on a quantum simulator)
All these are about complex unitary operations, which is now becoming experimentally accessible!
How to characterize the quantum machine
Boixo et al., arXiv:1608.00263
random unitary circuit
APS march meeting 20179 qubit, depth 10, error per gate 0.3%,
total fidelity 98%
How can we characterize complex many-body quantum dynamics universally?
Neutron scattering and quantum chaos (random unitary)
resonance level spacing: Wigner-Dyson distributionresonance strength: Porter-Thomas distribution
“Handbook of Nuclear Engineering - Chapter3”Evaluated Nuclear Data
P. Oblozinsky, M. Herman and S.F. Mughabghab
The scattering strength is proportional to which diagonalizes the Hamiltonian H.
|Uij |2
How to characterize the quantum machine
Boixo et al., arXiv:1608.00263
random unitary circuit
}Porter-Thomas distribution
| i =2nX
i=1
ci|ii
p(c1, c2, ..., c2n) / �
1�
X
i
|ci|2!
p(|c|2) / e�|c|2/ ¯|c|2
randomly chosen
Boixo et al., arXiv:1608.00263
How to characterize the quantum machine
probability × dim
prob
abilit
y de
nsity
noise less
uniform distribution PT entropy
42qubit
36qubit
PT entropy
How are OTOC decay and convergence of PT distribution related?
Outline• Quantum information perspective of random unitary:
blackhole as a mirror
• Random unitary and OTOC
• Quantum computational supremacy benchmark on a quantum computer
• How to characterize random unitary operation: quantum chaos
• Toward random unitary with a translation invariant system using universality (simulate random unitary on a quantum simulator)
All these are about complex unitary operations, which is now becoming experimentally accessible!
Universal quantum computer
|0i|0i|0i
|0i
initial state time evolution measurement
・・・
・・・unitary operator
・・・{n qu
bit
H1
H2
H1,2
on
off
quantum circuit = spatiotemporal modulationof Hamiltonian
Kelly et al., Nature 519, 66 (2015)
Can we achieve universality with more natural dynamics?
Feynman quantum computer R. Feynman, Quantum mechanical computers, Opt. News, vol. 11, pp. 11–46, 1985
H =TX
t=1
Ut ⌦ |tiht� 1|+ U†t ⌦ |t� 1iht|
working space clock
initial state
working space clock| ini|0i
time evolution TX
t=0
ct(⌧)UtUt�1 · · ·U1| ini|tie�iH⌧
(history state)
Hamiltonian dynamics with time-independent Hamiltonian can implement universal quantum computation!
様々な万能量子計算モデル
量子回路型
万能計算を埋め込むために利用する自由度
時間・空間的にハミルトニアンを変動
Feynman型 定常ハミルトニアン
断熱型Aharonov et al, ‘04
(QMA-hardness経由)
定常ハミルトニアン+断熱操作(横磁場イジングはダメ Bravyi et al ‘06)
オートマトン型 定常並進対称ハミルトニアン+初期状態
[4-local: Feynman’85, 2-local: Nagaj ’10&12]
[5-local: Kitaev ‘02, 3-local: Kempe-Kitaev-Regev’06; 2D 2-local Oliveira-Terhal ’08; 2D fermions: Schuch-Verstraete ’09, 1D 2-local: Aharonov et al ‘09; 2D XY: Cubitt-Montanaro ’13, and many more]
量子ウォーク型 定常ハミルトニアン[adjacency matrix: Childs’09,bosons&fermions: Childs-Gosset-Webb’13, and many more]
定常並進対称ハミルトニアン+時間的に変動 [2D local: Janzig-Wocjan ‘04]
[1D local: Raussendorf ‘05]
Scrambling (random unitary)with translation invariant system
ZZ X Z ZZ X Z
{
…
mea
sure
simul
tane
ously
Hzz
= g
N�1X
i=1
�z
i
�z
i+1, Hz
= hz
NX
i=1
�z
i
, Hx
= hx
NX
i=1
�x
i
{ { { { {
Time intervals are chosen randomly.
Convergence to PT distribution#
of c
ount
s
probability × dim
8 qubit, depth 300
0.01
0.1
1
10
100
0 1 2 3 4 5 6 7 8 9 10
depth=3depth=15depth=30
depth=300
Convergence to PT distribution
depth
entro
py
4 qubit
5 qubit
6 qubit
7 qubit
1.5 2
2.5 3
3.5 4
4.5 5
5.5 6
0 50 100 150 200 250 300
4 qubit5 qubit6 qubit7 qubit8 qubit9 qubit
PT entropy
8 qubit
9 qubit
S = log(2
n)� 1 + �
PT entropy:
Euler’s constant
Convergence of PT entropy and decay of OTO correlator
entro
pyO
TOC
More systematic understanding
would be required!-0.2
0
0.2
0.4
0.6
0.8
1
0 20 40 60 80 100 120 140
1 2 3 4 5 6 7 8 9
0 20 40 60 80 100 120 140
entropyPT entropy
Higher moment of PT distribution?
9qubit
Sensitivity of PT distribution against decoherence
0.01
0.1
1
10
100
1000
0 1 2 3 4 5 6 7 8 9 10 0.01
0.1
1
10
100
1000
0 1 2 3 4 5 6 7 8 9 10 0.01
0.1
1
10
100
1000
0 1 2 3 4 5 6 7 8 9 10
0.01
0.1
1
10
100
1000
0 1 2 3 4 5 6 7 8 9 10 0.01
0.1
1
10
100
1000
0 1 2 3 4 5 6 7 8 9 10 0.01
0.1
1
10
100
1000
0 1 2 3 4 5 6 7 8 9 10
0.01
0.1
1
10
100
1000
0 1 2 3 4 5 6 7 8 9 10 0.01
0.1
1
10
100
1000
0 1 2 3 4 5 6 7 8 9 10
p=0.001 p=0.002 p=0.005
p=0.01 p=0.02 p=0.03
p=0.05 p=0.1 Depolarizing error with prob. p.
PT distribution rapidly converges to uniform distribution.
Summary & Discussion• Random unitary (universal computation) connects quantum
computer, quantum chaos, and blackhole.
• OTOC is a good witness of “randomness” of unitary, which seems to be sensitive to delocalization.
• PT distribution is directly measurable from the output of random unitary.
• PT distribution is sensitive to decoherence.
• Experimental realization with cold atom quantum simulator.
• Integrable v.s. quantum chaotic system (random free-fermionic dynamics→ random matchgate)?