Parametric distribution

This feature is meant to speed up the solving time of parametric scenarios. To use it, you must check the Parametric distribution checkbox and choose the Parallel configuration.

Checking the Parametric distribution case will refresh the view with another set of inputs you can specify, as in the figure below:

The new information to provide is the Parallel configuration input. This allows you to choose how many cores to use:
  • Automatic configuration uses all available cores while respecting the amount of memory requested is available
  • Manual configuration allows you to set a number of cores N to use

Any configuration will create new jobs to compute the solution(s). The number of jobs created depends on the configuration and the scenario. The Manual configuration will create N jobs that will solve one or more steps and the Automatic will create as many jobs as parameters values in the scenario. So, in this last case only one solution is computed per job.

The amount of memory to allocate per job can also be specified using the input Memory (GB) per job.

If the dynamic memory is checked, that latter memory is only used by the scheduler to select the right nodes.

制約事項: As a first version, the only input archive supported must contain:
  • One Flux project
  • With only one parametric scenario