"""
Class definition for CaseRecorder, the base class for all recorders.
"""
from openmdao.core.system import System
from openmdao.core.driver import Driver
from openmdao.solvers.solver import Solver
from openmdao.core.problem import Problem
from openmdao.utils.mpi import MPI
from openmdao.utils.options_dictionary import OptionsDictionary
from openmdao.utils.record_util import check_path
# default pickle protocol version for serialization
PICKLE_VER = 4
[docs]class CaseRecorder(object):
"""
Base class for all case recorders and is not a functioning case recorder on its own.
Parameters
----------
record_viewer_data : bool, optional
If True, record data needed for visualization.
Attributes
----------
_record_viewer_data : bool
Flag indicating whether to record data needed to generate N2 diagram.
_counter : int
A global counter for execution order, used in iteration coordinate.
_inputs : dict
System inputs values, post-filtering, to be used by a derived recorder.
_outputs : dict
System or Solver output values, post-filtering, to be used by a derived recorder.
_abs_error : float
Solver abs_error value, to be used by a derived recorder.
_rel_error : float
Solver abs_error value, to be used by a derived recorder.
_iteration_coordinate : str
The unique iteration coordinate of where an iteration originates.
_parallel : bool
Flag indicating if this recorder will record on multiple processes.
_do_gather : bool
Flag indicating if this recorder will gather data from all ranks in the requestor's comm.
_record_on_proc : bool or None
Flag indicating if this recorder will record on the current process (None if unspecified).
_recording_ranks : list
List of ranks on which this recorder will record if running under MPI.
"""
[docs] def __init__(self, record_viewer_data=True):
"""
Initialize.
Parameters
----------
record_viewer_data : bool, optional
If True, record data needed for visualization.
"""
self._record_viewer_data = record_viewer_data
# global counter that is used in iteration coordinate
self._counter = 0
# For Systems
self._inputs = None
self._outputs = None
# For Solvers
self._abs_error = 0.0
self._rel_error = 0.0
# For Drivers, Systems, and Solvers
self._iteration_coordinate = None
# By default, this is False, but it will be set to True if the recorder
# will record data on multiple processes
self._parallel = False
# gather variables from all ranks in the requestor's comm if necessary
self._do_gather = False
# Flag indicating if recording will be performed on the current process.
# If the value is not set to True on any process (the default), then
# recording will be performed only on rank 0.
# If the value is set to True on any process, then the _parallel flag
# will be set and recording will occur on all processes for which the
# value is True.
self._record_on_proc = None
# List of ranks on which this recorder will record if running under MPI.
# Only used when running under MPI with communicator size greater than one.
self._recording_ranks = None
@property
def record_on_process(self):
"""
Determine if recording should be performed on this process.
"""
return self._record_on_proc
@record_on_process.setter
def record_on_process(self, record):
"""
Specify that recording should be performed on this process.
Parameters
----------
record : bool
If True, then recording should be performed on this process.
"""
self._record_on_proc = record
@property
def parallel(self):
"""
Return True if this recorder is recording on multiple processes.
"""
return self._parallel
[docs] def startup(self, recording_requester, comm=None):
"""
Prepare for a new run.
Parameters
----------
recording_requester : object
Object to which this recorder is attached.
comm : MPI.Comm or <FakeComm> or None
The MPI communicator for the recorder (should be the comm for the Problem).
"""
self._counter = 0
if MPI and comm and comm.size > 1:
record_on_ranks = comm.allgather(self._record_on_proc)
recording_ranks = [rnk for rnk, rec in enumerate(record_on_ranks) if rec]
if recording_ranks:
# recording ranks have been specified
self._recording_ranks = recording_ranks
self._parallel = len(recording_ranks) > 1
else:
# default to just record on rank 0
self._record_on_proc = comm.rank == 0
self._recording_ranks = [0]
self._do_gather = len(recording_ranks) < comm.size
def _get_metadata_system(self, system):
"""
Get system metadata.
Parameters
----------
system : System
The System for which to record metadata.
