| Noisy Intermediate-Scale Quantum (NISQ) era)
= Quantum volume: SAAS] SSO] SHE TALE Chet
ot QAS9| ASS SARCe DAYS BAF
= ALO] 7hSst THEO] 7H, 7H AAO] OF FH ESO]
AAs S Crest Qek7; EAYSt
Limiting error rate
1 10! 102 10108 1051010710
Number of Qubits
IMlustration of the qubit quality vs quantity relationship.
Image credit: lohn Martinis, Google. ] Quantum error correction ag
= Chap. 10
= Consider the classical communication channel
= The effect of the noise in the channel is to flip the bit
being transmitted with probability p > 0, while the bit will
be transmitted without error with probability of 1—p. >
Binary symmetric channel.
= Example: majority voting
» Asimple encoding scheme: 0 000, 1 111
© The bit string 000 and 111 are referred to as the logical 0
and logical 1.
© Limitation: the probability that two or more bits are flipped is
I 3p2(1 —p) + p?. ® Error probability is p, = 3p? — 2p?
© The code can make the transmission more reliable provided
Pe
|i) state is taken to X|w) with probability of p. X is
. sometimes called as bit flip operator.
| = Encoding: a0) + b|1) > a|0,) + b|1,) = a|000) + b|111) ] Quantum error correction ua
= Recovery
e Based on the measured syndrome, apply the bit flip operator
X on the corresponding qubit
¥ bo +
0 ° fK we
0)
* Errors occur in continuous fashion and probabilistically. Is the
previous complete bit flip model valid?
° Before the syndrome measurement, it is superposition of no-error
case and error case. However, once the error syndrome i Is
measured, the quantum state becomes pure state again except
the multi- -qubit errors. ] Quantum error correction ag
Quantum error model
© Phase flip channel: if we send qubits through a channel,
then that channel flips the phase between |0) and |1) with
probability p.
© D> |i) state is taken to Z|) with probability of p. Z is
sometimes called as phase flip operator.
Suppose we work in the qubit basis | +) = (0) + |1))/v2,
| -) = (0) — |1))/v2. In this basis, phase flip operator Z
flips between | +) and | —).
Encoding: |0,) =| +++), [1,) =|-——) 2)
By checking the parity in | +) and | —)
basis, phase-flipped qubit can be
identified if only one qubit is affected. I) ;
|0)
&] &) &) i] The Shor code ua
Combination of the three qubit phase flip and bit flip codes.
First encode the qubit using the phase flip code:
10) > | ++ +), 11) > |---)
Next, encode each of these qubits using the three qubit bit flip code:
= (0)+12)) _, (looo)+}111)) = @)=11) _, (000)-1111))
a ae ae I
Overall,
° |0)>
w) sl +
Jo0,) = e0dbe123))C000)+1111))(1000)+1211)
To 2v2 lo)
(J000)|111))(J000)~|211))(1000)~1111))
2 It) > [t= oe io
Called concatenation of code i) H ++
The Shor code (or sometimes called as Jo)
9-qubit code) can detect and correct 10)
single qubit error.
jo) 4H ++ ] Classical linear error correction code a
Classical linear error correction code (section 10.4.1)
Multiplication operation and all other arithmetic operations are done
modulo 2.
* Also note that bitwise-addition modulo 2 is equivalent to bitwise-XOR.
Terminology: codeword, generator matrix, parity check matrix
A linear code C encoding k bits of information into n bit code space is
specified by an n by k generator matrix G whose entries are all elements
of Zz, that Is, zeroes and ones. > [n,k] code
k —bits message y is encoded as x = Gy, where the message y is treated
as a column vector.
Example
1
"G= | > G[0] = [0,0,0]", G[1] = [1,1,1]” > called as [3, 1] code
1
1 0
10
_|10
"Flo 4
o1
0
1
oY. G0, 1] = [0,0,0,1,1,1]", G[1, 0] = [1,1,1,0,0,0]",
called as [6,2] code ] Classical linear error correction code ua
= Parity check matrix H
* A [n,k] code is defined to consist of all n -element vectors x over
Z, such that Hx = 0 where H is an n—k by n matrix known as
the parity check matrix, with entries all zeroes and ones.
© Example
- =[0 1B) > ae=0 only for x = [0,0,0]" and x = [11,1]
Faia il Hx = 0 only for x = [0,0,0]? and x = [1,1,1]
+ For other x's with one bit different from x = [0,0,0]” and x = [1,1,1]7 ,
Hx will caiculate the position.
2 Suppose that we encode the message y as x = Gy, but an error e
due to noise corrupts x giving the corrupted codeword x’ =x+e
(+ means bitwise modulo 2). Because Hx = 0 for all codewords,
Hx' = He which is called as the error syndrome.
. ° HG=0
I ° pomming distance d(x, y): the number of places at which x and y
iffer.
+ To correct up to t bits, the distance between the closest codewords
should be at least 2t + 1.
+ [n,k,d] code {| Classical linear error correction code g
= Hamming code
© Suppose r > 2 is an integer and let H be the matrix whose
columns are all 2” — 1 bit strings of length r which are not
identically 0. This parity check matrix defines a [2” — 1,2" —r—1]
linear code known as Hamming code.
» Example r = 3 > [7,4] code
ooo1111
H=|01 10011
1010101
a
"
Geeccon
HKorecoHS
cCHHOCHOOS
BeEEROOS ] Quantum error correction code 4a
= CSS code: Calderbank-Shor-Steane code
© Originally based on Hamming code [7,4]
© |0)z = 10000000) + |1010101) + |0110011) + 1100110) +
|0001111) + |1011010) + +]0111100) + {1101001)
© |1), = |1111111) + 10101010) + |1001100) + |0011001) +
|1110000) + |0100101) + |100011) + |0010110)
» Can correct both bit-flip and phase-flip error
= Quantum error correction code generally can be defined in
terms of stabilizer code. (Section 10.5)
« «= Surface code
I
© Data abit
Ancita qubit
Z stabiler
sp xsaoiner | Overview of quantum computer architecture I)
= OMA] SHAS! AAG AS PAE BS
= BelA AAG Chet PaAel PooyAl Atol7t EMS
71E UAE FzzE
A201 O121& BM
Bog 8M aol
Awe INGA oo,
Bol 7AM =
( asiel eee! ae
‘Se aoe gsle
BOW Bele architecture 2144
a 27 SAYS FOE we
(#21401 que Be
ange
we SE MO}O] SO MO} ALAS wet
, 0 BF cwne
FON Bey HY/So}
, ee
a
Somer
‘212Mo1 Molo af 4
ae Cig S12 a
91
1 at
#48 NAB Se
7 "
