Hi, I'd like to execute Optimizer automatically.
Is it possible to automate repeated executions of Optimizer while changing settings such as which parameter to select?
If we can do this, the parameter 1, parameter 2, ..., parameter n can be optimized separately.
This is a useful strategy when you can break a big problem into independent smaller ones.
Please tell me more if you know about it.
Thanks.
There isn't a built in way to do this. In version 22.0, you can define multiple optimizations up front. But there isn't an automated way to apply the optimal value found in a previous optimization to the next.
In most cases, however, this approach will yield worse results than optimizing all parameters together. When you optimize with only a single parameter, you'll have to determine what the other parameter values should be. If you choose different values for the constant parameters, the optimal value for the varying parameter will change. If you choose poor values for the other parameters on the first round, you'll get the best of a bad situation. If you then use that value in subsequent optimizations, you're locking in a choice that was made given the bad situation.
I bring that up because I think it would be better to run with all the parameters you plan to optimize, and just let the optimizer run for more iterations. In version 22.0, all the results of your model runs are saved in a separate file. If you run the optimizer again, you no longer have to re-simulate the scenarios that have already been run; the optimizer just looks up the results from the file. This makes it easy to run for a long while, get the best scenario, and then, if you like, use that scenario to start the next optimization.
In most cases, however, this approach will yield worse results than optimizing all parameters together.
I know this point, but the repeated execution of optimizer is a useful method for problems that consist of small independent problems.
Are there any plans to implement optimizer autorun in the future?
As said you can do nearly anything with FlexSim. You can connect FlexSim as master or client. You use an external program to control parameters within FlexSim. In such a case you can do an optimization by AI or ML A multiple step analysis is individually evaluated by an users opinion. You decide what result is an optimum in a step. Such an approach is Hard to automate inside of FlexSim, but you can try to run FlexSim by batch execution from outside and start a sequence of optimization steps.
14 People are following this question.
© 1993- FlexSim Software Products, Inc. All Rights Reserved.
FlexSim is a fully 3D simulation software environment. FlexSim can be used to simulate any process in any industry.
FlexSim®, FlexSim Healthcare™, Problem Solved.®, the FlexSim logo, the FlexSim X-mark, and the FlexSim Healthcare logo with stylized Caduceus mark are trademarks of FlexSim Software Products, Inc. All rights reserved.