Returns
-------
dict
dictionary of scaling vectors
OptionsDictionary
dictionary with recordable options for system
"""
# Cannot handle PETScVector yet
from openmdao.api import PETScVector
if PETScVector and isinstance(system._outputs, PETScVector):
return None, None # Cannot handle PETScVector yet
# collect scaling arrays
scaling_vecs = {}
for kind, odict in system._vectors.items():
scaling_vecs[kind] = scaling = {}
for vecname, vec in odict.items():
scaling[vecname] = vec._scaling
# create a copy of the system's metadata excluding what is in 'options_excludes'
excludes = system.recording_options['options_excludes']
if excludes:
user_options = OptionsDictionary()
user_options._all_recordable = system.options._all_recordable
for key in system.options._dict:
if check_path(key, [], excludes, True):
user_options._dict[key] = system.options._dict[key]
user_options._read_only = system.options._read_only
return scaling_vecs, user_options
else:
return scaling_vecs, system.options
[docs] def record_iteration(self, recording_requester, data, metadata, **kwargs):
"""
Route the record_iteration call to the proper method.
Parameters
----------
recording_requester : object
System, Solver, Driver in need of recording.
data : dict
Dictionary containing desvars, objectives, constraints, responses, and System vars.
metadata : dict, optional
Dictionary containing execution metadata.
**kwargs : keyword args
Some implementations of record_iteration need additional args.
"""
if not self._parallel or self._record_on_proc:
self._counter += 1
self._iteration_coordinate = \
recording_requester._recording_iter.get_formatted_iteration_coordinate()
if isinstance(recording_requester, Driver):
self.record_iteration_driver(recording_requester, data, metadata)
elif isinstance(recording_requester, System):
self.record_iteration_system(recording_requester, data, metadata)
elif isinstance(recording_requester, Solver):
self.record_iteration_solver(recording_requester, data, metadata)
elif isinstance(recording_requester, Problem):
self.record_iteration_problem(recording_requester, data, metadata)
else:
raise ValueError("Recorders must be attached to Drivers, Systems, or Solvers.")
[docs] def record_iteration_driver(self, recording_requester, data, metadata):
"""
Record data and metadata from a Driver.
Parameters
----------
recording_requester : Driver
Driver in need of recording.
data : dict
Dictionary containing desvars, objectives, constraints, responses, and System vars.
metadata : dict
Dictionary containing execution metadata.
"""
raise NotImplementedError("record_iteration_driver has not been overridden")
[docs] def record_iteration_system(self, recording_requester, data, metadata):
"""
Record data and metadata from a System.
Parameters
----------
recording_requester : System
System in need of recording.
data : dict
Dictionary containing inputs, outputs, and residuals.
metadata : dict
Dictionary containing execution metadata.
"""
raise NotImplementedError("record_iteration_system has not been overridden")
[docs] def record_iteration_solver(self, recording_requester, data, metadata):
"""
Record data and metadata from a Solver.
Parameters
----------
recording_requester : Solver
Solver in need of recording.
data : dict
Dictionary containing outputs, residuals, and errors.
metadata : dict
Dictionary containing execution metadata.
"""
raise NotImplementedError("record_iteration_solver has not been overridden")
[docs] def record_iteration_problem(self, recording_requester, data, metadata):
"""
Record data and metadata from a Problem.
Parameters
----------
recording_requester : Problem
Problem in need of recording.
data : dict
Dictionary containing desvars, objectives, constraints.
metadata : dict
Dictionary containing execution metadata.
"""
raise NotImplementedError("record_iteration_problem has not been overridden")
[docs] def record_derivatives(self, recording_requester, data, metadata, **kwargs):
"""
Route the record_derivatives call to the proper method.
Parameters
----------
recording_requester : object
System, Solver, Driver in need of recording.
data : dict
Dictionary containing derivatives keyed by 'of,wrt' to be recorded.
metadata : dict
Dictionary containing execution metadata.
**kwargs : keyword args
Some implementations of record_derivatives need additional args.
"""
self._iteration_coordinate = \
recording_requester._recording_iter.get_formatted_iteration_coordinate()
self.record_derivatives_driver(recording_requester, data, metadata)
[docs] def record_derivatives_driver(self, recording_requester, data, metadata):
"""
Record derivatives data from a Driver.
Parameters
----------
recording_requester : Driver
Driver in need of recording.
data : dict
Dictionary containing derivatives keyed by 'of,wrt' to be recorded.
metadata : dict
Dictionary containing execution metadata.
"""
raise NotImplementedError("record_derivatives_driver has not been overridden")
[docs] def record_viewer_data(self, model_viewer_data):
"""
Record model viewer data.
Parameters
----------
model_viewer_data : dict
Data required to visualize the model.
"""
raise NotImplementedError("record_viewer_data has not been overridden")
[docs] def shutdown(self):
"""
Shut down the recorder.
"""
pass
