Logical Qubit Allocation Using Connectivity Graphs and Qubit Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for allocating logical qubits in quantum processors, such as those using exchange operations, lead to increased error risk and require excessive computational time, especially for algorithms with a large number of qubits, and are not optimized for neutral-atom quantum processors.
Innovation Solution
A method involving a connectivity graph and logical operation graph to rank qubits by weight, iteratively allocate them to physical qubits using random selection, and calculate scores to minimize exchange operations, allowing efficient allocation of a large number of qubits with polynomial time complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If exchange operations are used to make remote physical qubits interact, then connectivity between qubits is improved, but the number of operations increases and error risk increases
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing optimal exchange operation sequences in a lookup table before execution. The allocation unit determines the minimal sequence of exchange operations needed to bring remote qubits within interaction range, and these pre-computed paths are stored for rapid retrieval during algorithm execution, minimizing runtime overhead and reducing error opportunities
2Reliability
If exact allocation methods are used to minimize exchange operations, then reliability is improved, but computational time increases exponentially
Solution Approach 1:
The patent segments the allocation problem into two independent parts: (1) the allocation unit determines optimal qubit allocation and minimal exchange operation sequences, and (2) the execution unit retrieves pre-computed solutions from a lookup table. This segmentation allows the computationally intensive optimization to be performed once offline, while runtime execution becomes extremely fast, resolving the contradiction between optimization quality and computational speed
Solution Approach 2:
The patent performs the computationally intensive exact allocation optimization in advance, storing results in a lookup table. During actual quantum algorithm execution, the system simply retrieves pre-computed allocation configurations, reducing computational time from exponential to constant time while maintaining optimal reliability
3Productivity
If the number of physical qubits is increased to allocate more logical qubits, then productivity is improved, but device complexity increases
Solution Approach 1:
The patent introduces a temporal dimension by separating allocation optimization (offline, pre-computation) from execution (online, real-time). The lookup table stores allocation configurations across multiple dimensions (different algorithm types, different qubit counts), allowing the system to handle larger numbers of logical qubits without proportionally increasing runtime complexity or physical hardware requirements
Data Source
AI summary
According to one aspect, the subject matter of the present description is a method for allocating logical qubits of a quantum algorithm to physical qubits of a quantum processor arranged according to a given architecture, the method including: establishing, from the architecture, a connectivity graph; the establishing, from the quantum algorithm, a graph of the logical operations; assigning a weight to each logical qubit; ranking the logical qubits according to the weight; allocating a first logical qubit chosen according to the weight, to a first physical qubit of the quantum register; iteratively allocating the logical qubits remaining to be allocated, including calculating a score; and reproducing the iterative allocation procedure until allocating the last logical qubit remaining to be allocated, thus resulting in obtaining a configuration for allocating logical qubits of the quantum algorithm to physical qubits of the quantum processor.