Phase estimation

estimation

One version of Shor’s algorithm is based on phase estimation. This appli-

cation of the quantum Fourier transform is used to estimate the eigenvalue of

a unitary operator, which is a phase:

ˆ

U|ui = e

|ui; θ = 2πφ (8.63)

where φ is a fraction.

As a preliminary to this algorithm, let’s look at a toy version. Suppose you

are given U and an eigenstate |ui. We have seen that the circuit of Figure 7.14

simulates a measurement of U.

|0i

H

H

|ui

U

|ui

Here,

1

2

|0i + |1i

⊗ |ui →

1

2

|0i + e

2πiφ

|1i

⊗ |ui. (8.64)

If φ were a single bit, then you can see that the output is 0 if φ = 0.0 and 1

if φ = 0.1. This circuit thus gives us the value in one run. But in general φ

will be several bits long. A measurement of the upper register in the H basis

will yield a 0 or 1 with probabilities cos

2

πφ and sin

2

πφ. A statistically large

number of measurements will allow us to recover φ from the counts. But this

is an inefficient method.

Note that the H transform on the upper register is the one-bit Fourier

transform. In order to estimate φ to more bits of accuracy, we must have

a qubit for each significant figure of φ and then perform an inverse Fourier

transform, as shown in the circuit of Figure 8.13.

|0i

H

. . .

QF T

−1

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

t qubits |0i

H

. . .

|0i

H

. . .

|ui /

n

U

U

2

. . .

U

2

t−1

|ui

FIGURE 8.13: Circuit for phase estimation.

Imagine φ upto t bits as

φ = 0.φ

1

φ

2

···φ

t

=

φ

0

2

t

, φ

0

= φ

t

φ

t−1

···φ

1

. (8.65)

Quantum Algorithms 167

Then we start with t working qubits in the input register, and use them to

control gates of the form U

2

k

. After the control gates, the output on the k

th

line is

1

2

(|0i + |1i) −→

1

2

|0i + e

2πi2

k

φ

|1i

(8.66)

=

1

2

|0i + e

2πi(0.φ

1

φ

2

···φ

k

)

|1i

. (8.67)

You can see that just before the QFT gate, the state of the upper register is

1

2

t

|0i + e

2πiφ

|1i

|0i + e

4πiφ

|1i

⊗ . . . ⊗

|0i + e

2πi.2

t

φ

|1i

(8.68)

=

1

2

t

2

t

−1

X

k=0

e

2πiφk

|ki (8.69)

This is just the QFT mod 2

t

of φ

0

and an inverse Fourier transform will give

you φ

0

exact to t significant figures.


Comments

Leave a Reply

Your email address will not be published. Required fields are marked *