Methods for solving computational problems using a quantum processor
View Patent ↗Methods for solving a computational problem including minimizing an objective including a set of weights and a dictionary by casting the weights as Boolean variables and alternately using a quantum processor and a non-quantum processor to successively optimize the weights and the dictionary, respectively. A first set of values for the dictionary is guessed and the objective is mapped to a QUBO. A quantum processor is used to optimize the objective for the Boolean weights based on the first set of values for the dictionary by minimizing the resulting QUBO. A non-quantum processor is used to optimize the objective for the dictionary based on the Boolean weights by updating at least some of the columns of the dictionary. These processes are successively repeated until a solution criterion is met. Minimization of the objective may be used to generate features in a learning problem and/or in data compression.
1. A method of minimizing an objective including a set of weights and a dictionary, the method comprising:
casting the set of weights in the objective as Boolean variables via a digital processor;
setting a first set of values for the dictionary via the digital processor;
optimizing the objective for a first set of values for the Boolean weights based on the first set of values for the dictionary by: mapping the objective to a first quadratic unconstrained binary optimization (“QUBO”) problem and by at least approximately minimizing the first QUBO problem via at least one of adiabatic quantum computation or quantum annealing performed by a quantum processor;
optimizing the objective for a second set of values for the dictionary based on the first set of values for the Boolean weights by updating at least some of the values for the dictionary via a non-quantum processor;
optimizing the objective for a second set of values for the Boolean weights based on the second set of values for the dictionary by mapping the objective to a second QUBO problem and by at least approximately minimizing the second QUBO problem via the quantum processor; and
optimizing the objective for a third set of values for the dictionary based on the second set of values for the Boolean weights by updating at least some of the values for the dictionary via the non-quantum processor.
2. The method of claim 1 , further comprising:
optimizing the objective for a t th set of values for the Boolean weights, where t is an integer greater than 2, based on the third set of values for the dictionary, by mapping the objective to a t th QUBO problem and by at least approximately minimizing the t th QUBO problem via the quantum processor;
optimizing the objective for a (t+1) th set of values for the dictionary based on the t th set of values for the Boolean weights by updating at least some of the values for the dictionary via the non-quantum processor; and
optimizing the objective for a (t+1) th set of values for the Boolean weights based on the (t+1) th set of values for the dictionary by mapping the objective to a (t+1) th QUBO problem and by at least approximately minimizing the (t+1) th QUBO problem via the quantum processor.
3. The method of claim 2 wherein the optimizing the objective for a (t+1) th set of values for the dictionary based on the t th set of values for the Boolean weights and the optimizing the objective for a (t+1) th set of values for the Boolean weights based on the (t+1) th set of values for the dictionary are repeated for incremental values of t until at least one solution criterion is met.
4. The method of claim 3 wherein the at least one solution criterion includes either convergence of the set of values for the Boolean weights or convergence of the set of values for the dictionary.
5. The method of claim 1 wherein minimizing an objective comprises generating features in a learning problem.
6. The method of claim 5 wherein the learning problem is selected from the group consisting of: pattern recognition, training an artificial neural network, and software verification and validation.
7. The method of claim 5 wherein the learning problem corresponds to either a machine learning problem or an application of artificial intelligence.
8. The method of claim 1 wherein the set of values for the dictionary includes a set of columns, and wherein updating at least some of the values for the dictionary includes updating at least some of the columns of the dictionary.
9. The method of claim 1 wherein minimizing an objective includes solving a sparse least squares problem.
10. The method of claim 1 wherein minimizing an objective includes performing data compression.
11. A method of minimizing an objective including a set of weights and a dictionary, the method comprising:
casting the set of weights in the objective as Boolean variables via a digital processor;
setting a first set of values for the Boolean weights via the digital processor;
optimizing the objective for a first set of values for the dictionary based on the first set of values for the Boolean weights by updating at least some of the values for the dictionary via a non-quantum processor;
optimizing the objective for a second set of values for the Boolean weights based on the first set of values for the dictionary by mapping the objective to a first quadratic unconstrained binary optimization (“QUBO”) problem and by at least approximately minimizing the first QUBO problem via at least one of adiabatic quantum computation or quantum annealing performed by quantum processor;
optimizing the objective for a second set of values for the dictionary based on the second set of values for the Boolean weights by updating at least some of the values for the dictionary via the non-quantum processor; and
optimizing the objective for a third set of values for the Boolean weights based on the second set of values for the dictionary by mapping the objective to a second quadratic unconstrained binary optimization (“QUBO”) problem and by at least approximately minimizing the second QUBO problem via the quantum processor.
12. The method of claim 11 , further comprising:
optimizing the objective for a t th set of values for the dictionary, where t is an integer greater than 2, based on the third set of values for the Boolean weights, by updating at least some of the values for the dictionary via the non-quantum processor;
optimizing the objective for a (t+1) th set of values for the Boolean weights based on the t th set of values for the dictionary by mapping the objective to a t th quadratic unconstrained binary optimization (“QUBO”) problem and by least approximately minimizing the t th QUBO problem via a quantum processor; and
optimizing the objective for a (t+1) th set of values for the dictionary based on the (t+1) th set of values for the Boolean weights by update-updating at least some of the values for the dictionary via the non-quantum processor.
13. The method of claim 12 wherein the optimizing the objective for a (t+1) th set of values for the Boolean weights based on the t th set of values for the dictionary and the optimizing the objective for a (t+1) th set of values for the dictionary based on the (t+1) th set of values for the Boolean weights are repeated for incremental values of t until at least one solution criterion is met.
14. The method of claim 13 wherein the at least one solution criterion includes either convergence of the set of values for the Boolean weights or convergence of the set of values for the dictionary, and further comprising determining whether convergence has occurred.
15. The method of claim 11 wherein minimizing an objective comprises generating features in a learning problem.
16. The method of claim 15 wherein the learning problem is selected from the group consisting of: pattern recognition, training an artificial neural network, and software verification and validation.
17. The method of claim 15 wherein the learning problem corresponds to either a machine learning problem or an application of artificial intelligence.
18. The method of claim 11 wherein the set of values for the dictionary includes a set of columns, and wherein updating at least some of the values for the dictionary includes updating at least some of the columns of the dictionary.
19. The method of claim 11 wherein minimizing an objective includes solving a sparse least squares problem.
20. The method of claim 11 wherein minimizing an objective includes performing data compression.