2Q RB generated by cross-resonance CNOT #276
Conversation
| * Calibrate readout pulses and optimize state discrimination. | ||
| * Calibrate and optimize qubit XY pulses. Single qubit operations are based on x90, -x90, x180, y90, -y90, and y180. | ||
| * Our two qubit RB is based on CNOT as the entangling operation which is implemented by direct CR gate. Replace the pertinent lines in the definition of the function `play_sequence` in [run_two_qubit_rb_CR_CNOT.py](run_two_qubit_rb_CR_CNOT.py) (switch case 49) if echoed CR is to be used. | ||
| * No baking will be directly called by the user. The user can easily customize their gates, especially the two-qubit gate pulses. |
There was a problem hiding this comment.
I think we can delete this note since not everyone is familiar with baking. Maybe we can write:
The user can easily customize their gates, especially the two-qubit gate pulses.
| * Calibrate and optimize qubit XY pulses. Single qubit operations are based on x90, -x90, x180, y90, -y90, and y180. | ||
| * Our two qubit RB is based on CNOT as the entangling operation which is implemented by direct CR gate. Replace the pertinent lines in the definition of the function `play_sequence` in [run_two_qubit_rb_CR_CNOT.py](run_two_qubit_rb_CR_CNOT.py) (switch case 49) if echoed CR is to be used. | ||
| * No baking will be directly called by the user. The user can easily customize their gates, especially the two-qubit gate pulses. | ||
| * Running gate sequences other than those for RB is also allowed if one manually set the `sequence_list`. (See ##### TEST EXAMPLE in [run_two_qubit_rb_CR_CNOT.py](run_two_qubit_rb_CR_CNOT.py).) |
There was a problem hiding this comment.
(See example in run_two_qubit_rb_CR_CNOT.py.)
There was a problem hiding this comment.
correcting myself:
(See TEST EXAMPLE in run_two_qubit_rb_CR_CNOT.py.)
| The configuration consists of 4 elements: | ||
| * `rr1` and `rr2` send readout pulses and measure the transmitted signal for each qubit | ||
| * `q1_xy` and `q2_xy` send drive XY pulses to each qubit | ||
| * `cr_drive_c1t2`, `cr_drive_c2t1` send cross-resonance drive to the control qubit (at target frequency) |
There was a problem hiding this comment.
Let's add - In cr_drive_c1t2, qubit 1 acts as the control qubit, and qubit 2 as the target. Conversely, in cr_drive_c2t1, qubit 2 is the control qubit, and qubit 1 is the target.
| * `rr1` and `rr2` send readout pulses and measure the transmitted signal for each qubit | ||
| * `q1_xy` and `q2_xy` send drive XY pulses to each qubit | ||
| * `cr_drive_c1t2`, `cr_drive_c2t1` send cross-resonance drive to the control qubit (at target frequency) | ||
| * `cr_cancel_c1t2`, and `cr_cancel_c2t1` send cancellation drive the target qubit (also at target frequency) |
There was a problem hiding this comment.
Let's add - In cr_cancel_c1t2, qubit 1 acts as the control qubit, and qubit 2 as the target. Conversely, in cr_cancel_c2t1, qubit 2 is the control qubit, and qubit 1 is the target.
|
|
||
| The qubit elements `q1_xy` and `q2_xy` have several operations defined that correspond to X/2, Y/2, -X/2, -Y/2, X and Y gates. The envelopes are Gaussian (although this can be changed by the user), with optional Gaussian DRAG corrections, parameters (amplitudes, lengths, Gaussian sigmas, etc.), that need to be calibrated. | ||
|
|
||
| The qubit element `cr_drive_c1t2` and `cr_drive_c2t1` have square pulse and blackman function options for the CR envelopes. |
| # Only played the depth corresponding to target_depth | ||
|
|
||
| with for_(n, 0, n < n_avg_, n + 1): | ||
| align() |
There was a problem hiding this comment.
no need for this align :)
| depth_list = [0, 1, 5, 7, 14, 20, 26, 32] | ||
| np.random.seed(seed=int(time.time()%1*1e8)) | ||
| # np.random.seed(seed = 0) # can set the seed for consistent comparison | ||
| sequence_list, len_list = pre_generate_sequence(num_of_sequences, depth_list) # standard RB |
There was a problem hiding this comment.
When running this lune I get an error "TypeError: LineQubit.new() missing 1 required positional argument: 'x'"
After a bit of debugging it looks like it comes from the function generate_sequence_list inside the file TwoQ_RB_Sequence_Generation_CNOT_CR. Seems like that these lines create the error:
with open('2q_Clifford_gen_CNOT_circuit.pkl', 'rb') as file:
circuit_list = pickle.load(file)
are you familiar with it? maybe it's the cirq version? which one are you using?
There was a problem hiding this comment.
I updated my cirq to 1.4.1 and reran the clifford generation code. Now it's working.
| @@ -0,0 +1,363 @@ | |||
|
|
|||
There was a problem hiding this comment.
Can you put a reference paper?
| instruct.extend(instruct_list[sequence_ints[d_]]) | ||
| unitary = cirq.unitary(circuit) | ||
| for c_ in range(11520): | ||
| if abs((unitary @ unitary_list[c_]).trace())>3.95: |
There was a problem hiding this comment.
can you please explain about this loop?
| ### 3.2 Play multiple random sequences with different depths | ||
|
|
||
| #### 3.2.1 Stacking all `instruct_integers` into `sequence_list` | ||
| It appears that we cannot let qm loop over an numpy array. However, we have one `instruct_integers` for each randomization and for each different depth. The preferable way is to use a single list to store all `instruct_integers` from different randomizations and for different `depth`'s. This is done by the function `pre_generate_sequence`. |
There was a problem hiding this comment.
It is possible to loop over the an array in QUA using for_each_ and then the logic of the loop would be a bit different. (you can read more about it here - https://docs.quantum-machines.co/latest/docs/Guides/features/#for_each) but since it's a bit less efficient I think that the logic of what you did is great.
We can just remove the comment since it's might be confusing and keep the explanation of the logic.
| align(qt_xy, qc_xy, cr_drive, cr_cancel) | ||
| frame_rotation_2pi(+0.25, qc_xy) | ||
| # check the sign before 0.25 -- could be either way depending on cr_drive_phase | ||
| play('y90' * amp(0.00001), qc_xy, duration=1) |
Hi Michal and Kevin,
cc @bornman-nick
This is the pull request to upload our homemade 2q RB code for cross-resonance CNOT gates. These codes describe our method to get around the baker library which we found less convenient.
To summarize, a new folder named "Use_Case_1_Two_qubit_RB_without_baking" has been added into directory
/qua-platform/qua-libs/tree/main/Quantum-Control-Applications/Superconducting/Two-Fixed-Coupled-Transmons/Please let me know if you have any comments!
Best,
Ziwen