8.1
By expanding the tensor product, we have the all possible binary numbers.
As discussed in 6.1.4, for a single qubit, Hadamard gate can be implemented as:
using the coherent drive:
that's on resonance and .
To apply it to identical flux qubits, we have to have different gaps on neighboring qubits so that the drive doesn't affect neighboring qubits.
8.2
Since and ,
8.3
where . Hence, we have:
Then,
So in the interaction picture with , the system is described with itself. Note that, in this picture, the photon has no energy. (Photon energy is absorbed in ).
We split as:
and take as a weak perturbation.
Now we apply the nondegenerate perturbation theory. In general, the transformation generator is given by:
We consider the states with minimum number of photons (or subspace of JC-Ladder), such as:
Since photons have no energy, their respective energy are:
Then, can be expanded as:
Note that, with similar calculations, you can confirm that the above expression for is valid for any subspace of JC-Ladder though we explicitly calculated in subspace.
So the perturbative correction to the Hamiltonian is given by:
Now,
Hence,
So, the effective Hamiltonian (in the interaction picture) is:
--- (1-1)
Since , we can go back to the original Schroedinger picture without modifying . So getting back to the original Schroedinger picture and dropping the photon energy , we have:
(2) Defining , we have:
Hence states are eigenstates of whereas states are mixed together via time-evolution.
As discussed in 8.3.2, when , the time evolution operator becomes:
So you can implement iSwap with up to local transformations:
--- (2-1)
(3)
From (1-1), the effective energy of qubits increase . This would influence the local transformation in (2-1). But I don't see how it affects the two-qubit exchange???
8.4
We use the unit system with , and a consistent notation for energy levels (instead of ).
(1) The eigenvalues of subspace are:
where
The eigenvalues of subspace are:
where
(2) From the results of (1), energy levels degenerate at and , that is, and
(3) In , for large , we can ignore . So approximating as , we have:
They respectively correspond to states .
(4) Relabel the eigenvalues as:
Then, the phase changes are described as:
(5) Define:
Then,
If we modify the adiabatic change times slower, is modified as:
So, by adjusting the speed of the adiabatic change, we can achieve arbitrary value of .
Note that once we achieve the transformation with , by applying single qubit operations, we can achieve the transformation:
(6) N-bit Quantum Fourier transformation can be implemented with control-phase gates:
So we need .
8.5
Hence,
8.6
(1) Asymmetric totally depolarizing channel:
This applies the phase flip (along the axe) with probability .
Since , by setting , we have:
In this symmetric case, with probability p, the system is replaced with the completely mixed status .
(2) Random flip channel is implemented as:
This is a special case of the asymmetric totally depolarizing channel .
(3) For ,
So applying it times, we have:
(4) Kraus operators for asymmetric totally depolarizing channel are:
Kraus operators for dephasing channel are:
Actually,
Since , for asymmetric totally depolarizing channel,
For other channels, you can confirm with similar calculations.
8.7
Define and you can confirm with a direct calculation that:
.
SInce [tex|displaystyle {\rm tr}(A\otimes B)={\rm tr}(A){\rm tr}(B)], it's also true that:
.
So, under the inner product they are linear-independent and become orthobormal basis of dimension linear space.
As the matrix has elements, it can be expressed by a liner combination of .
And the coefficients are determined by .
8.8
In the textbook, matrix representation and operator notation are somewhat mixed up. In the operator notation, we have:
--- (1)
where
--- (2)
where
From (1), we have:
.
From (2), we have:
Now we prove (8.16). For any ,
On both sides, we apply from left:
Since , we have:
--- (3)
Now we take , and using the relationship , we have:
8.9
(1) Deduce and for .
For , from Exercise 8.8 (1), we have:
Applying and taking , we have:
For ,
Hence, from Exercise 8.8 (3),
(2) and for .
We use the Einstein summation convention in the following calculations.
For ,
where
If either or is , .
For other cases, from the relationships:
we have:
(1st term)
(2nd term)
(3rd term)
For ,
where
We use the convention that , then we have:
(3) and for CNOT .
Hence for ,
For ,
8.10
The error mapping process is described as:
In the operator form, this can be described as:
Now the matrix elements of are calculated through the relationships:
For our particular case, using the notation :
Non-zero elements are:
By taking the limit , the error-free version is:
Hence, the process fidelity is:
The average fidelity is:
The matrix elements of can be calculated through the relationship:
8.11
8.12
Define as a probability of depolarization, and the totally depolarization channel is:
Hence, for a pure state
So the fidelity for this particular state is:
Since this is independent of the state, the average over all pure states is given by:
Note that the depolarization parameter in the textbook (p.220) is a probability of not decaying: . So (8.24) should be: