2.4. pyopus.evaluator.auxfunc — Auxiliary functions for parameter management

Auxiliary functions for parameter management

Parameter descriptions in list form consist of arrays or lists where every one of them specifies one propery for every parameter. An additional list is needed that holds the names of the parameters. An example:

paramLow =[0.1, 0.2, 0.5, 0.2]
paramHigh=[0.5, 0.9, 0.7, 0.5]
paramName=['a', 'b', 'c', 'd']

The three lists specify the lower and the upper bounds, and the names of four parameters (a, b, c, and d). Every list specifies one aspect of all parameters. List form specifies an ordering for the parameters.

An alternative is to organize this information in dictionary form. This way the parameters are organized in a dictionary with parameter name as key and parameter information as value. Parameter information is a dictionary with parameter aspect name as key and parameter aspect as value. Dictionary form does not imply parameter ordering, except for the one obtained by the keys() method of the dictionary object. The above example could be organized in dictionary form as:

params={
  'a': {
    'lo': 0.1, 
    'hi': 0.5
  }, 
  'b': {
    'lo': 0.2, 
    'hi': 0.9
  }, 
  'c': {
    'lo': 0.5, 
    'hi': 0.7
  }, 
  'd': {
    'lo': 0.2, 
    'hi': 0.5
  }
}

Conevrsion between list form and dictionary form of parameter description is performed by the dictParamDesc() and the listParamDesc() function. The list form can be converted to the dictionary form using:

params=dictParamDesc({'lo': paramLow, 'hi': paramHi}, paramName)

To extract only selected parameters from the list form, one can specify the indices:

# Extract only the first two parameters
params=dictParamDesc({'lo': paramLow, 'hi': paramHi}, paramName, indices=[0,1])

Conversion in the opposite direction can be performend with:

# paramName is the parameter name list specifying the parameter order
paramLists=listParamDesc(params, paramName)
# Extract the paramLow and the paramHigh list
paramLow=paramLists['lo']
paramHigh=paramLists['hi']

Selected parameters can be extracted by specifying indices:

# Extract only the first two parameters specified in the nameList
paramLists=listParamDesc(params, paramName, indices=[0,1])

or by specifying a shorter parameter list:

# Extract parameters a and b
paramLists=listParamDesc(params, ['a', 'b'])

Finally, one can extract only one property by specifying the propName parameter:

paramLow=listParamDesc(params, paramName, propName='lo')

Parameter values can be given as lists or as dictionaries. In list form the corresponding parameter names must be specified by a list of names. This list also implies the parameter ordering:

paramValues=[0.1, 0.5, 0.7, 0.3]
paramNames =['a', 'b', 'c', 'd']

In dictionary form parameter names are specified implicitly with keys, but no ordering is implied apart from that obtained with the keys() method of the dictionary:

paramDictionary={
  'a': 0.1, 
  'b': 0.5, 
  'c': 0.7, 
  'd': 0.3
}

Conversion between the two forms of parameter values is performed using the paramList() and the paramDict() function:

# To dictionary form
paramDictionary=paramDict(paramValues, paramNames)

# To list form, paramNames specifies the ordering for the list form
paramValues=paramList(paramDictionary, paramNames)

Selected parameters can be extracted by specifying the indices parameter:

# Extract only the first two parameters given in the paramValues list, 
# return the parameters in dictionary form
paramDictionary=paramDict(paramValues, paramNames, indices=[0,1])

# Extract the parameters corresponding to the first two names in paramNames, 
# return parameters in list form
paramValues=paramList(paramDictionary, paramNames, indices=[0,1])
pyopus.evaluator.auxfunc.dictParamDesc(inputList, nameList, indices=None)

Processes property value lists stored in a dictionary with property name for key. Returns a parameter definition dictionary with parameter name for key and a property dictionary corresponding to a parameter for value.

nameList is a list of parameter names corresponding to individual propeerty list positions.

If indices is given only the parameters with specified indices are included in the output.

If property value is None it is not put into the corresponding parameter’s dictionary.

pyopus.evaluator.auxfunc.listParamDesc(inputDict, nameList, propName=None, indices=None)

Processes parameter definition dictionary and produces ordered property value lists.

A parameter definition dictionary has entries with parameter names for keys and property dictionaries for values. A property dictionary has property names for keys and stores peroperty values.

nameList is the list of parameter names used in vectorization.

Returns a dictionary with property names for keys and property value lists for values. The order of values in the property value vectors corresponds to the ordered parameter names.

If some property is not given for a parameter the corresponding entry in the property value list is is set to None.

If propname is given a single list is returned corresponding to the specified property.

If indices is given, only the parameters with specified indices are included in the output.

pyopus.evaluator.auxfunc.paramList(inputDict, nameList, indices=None)

Coverts a dictionary of values into a vector of values. nameList is a list of keys specifying the order of values.

If indices is given only the parameters with specified indices are included in the output.

pyopus.evaluator.auxfunc.paramDict(inputList, nameList, indices=None)

Converts a list of parameter values into a dictionary. nameList specifies the keys corresponding to values in the list.

If indices is given only the parameters with specified indices are included in the output.

Previous topic

2.3. pyopus.evaluator.aggregate — Performance aggregation

Next topic

3. pyopus.parallel — Parallel processing support

This Page