Tensor Network Contraction Control Program Memory Feasibility
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining whether the contraction of a tensor network is possible within the available memory capacity are time-consuming, as they require waiting for the completion of the contraction order calculation, which can take a long time and is inefficient.
Innovation Solution
A program that quickly determines if the contraction of a tensor network is feasible by assessing the number of edges in the network and comparing it to a reference value, allowing for early determination of memory capacity sufficiency without completing the entire contraction order calculation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the completion of contraction order calculation is waited for to determine memory capacity sufficiency, then the determination accuracy is improved, but the processing time increases significantly
Solution Approach 1:
The patent performs preliminary estimation of memory capacity requirements during the contraction order calculation process itself, rather than waiting for completion. The estimation unit calculates estimated memory capacity requirements at intermediate stages and compares them against available memory capacity to enable early determination of feasibility, thus resolving the contradiction between accurate determination and processing time.
2Reliability
If the entire contraction order calculation is completed before determining feasibility, then the determination reliability is improved, but the productivity decreases due to unnecessary calculations
Solution Approach 1:
The patent performs only partial contraction order calculation - sufficient to obtain a reliable estimate of memory capacity requirements - rather than completing the entire calculation. The estimation unit determines feasibility based on intermediate calculation results and estimated memory usage, allowing the system to stop calculations early when memory insufficiency is detected, thus improving productivity while maintaining adequate determination reliability.
Solution Approach 2:
The patent implements a feedback mechanism where the estimation unit continuously monitors estimated memory capacity requirements during the calculation process and compares them against available memory capacity. When the estimated requirements exceed available capacity, the system provides feedback to terminate further calculations, enabling efficient feasibility determination without completing unnecessary calculations.
Data Source
AI summary
A non-transitory computer-readable recording medium stores a tensor network contraction control program for causing a computer to execute a process including: determining whether or not contraction of a tensor network that includes a plurality of tensors coupled to each other is accomplishable in a range of an available memory capacity based on a number of edges included in the tensor network.