faropt.templates package

Submodules

faropt.templates.blackbox module

class faropt.templates.blackbox.AsyncOpt(stackname='faropt')

Bases: object

AsyncOpt class for typical black-box optimizattion problems where the function evaluation is slow. Take for example an engine simulation - each simulation may take an hour, and you don’t want to make the optimizer wait for an update. This class implements a serverless ask- and tell-optimizer interface to ask an optimizer for the next point to evaluate the function at, based on expected improvement (Bayesian opt), and users can tell the optimizer the results of an optimization to update the response surface

ask_model(modelname='')

asks the model for a new point to evaluate at, based on expected improvement

Parameters

modelname (string, opional) – UUID of the model

create_model(bounds, modeltag='')

Updates the model by pickling and uploading (overwriting) the opt model to a special async bucket generated by the stack

Parameters
  • bounds (list of tuples, like [(-1.0,1.0),(-5,5),(-10.5,10.5)..]) – UUID of the model

  • modeltag (string, optional) – Friendly name of the optimizer

list_models()

Prints a list of all models created so faar with UUID and modeltags

tell_model(xval, fval, modelname='')

tells the model what the fuction value was for a new point (xval). The model’s response surface is then updated based on this information

Parameters
  • xval (list) – Point that was evaluated, like [0,0], that is within the bounds set in create_model

  • fval (float) – function value measured at xval

  • modelname (string, opional) – UUID of the model

update_model(modelname, opt)

Updates the model by pickling and uploading (overwriting) the opt model to a special async bucket generated by the stack

Parameters
  • modelname (string) – UUID of the model

  • opt (scikit-opt) – object containing params of the optimizer

faropt.templates.opsplan module

class faropt.templates.opsplan.OpsPlanOpt(stackname='faropt')

Bases: object

faropt.templates.routing module

class faropt.templates.routing.RoutingOpt(stackname='faropt')

Bases: object

Module contents