Skip to main content
Quantum Error Correction
Phelim Bradley
Contents
• Why we need (Quantum) Error Correcting
  (QEC)
• Classical error correction
• Barriers to QEC
• Types of Errors
• Quantum Error Correcting Codes
Why do we need (Quantum) Error Correction?

• Theoretically operators and states are pristine
  and perfect.
• In a lab:
  ▫ Approximations must be made.
  ▫ Operators don’t always do as they should.
  ▫ Fidelity of the prepared states and the theoretical
    states is not perfect .
• Quantum Error Correction (QEC) deals with the
  imperfection of the real world.
Classical Error
• Say, probability failure per gate = p
• Probability getting right answer with n gates:
                         (1 p)n
Classical Error Correction
If you have a possible bit-flip error like:
    |0         (1 p) | 0        p |1
    |1        (1 p) | 1         p|0
 Can try an error correction code like:
     | 00       (1 p) | 00       p | 10
     | 11      (1 p) | 11       p | 01
Classical Error Correction
  But this hasn’t done much, because it’s
  possible to have a bit-flip in either bit.
| 00    (1 p)(1 q) | 00    p | 10 q | 01    pq | 11
| 11   (1 p)(1 q) | 11    p | 01 q | 10    pq | 11
   So now we can’t distinguish between
   states.
Classical Error Correction
However, with 3 bits we can make progress:
          Let p=q for simplicity.
                    3
| 000      (1 p) | 000              ...

                        {
                            | 001
                    2
           (1 p) p          | 010         ...
                            | 100


                        {
                        2   | 011
          (1       p) p     | 101         ...
                            | 110
               3
            p | 111
Classical Error Correction

• Recover by looking at the majority of bits.
• i.e. | 001    | 000     |0 L
         | 101       | 111        |1    L
•   Ideally p    0 in reality p
                                        2
                                  c1p       c2 p3   o( p 2 )
                                     2
•   Gain in accuracy available for cp    p
•   Increase in resources to reduce error
•   e.g. p    p17
                needs 18 bits.
Barriers to Quantum Error Correction
 • Measurement of error destroys
   superpositions. (Classically we can
   observe all bits)
Barriers to Quantum Error Correction
• No-cloning theorem prevents repetition.
      0 +    1       0 0 +        1 1
                 (   0 +   1 )(    0 +   1)
• Not the same state. => contradiction proves a
  cloning operator cannot exist.
• Means you can’t just keep “back up” of states.
  Must protect original.
• Also prevents some error correction codes.
Barriers to Quantum Error Correction
  • Must correct multiple types of errors
    (not just bit flips).
  • Must correct continuous errors and
    decoherence.
Know thy enemy - Errors
 • A general operator :

                |         A |
                                *
                      A A
 • A density matrix ρ describes the statistical
   state of a system.
Know thy enemy - Errors
 • A density matrix ρ describing the ensemble:
                                  *
                       p A A
 • Normalise
                          *
                    p A A     1
Know thy enemy – Errors
 Bit Flip X
        |0         |1           |1          |0
•2 possibilities => 2 operators
         10                           01
   A0                       A1
         01                           10
                        *               *
        (1 p) A0 A0             pA A
                                  1    1

              00   01            11         00   10   01
   (1 p)                    p
              10   11            01         10   00   11
Know thy enemy - Errors
 Phase Flip Z:      |0      |1       |0     |1

 Complete dephasing:
                                 (1 p)    pI / 2
 (depolarisation)

 Rotation R: R 0 = 0 , R 1 =ei 1
             At          R->Z
Quantum Error Correction
•Let us apply the classical repetition code to a
quantum state to try to correct a single bit flip
error:
              0 + 1         000 + 111
•This is just a tensor product of three
qubits, and different from cloning a state.
•Bit flip error (X) on 2nd qubit:
                   010 +     101

 Can’t measure the data!
Measure the error, not the data.
• Simultaneously measure:
                  (1)    ( 2)
                  z      z
                  ( 2)   ( 3)   On   010 +   101
                  z      z
•This will check for the orthogonality of the
qubits.
•We will measure -1 for both, showing a bit-
flip in the 2nd qubit.
•Correct by applying a bit-flip X
operator.
Classical Quantum Error Correction
• Recover by looking at the majority parity of bits.
• i.e. | 001     | 000     |0 L
         | 101     | 111        |1   L
•   Ideally p  0 in reality p   c1p 2 c2 p3   o( p 2 )
•   Gain in accuracy available for cp2 p
•   Increase in resources to reduce error
•   e.g. p p17needs 18 bits (per error type).
Correcting Phase (Z) Errors
  Hadamard transform H exchanges bit flip
  and phase errors:           |    (| 0   |1 ) / 2
     H( 0 + 1 )= + + - |           (| 0   |1 ) / 2
      X + = + , X - = - - (acts like phase flip)
      Z + = - , Z - = + (acts like bit flip)
  Repetition code corrects a bit flip error
   The same code in a new basis
   corrects a phase error!
         + +    -      +++ +      ---
Shor’s Code



•This is simply a combination of the two codes
above.
•Had to stick to one basis, so it’s a little less
intuitive.
•Correct for both X (bit-flip) and Z(phase-flip)
•Also, get Y errors for free Y=iXY!
Summary
•Applied concepts of classical error
correction to QEC.
•Learnt about quantum errors.
•Circumvented the problem caused by no
cloning and superposition.
•Learnt codes to correct for multiple types
of errors that can occur in quantum
computing.
Questions?