January 12, 2018

Using Cache in pSeven

It is common for an engineer to work with some computational model that substitutes a product being developed. As product complexity increases day by day, the complexity of corresponding models also increases. This fact leads to an increase in time, efforts and other resources needed to run these computational models, not to mention their comprehensive analysis and optimization. Therefore, it is necessary to provide an engineer with an opportunity to save all processed data, in order to be able to re-use it in the later experiments. In other words, to avoid running model calculations with parameter values, for which the model outputs have already been computed.

pSeven allows you to create cached regions in a workflow using Composite blocks to re-use evaluation data. The user has to select the "Enable Cache" checkbox and to give a name for the file, which all processed data will be saved in.

You can consider the following situations where it can be useful:

1. The optimization process is interrupted due to some technical problems.
• The data analysis and optimization algorithms implemented in pSeven can be run in a deterministic mode, so the interrupted process will continue from the latest not-finished iteration.
2. The problem statement is changed, but the computational model remains the same.

An example can be a recent problem from our customer. This is a single-objective optimization problem with 9 variable parameters (including discrete integer parameters) and 15 constraints. A single run of the third-party computational model to evaluate the objective function and constraints took about 12 minutes. This optimization problem was solved with the activated cache, so during this process, all data was stored in a file. After analyzing the optimal solution, the customer changed the problem statement. The second run of the problem with this cache reduced a total time for solving the problem by 2 times in comparison with a time needed for solving the original problem.