1
$\begingroup$

I am trying to write Toffoli gate matrice by using one and two qubit gates matrices. I follow this circuit

link for the circuit

enter image description here

I first started to write the matrices of one and two qubit gates:

identity = np.array([[1, 0], [0, 1]])
xgate = np.array([[0, 1], [1, 0]])
ygate = np.array([[0,-1j],[1j,0]])
zgate = np.array([[1,0],[0,-1+0j]])
hgate = 1/math.sqrt(2)*(xgate+zgate)
sgate = np.sqrt(zgate)
tgate = np.sqrt(sgate)
tdag = tgate.conj().T
cnot = 0.5*(np.kron(identity,identity)+np.kron(identity,xgate)+np.kron(zgate,identity)-np.kron(zgate,xgate))
deneme = np.kron(identity, swap)
deneme2 = np.kron(cnot,identity)
deneme3 = np.kron(identity,swap)
#deneme_ = deneme*deneme2*deneme3
deneme_ = np.matmul(deneme,deneme2)
non_adjacent  = np.matmul(deneme_,deneme3)

Then, I started to write each part of the circuit in the following way:

tof1 = np.kron(np.kron(identity,identity),hgate)
tof2 = np.kron(identity,cnot)
tof3 = np.kron(np.kron(identity,identity),tdag)
tof4 = non_adjacent
tof5 = np.kron(np.kron(identity,identity),tgate)
tof6 = np.kron(identity,cnot)
tof7 = np.kron(np.kron(identity,identity),tdag)
tof8 = non_adjacent
tof9 = np.kron(np.kron(identity,tdag),tgate)
tof10 = np.kron(cnot,hgate)
tof11 = np.kron(np.kron(identity,tdag),identity)
tof12 = np.kron(cnot,identity)
toffoli = tof1*tof2*tof3*tof4*tof5*tof6*tof7*tof8*tof9*tof10*tof11*tof12

And the final matrix is that: enter image description here

My guess is that: I did a mistake when I tried to write cnot gate for the first and third qubits or maybe I am wrong to write identities

Can someone explain to me what I missed? Sorry for this dumb question Thanks in advance

Improve this question
$\endgroup$
7
  • $\begingroup$ Here is a "manual" how to write a matrix representing CNOT working on non-adjacent qubits (in your case tof4 and tof8 steps which you have wrong): quantumcomputing.stackexchange.com/questions/9180/… $\endgroup$
    – Martin Vesely
    Dec 28, 2021 at 10:19
  • $\begingroup$ @MartinVesely thanks, I wrote swap gate and then I wrote this code lines: ``` tof4 = np.matmul(swap,cnot) tof4 = np.kron(tof4,identity)# ``` and I did the same for tof8 too but the result is still weird $\endgroup$
    – quest
    Dec 28, 2021 at 11:43
  • $\begingroup$ also I tried to write the matrix which is described here: quantumcomputing.stackexchange.com/questions/9180/… directly but still did not work for me $\endgroup$
    – quest
    Dec 28, 2021 at 11:47
  • $\begingroup$ I see now, you have to multiply the matrices instead of summing them - Toffoli = tof13*tof12....tof1. Moreover, check again each step, I see a S gate in tof13 however the last step is CNOT acting on first and second qubits. $\endgroup$
    – Martin Vesely
    Dec 28, 2021 at 12:01
  • 1
    $\begingroup$ The issue is you're using * instead of @ to combine the pieces. For numpy, * is pair-wise product, and @ is matrix multiplication. $\endgroup$
    – Craig Gidney
    Dec 29, 2021 at 0:08

1 Answer 1

Reset to default
1
$\begingroup$

Here I am providing working code:

import numpy as np
import math as m

idn = np.array([[1, 0], [0, 1]])

h = (1/m.sqrt(2))*np.array([[1, 1], [1, -1]])

t =    np.array([[1, 0], [0, (1/m.sqrt(2))*(1+1j)]])
tdag = np.array([[1, 0], [0, (1/m.sqrt(2))*(1-1j)]])

cnot_adj = np.array([[1, 0, 0, 0],
                     [0, 1, 0, 0],
                     [0, 0, 0, 1],
                     [0, 0, 1, 0]])

cnot_non_adj = np.array(
                        [[1, 0, 0, 0, 0, 0, 0, 0],
                         [0, 1, 0, 0, 0, 0, 0, 0],
                         [0, 0, 1, 0, 0, 0, 0, 0],
                         [0, 0, 0, 1, 0, 0, 0, 0],
                         [0, 0, 0, 0, 0, 1, 0, 0],
                         [0, 0, 0, 0, 1, 0, 0, 0],
                         [0, 0, 0, 0, 0, 0, 0, 1],
                         [0, 0, 0, 0, 0, 0, 1, 0],
                        ]
                        )

toffoli = np.kron(np.kron(idn, idn), h)
toffoli = np.dot(np.kron(idn, cnot_adj), toffoli)
toffoli = np.dot(np.kron(np.kron(idn, idn), tdag), toffoli)
toffoli = np.dot(cnot_non_adj, toffoli)
toffoli = np.dot(np.kron(np.kron(idn, idn), t), toffoli)
toffoli = np.dot(np.kron(idn, cnot_adj), toffoli)
toffoli = np.dot(np.kron(np.kron(idn, idn), tdag), toffoli)
toffoli = np.dot(cnot_non_adj, toffoli)
toffoli = np.dot(np.kron(np.kron(idn, t), t), toffoli)
toffoli = np.dot(np.kron(cnot_adj, h), toffoli)
toffoli = np.dot(np.kron(np.kron(t, tdag), idn), toffoli)
toffoli = np.dot(np.kron(cnot_adj, idn), toffoli)

np.set_printoptions(precision=3) #"rounding"
np.set_printoptions(suppress=True) #supressing scientific format"
print('Real parts')
print(toffoli.real)
print('\nImaginary parts')
print(toffoli.imag)

Variable cnot_non_adj is CNOT gate with control on the uppermost qubit and target on the lowermost qubit. The matrix is designed according to manual I provided here.

Note that np.dot is a matrix multiplication. You can see, that the order in the multiplication is reversed in comparison with the diagram.

The result of the code is

Real parts
[[1. 0. 0. 0. 0. 0. 0. 0.]
 [0. 1. 0. 0. 0. 0. 0. 0.]
 [0. 0. 1. 0. 0. 0. 0. 0.]
 [0. 0. 0. 1. 0. 0. 0. 0.]
 [0. 0. 0. 0. 1. 0. 0. 0.]
 [0. 0. 0. 0. 0. 1. 0. 0.]
 [0. 0. 0. 0. 0. 0. 0. 1.]
 [0. 0. 0. 0. 0. 0. 1. 0.]]

Imaginary parts
[[-0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0. -0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  0. -0. -0.]
 [ 0.  0.  0.  0.  0.  0. -0. -0.]]

This is a matrix description of the Toffoli gate.

Improve this answer
$\endgroup$
2
  • 1
    $\begingroup$ Ah that is : np.set_printoptions(precision=3) #"rounding" np.set_printoptions(suppress=True) #supressing scientific format" Thank you very much now my code is also working agter your warnings. I deleted tof13 and I multiplied like you did and it is now working. Many thanks. This is great :)))) $\endgroup$
    – quest
    Dec 28, 2021 at 12:51
  • 1
    $\begingroup$ @quest: You are very welcome. I am glad I helped! $\endgroup$
    – Martin Vesely
    Dec 28, 2021 at 13:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.