High-Throughput Material Simulation Scheduling Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current high-throughput computational material simulation methods struggle to optimize the execution time of a group of compute-intensive simulation jobs, leading to inefficiencies in designing new materials.
Innovation Solution
A high-throughput computational material simulation optimization method based on time prediction, which involves establishing a predictive model using a deep neural network to forecast execution times and optimize job scheduling across multiple jobs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If traditional single-job optimization methods are used, then optimization implementation is simple, but overall optimization performance is restricted by local factors
Solution Approach 1:
The patent segments the optimization problem into two levels: individual job optimization (local) and group job optimization (global). By dividing the computation tasks into job groups based on physical relationships, the system can apply different optimization strategies at each level, achieving both implementation simplicity and improved overall performance through hierarchical optimization.
2Ease of operation
If local optimization for single jobs is performed, then optimization method is easy to implement, but macroscopic optimization potential is ignored
Solution Approach 1:
The patent transitions from single-job optimization to multi-job group optimization by adding a new dimension of analysis. It introduces job grouping based on physical relationships and uses a predictive model that considers group-level characteristics, thereby reducing execution time through macroscopic optimization while maintaining implementation feasibility.
3Ease of manufacture
If compute-intensive simulation jobs are executed without optimization, then job execution is straightforward, but execution time is excessively long
Solution Approach 1:
The patent applies preliminary action by building a predictive model before executing simulation jobs. The model is trained on historical execution data to predict execution times of compute-intensive jobs, allowing the system to pre-identify optimization opportunities and schedule jobs more efficiently, thereby reducing execution time while keeping the execution process straightforward.
4Extent of automation
If automation procedure focuses on computational stages only, then automation is achieved, but execution time optimization is insufficient
Solution Approach 1:
The patent introduces feedback mechanisms by using predictive models that learn from actual execution times of simulation jobs. The system continuously refines its predictions based on feedback from completed jobs, enabling it to progressively improve execution time optimization while maintaining automation of computational stages. This closed-loop approach addresses the insufficient execution time optimization in traditional automated workflows.
Data Source
AI summary
Provided in the present invention is a high-throughput material simulation calculation optimization method based on time prediction, relating to the field of materials science. The method comprises: first constructing a prediction model of task configurations and corresponding time predictions, and using the prediction model to generate the execution time of all of the tasks in a high-throughput material simulation calculation under different conditions; then generating an optimal scheduling plan for each model in the high-throughput material simulation calculation by means of directed graphs; and, according to the optimal scheduling plan for each model, sequentially executing all of the tasks until all of the tasks are completed. Further, a high-throughput computing simulation optimization apparatus based on a time prediction and a storage medium are provided.