Sensitivity run

[Quame]

Dear All,

Is there a way to view sensitivity run results while model is running? Also,is it possible to run Vensim Monte Carlo simulation in a batch mode?

Thanks

[Quame]

I am running a huge number of simulations and wanted to view the results, say on another computer before the simulation ends.

[tomfid]

You can script everything with a command script - see RUN_SENSITIVITY.

You can't interrupt a sensitivity run for viewing, but you could break the test into subsets.

[Quame]

Thanks a lot. I will look into the scripting for running in a batch mode then. Kind regards

[Quame]

I have gone through the manual trying to implement Monte Carlo in a batch mode. In a normal Vensim Monte Carlo simulation, after specifying the range (min-max) values for parameters of interest, the model automatically samples from this space during each run. Now in a batch mode using the command script, how do I do this without repeating runs of same parameter values? For other applications I have used before, random numbers within the parameter space is first generated after which the numbers are incessantly feed into the model for each subset. How do I do a similar thing, using the Vensim command script?

Thanks,

[tomfid]

I think what you'd have to do is make multiple copies of your control file (.vsc), each with a different random seed. Then use the command script to load each vsc and run_sensitivity in succession.

[Quame]

Okay, greatly appreciate it. I will try that. Thanks,

[Quame]

This works great for batching, except all subset simulations start from 1 to the end specified. Just wondering if there is a way to serially do the runs: say first subset = run 1- run 500, second subset = run 501- run 1000 etc, using the .vsc file. Thanks

[Administrator]

This works great for batching, except all subset simulations start from 1 to the end specified. Just wondering if there is a way to serially do the runs: say first subset = run 1- run 500, second subset = run 501- run 1000 etc, using the .vsc file. Thanks

There is no way to do this.

[Quame]

Good to know. Thanks for your assistance