Virtual Solver Abstraction for Quantum Processor Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hybrid computing systems face challenges in achieving high availability, failover, and load balancing when dealing with heterogeneous resources such as quantum processors, as existing methods are not atomic, automatic, or transparent, and do not guarantee load balancing.
Innovation Solution
The introduction of a contractual layer that abstracts heterogeneous quantum processors as uniform virtual solvers, each with defined requirements and features, allowing for dynamic resource allocation and failover, and implementing load balancing within the pool of physical resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If heterogeneous quantum processors are directly used as resources, then resource utilization flexibility is improved, but system reliability and availability deteriorate due to lack of uniform abstraction
Solution Approach 1:
The patent introduces a virtual solver as an intermediary layer between the user and physical quantum processors. This virtual solver abstracts the heterogeneous physical resources into a uniform interface, allowing users to interact with quantum processors without needing to understand their specific hardware details. The virtual solver manages resource allocation, failover, and load balancing, thereby improving both flexibility and reliability simultaneously.
Solution Approach 2:
The patent applies homogeneity by creating a unified virtual solver interface that presents all physical quantum processors as equivalent resources. Despite the underlying physical processors being heterogeneous (different architectures, qubit counts, error rates), the virtual solver masks these differences and presents a consistent, homogeneous API to users, enabling reliable resource allocation and failover mechanisms.
2Ease of operation
If manual resource selection is used, then resource allocation control is improved, but system complexity and operational difficulty worsen
Solution Approach 1:
The virtual solver implements self-service by automatically managing resource allocation, selection, and failover without requiring manual user intervention. The system monitors the health of physical quantum processors and automatically selects appropriate resources based on predefined criteria, while the load balancer dynamically distributes workloads. This automation reduces operational complexity while maintaining user control through high-level abstractions.
Solution Approach 2:
The system performs preliminary actions by pre-configuring virtual solvers with resource selection criteria, health check mechanisms, and failover policies before actual computation begins. This preparation enables the system to automatically respond to resource failures and allocation needs without requiring complex manual configuration, thereby reducing operational burden while maintaining control.
3Reliability
If atomic execution is not enforced, then resource sharing flexibility is improved, but data integrity and execution reliability worsen
Solution Approach 1:
The patent segments the execution control by isolating each problem execution as an independent atomic unit within the virtual solver queue. Each problem is assigned a unique execution context and is processed independently, preventing interference between concurrent executions. This segmentation ensures atomicity without requiring complex global synchronization mechanisms, as each problem's execution is encapsulated and managed separately by the load balancer and virtual solver.
Data Source
AI summary
Systems, methods and article provide the services of heterogeneous resources, for example the services analog processors, e.g., quantum processors, in a robust manner that can include high availability, failover, and load balancing of the heterogeneous resources. A virtual solver is selected based at least in part on a first set of requirements, a first set of analog processors is identified based at least in part on the first set of requirements, and a first handle returned to the first virtual solver. A load balancer may balance loads. Failure over may be implemented.