Raw post-processing routines.
This module implements class DataSet, unit element in the
post-processing and class DataSetList, sequence of instances
of DataSet.
Futhermore it implements methods for dealing with a third data structure
which is a dictionary of DataSetList which is handy when
dealing with DataSetList instances from multiple algorithms
for comparisons.
| Class | |
Unit element for the COCO post-processing. |
| Class | |
List of instances of DataSet. |
| Class | |
Undocumented |
| Class | |
a class instance call returns f-target values based on reference runlengths: |
| Class | |
store and retrieve a list of target function values: |
| Function | align |
Undocumented |
| Function | as |
Undocumented |
| Function | cocofy |
Replaces cocopp references in pickles files with coco_pproc This could become necessary for future backwards compatibility, however rather should become a class method. |
| Function | dict |
Returns a dictionary with problem dimension as key from a dictionary of DataSet lists. |
| Function | dict |
Returns a dictionary with problem dimension as key. |
| Function | dict |
Returns a dictionary with function id as key. |
| Function | dict |
Returns a dictionary with function group as key. |
| Function | dict |
Returns a dictionary with noise group as key. |
| Function | get_ |
try to load pickle file or fall back to DataSetList constructor. |
| Function | parseinfo |
Extract data from a header line in an index entry. |
| Function | parseinfoold |
Deprecated: Extract data from a header line in an index entry. |
| Function | process |
Undocumented |
| Function | process |
Process command line arguments. |
| Function | set_unique_alg |
on return, elements in ds_list do not have an algId attribute value from taken_ids or from ds_list_reference if taken_ids is None. |
| Function | store |
Undocumented |
| Variable | maximal |
Undocumented |
| Variable | targets |
Undocumented |
| Function | _ |
insert a line for each target value, never used (detEvals(targets) does the job on the fly). |
Replaces cocopp references in pickles files with coco_pproc This could become necessary for future backwards compatibility, however rather should become a class method.
Returns a dictionary with problem dimension as key from a dictionary of DataSet lists.
The input argument is a dictionary with algorithm names as
keys and a list of DataSet instances as values.
The resulting dictionary will have dimension as key and as values
dictionaries with algorithm names as keys.
Returns a dictionary with problem dimension as key.
The difference with dictAlgByDim is that there is an
entry for each algorithm even if the resulting
DataSetList is empty.
This function is meant to be used with an input argument which is a
dictionary with algorithm names as keys and which has list of
DataSet instances as values.
The resulting dictionary will have dimension as key and as values
dictionaries with algorithm names as keys.
Returns a dictionary with function id as key.
This method is meant to be used with an input argument which is a
dictionary with algorithm names as keys and which has list of
DataSet instances as values.
The resulting dictionary will have function id as key and as values
dictionaries with algorithm names as keys.
Returns a dictionary with function group as key.
This method is meant to be used with an input argument which is a
dictionary with algorithm names as keys and which has list of
DataSet instances as values.
The resulting dictionary will have a string denoting the function
group and as values dictionaries with algorithm names as keys.
Returns a dictionary with noise group as key.
This method is meant to be used with an input argument which is a
dictionary with algorithm names as keys and which has list of
DataSet instances as values.
The resulting dictionary will have a string denoting the noise group
('noiselessall' or 'nzall') and as values dictionaries with
algorithm names as keys.
try to load pickle file or fall back to DataSetList constructor.
Also write pickle file if reading failed. Global side effect:
testbedsettings.load_current_testbed is called as it is in
DataSet.__init__.
args[0] is expected to be either a list with one element which is a
repository filetype name or the name itself. Otherwise, the fallback is
executed.
Extract data from a header line in an index entry.
Use a 'smarter' regular expression than parseinfoold.
The header line should be a string of comma-separated pairs of
key=value, for instance: key = value, key = 'value'
Keys should not use comma or quote characters.
Deprecated: Extract data from a header line in an index entry.
Older but verified version of parseinfo
The header line should be a string of comma-separated pairs of key=value, for instance: key = value, key = 'value'
Keys should not use comma or quote characters.
Process command line arguments.
Returns several instances of DataSetList, and a list of
algorithms from a list of strings representing file and folder names,
see below for details. This command operates folder-wise: one folder
corresponds to one algorithm.
It is recommended that if a folder listed in args contain both :file:`info` files and the associated :file:`pickle` files, they be kept in different locations for efficiency reasons.
| Parameters | |
| args | Undocumented |
| process | Undocumented |
| list args | string arguments for folder names |
| bool process | option to process also background algorithms |
| Returns | |
| |
on return, elements in ds_list do not have an algId attribute value from taken_ids or from ds_list_reference if taken_ids is None.
In case, BFGS becomes BFGS 2 etc.