Granted Patent B2

US 12,699,913 · App. 18/953,748 · Granted Aug 4, 2026

Inventors: Chad Tyler Rigetti (Walnut Creek, CA); William J. Zeng (Berkeley, CA); Blake Robert Johnson (El Cerrito, CA); Nikolas Anton Tezak (Oakland, CA)

Assignee: Rigetti & Co, LLC

G06N10/40G06F13/1663G06F15/16G06N10/00G06N10/20G06N10/70G06F9/544

Abstract

Claims (70)

1 . A method of using a first quantum processor unit (QPU) to calibrate a second QPU in a hybrid quantum-classical computing platform, the method comprising:

obtaining control parameters for the second QPU as an initial calibration;

generating sampled data based on operating the second QPU with the control parameters;

storing the sampled data in shared classical memory;

using the sampled data in a quantum Hamiltonian learning process, wherein the quantum Hamiltonian learning process produces a learned Hamiltonian of the second QPU based on operating the first QPU as a trusted simulator; and

updating the control parameters for the second QPU based on evaluating the learned Hamiltonian against a calibration objective function.

2 . The method of claim 1 , wherein the quantum Hamiltonian learning process uses the first QPU to simulate subsystems of qubits of the second QPU.

3 . The method of claim 1 , further comprising performing an iterative process until a termination criterion is reached, each iteration of the iterative process comprising:

generating sampled data for the iteration based on operating the second QPU with the updated control parameters;

storing the sampled data for the iteration in the shared classical memory;

using the sampled data for the iteration in the quantum Hamiltonian learning process, wherein the quantum Hamiltonian learning process produces a learned Hamiltonian for the iteration; and

updating the control parameters based on evaluating the learned Hamiltonian for the iteration against the calibration objective function.

4 . The method of claim 3 , wherein the termination criterion is a target calibration accuracy.

5 . The method of claim 1 , wherein the calibration objective function is the average distance from a set of random sequences to the identity.

6 . The method of claim 1 , wherein the first QPU operates at a higher clock rate than the second QPU.

7 . The method of claim 1 , wherein the first QPU operates at a higher logical clock rate than the second QPU.

8 . The method of claim 1 , wherein the first QPU operates at a higher repetition rate than the second QPU.

9 . The method of claim 1 , wherein the second QPU has higher fidelity than the first QPU.

10 . The method of claim 1 , wherein the shared classical memory is connected to both the first QPU and the second QPU, and the shared classical memory is configured to share data between the first QPU and the second QPU.

11 . The method of claim 1 , wherein the first QPU comprises a first classical local memory, the second QPU comprises a second classical local memory, and the first classical local memory and the second classical local memory are connected to the shared classical memory for transferring data between the first QPU and the second QPU.

12 . The method of claim 1 , wherein the hybrid quantum-classical computing platform comprises a quantum communication link between the first QPU and the second QPU for teleportation of quantum states.

13 . The method of claim 1 , wherein generating the sampled data comprises sampling Pr(D|H), wherein a likelihood of the data D being from a given Hamiltonian H is given by

Pr

⁡

(

H

|

D

)

=

Pr

⁡

(

D

|

H

)

⁢

Pr

⁡

(

H

)

Pr

⁡

(

D

)

.

14 . A hybrid quantum-classical computing platform comprising:

a first quantum processor unit (QPU);

a second QPU;

shared classical memory; and

a classical computing resource configured to:

obtain control parameters for the second QPU as an initial calibration;

generate sampled data based on operating the second QPU with the control parameters;

store the sampled data in the shared classical memory;

use the sampled data in a quantum Hamiltonian learning process, wherein the quantum Hamiltonian learning process produces a learned Hamiltonian of the second QPU based on operating the first QPU as a trusted simulator; and

update the control parameters for the second QPU based on evaluating the learned Hamiltonian against a calibration objective function.

15 . The platform of claim 14 , wherein the quantum Hamiltonian learning process uses the first QPU to simulate subsystems of qubits of the second QPU.

16 . The platform of claim 14 , wherein the classical computing resource is configured to perform an iterative process until a termination criterion is reached, each iteration of the iterative process comprising:

generating sampled data for the iteration based on operating the second QPU with the updated control parameters;

storing the sampled data for the iteration in the shared classical memory;

using the sampled data for the iteration in the quantum Hamiltonian learning process, wherein the quantum Hamiltonian learning process produces a learned Hamiltonian for the iteration; and

updating the control parameters based on evaluating the learned Hamiltonian for the iteration against the calibration objective function.

17 . The platform of claim 14 , wherein the termination criterion is a target calibration accuracy.

18 . The platform of claim 14 , wherein the calibration objective function is the average distance from a set of random sequences to the identity.

19 . The platform of claim 14 , wherein the shared classical memory is connected to both the first QPU and the second QPU, and the shared classical memory is configured to share data between the first QPU and the second QPU.

20 . The platform of claim 14 , comprising:

a quantum communication link between the first QPU and the second QPU for teleportation of quantum states.

Assignments (2)

Continuity (4)

References Cited (144)