Integration of quantum processing devices with distributed computers
Quantum processing devices are integrated with conventional distributed computing paradigms. In one aspect, ideas from classical distributed and high-performance computing are brought into the quantum processing domain. Various architectures and methodologies enable the bilateral integration of quantum processing devices and distributed computers. In one aspect, a system is composed of a high-level API and library, a quantum data model, and a set of software processes to prepare this data model for computation on a quantum processing device and to retrieve results from the quantum processing device. This provides a way for distributed computing software frameworks to integrate one or more quantum processing devices into their workflow.
1. A computing system comprising:
a master machine;
a physical quantum processing device that functions as a worker machine controlled by the master machine, wherein said physical quantum processing device is a gate-model quantum computing device;
an API stack, that provides an interface for the master machine to control any of a plurality of different types of conventional computers and quantum processing devices including said physical quantum processing device, comprising:
an interface communicating with the master machine using a conventional software form, including receiving a problem from the master machine and sending results based on the problem to the master machine using the conventional software form;
an interface to said physical quantum processing device, including configuring the problem on said physical quantum processing device and receiving low-level results based on the problem from said physical quantum processing device;
a conversion module that (A) converts the problem received from the master machine in conventional software form to a quantum data model amendable to solution on quantum processing devices including said physical quantum processing device; and (B) converts the low-level results based on the problem received from said physical quantum processing device to the conventional software form to send to the master machine; and
a device-specific optimization module that optimizes the quantum data model for solution on said physical quantum processing device; and
a domain-specific library containing routines to prepare the problem within a domain for solution by said physical quantum processing device, wherein the master machine calls the routines using conventional software calls and results of the routines are passed to the API stack, wherein the domain-specific library is a machine learning library.
2. The computing system of claim 1 wherein the API stack includes a module that decomposes the problem received from the master machine into computational tasks, wherein at least one of the computational tasks is assigned to be computed by said physical quantum processing device.
3. The computing system of claim 1 wherein the API stack includes a module that schedules computational tasks to be computed by said physical quantum processing device.
4. The computing system of claim 1 , wherein the API stack includes a module that allocates tasks among the quantum processing devices.
5. The computing system of claim 1 wherein the API stack includes a manual user interface for a module that allocates tasks among the quantum processing devices.
6. The computing system of claim 1 further comprising another domain-specific library that is a graph analytics library.
7. The computing system of claim 1 wherein the master machine includes a big data compute framework.
8. The computing system of claim 1 further comprising:
a distributed computing cluster that functions as a worker machine controlled by the master machine, the master machine allocating tasks between the distributed computing cluster and the physical quantum processing device.
9. The computing system of claim 1 wherein the API stack is a plug-in to the master machine.
10. The computing system of claim 1 wherein the physical quantum processing device is located remotely from the master machine.
11. A non-transitory computer readable medium containing:
an API stack that provides an interface for a master machine to control any of a plurality of different types of conventional computers and quantum processing devices including a physical quantum processing device, the API stack comprising:
an interface communicating with the master machine using a conventional software form, including receiving a problem from the master machine and sending results based on the problem to the master machine using the conventional software form;
an interface to said physical quantum processing device, wherein said physical quantum processing device is a gate-model quantum computing device, including configuring the problem on said physical quantum processing device and receiving low-level results based on the problem from said physical quantum processing device;
a conversion module that (A) converts the problem received from the master machine in conventional software form to a quantum data model amendable to solution on quantum processing devices including said physical quantum processing device; and (B) converts the low-level results based on the problem received from said physical quantum processing device to the conventional software form to send to the master machine; and
a device-specific optimization module that optimizes the quantum data model for solution on said physical quantum processing device; and
a domain-specific library containing routines to prepare the problem within a domain for solution by said physical quantum processing device, wherein the master machine calls the routines using conventional software calls and results of the routines are passed to the API stack, wherein the domain-specific library is a machine learning library.
12. The non-transitory computer readable medium of claim 11 wherein the API stack further comprises:
a module that allocates tasks among the quantum processing devices.
13. The non-transitory computer readable medium of claim 11 wherein the API stack is accessible from any of a plurality of remotely located master machines.
14. The non-transitory computer readable medium of claim 11 wherein the API stack further comprises:
a module that decomposes the problem received from the master machine into computational tasks, wherein at least one of the computational tasks is assigned to be computed by said physical quantum processing device.
15. The non-transitory computer readable medium of claim 11 wherein the API stack further comprises:
a module that schedules computational tasks to be computed by said physical quantum processing device.
16. The non-transitory computer readable medium of claim 11 wherein the API stack further comprises:
a manual user interface for a module that allocates tasks among the quantum processing devices.
17. The non-transitory computer readable medium of claim 11 further containing:
another domain-specific library that is a graph analytics library.
18. The non-transitory computer readable medium of claim 11 wherein the API stack is a plug-in to the master machine.