run

class batchpy.run.Run(batch, saveresult=True, **parameters)

A batchpy run base class

This class is intended as a base class from which custom user defined run objects can inherit. In the custom run class the run() method needs to be redefined to actually run the wanted computations and return the result as a dictionary.

Parameters:

batch : Batch() object

The batch the run belongs to

saveresult : boolean, optional

Save the results to disk or not, if not the result is available in the _result attribute

**parameters :

Keyword parameters which modify the run instance

Notes

Batchpy runs should not be created directly but through the batch methods add_run() or add_factorial_runs().

Examples

>>> class Myrun(batchpy.Run):
...     def run(self,mypar=5):
...         # some conplicated computation
...         return {'val': 2*mypar}
...
>>> batch = batchpy.Batch('mybatch')
>>> run = Myrun(batch,saveresult=False,mypar=5)
>>> run()
{'val': 10}
>>> 

run()

Perform calculations and return the result

This method should be overwritten in a user defined child class to perform the actual computations.

Parameters:

parameters

parameters can be defined as named parameters. The use of **kwargs is not supported.

Returns:

res : dict

a dictionary with the results of the run

Examples

>>> class Myrun(batchpy.Run):
...     def run(self,mypar=5):
...         # some conplicated computation
...         return {'val': 2*mypar}
...

__call__()

Checks if the run results are already computed and compute them if not.

When a run is called the class checks if the results are available in memory or on the disk.

When the result is available in memory, it is returned. When it is available on disk, it is loaded and returned.

When the result is not available it is computed using the run method and the results are stored on disk (if the _saveresult attribute is true) or in the _result attribute (otherwise).

The computation is timed and the runtime is saved in the runtime attribute.

Returns:

res : anything

Results.

Examples

>>> batch = batchpy.Batch('mybatch')
>>> run = Myrun(batch,saveresult=False,mypar=5)
>>> run()
{'val': 10}

load()

Checks if the run results are already computed and return them if so.

When the result is available in memory, it is returned. When it is available on disk, it is loaded and returned. When the result is not computed yet this returns None.

Returns:

res : anything, None

Results, returns None if the results are not available.

Examples

>>> run.load()
{'val': 10}

clear()

Tries to erase the run results from the disk

Returns:

success : bool

True if the run was deleted from the disk, False otherwise.

Examples

>>> run.clear()
True

id

Property returning the id.

parameters

Property returning the run parameters.

result

Property alias to self.load().

done

Property returning if the run is done or not.

runtime

Property returning the computation time of a run

index

Property returning the run index in its batch.

filename

Property returning the filename of the run.

generate_id(parameters)

Generates an id hash from the parameters.

The id hash is used to identify a run. It is hashed from the parameters used to create the run to ensure that when even a single parameter is changed the run ids are different. The id is stored in the :code`id` attribute.

Parameters:

parameters : dict

a dictionary with parameters from which to compute the hash.

Returns:

id : string

the id hash of this run

Notes

When classes, methods or functions are supplied as parameters, the hash is created using their name attribute. This avoids ids being different when python is restarted. A hash created from the function itself would be different each time python starts as the object resides in a different memory location.

Examples

>>> run.generate_id(run.parameters)
'10ae24979c5028fa873651bca338152dc0484245'

class batchpy.run.ResultRun(batch, id)

A class to load run results.

Parameters:

batch : Batch() object

The batch the run belongs to.

id : str or list of str

An id or list of ids specifying previously saved runs.

Notes

Batchpy result runs should not be created directly but through the batch methods add_resultrun().

Examples

>>> batch = batchpy.Batch('mybatch')
>>> run = batchpy.ResultRun(batch,'3ecc784a9d5cf26eb6420de2a43f04b310073925')

batchpy.run.convert_run_to_newstyle(run)

Converts a saved run result to include the parameters.

Parameters:

run : Run() object or child class

The run to convert the result from.