Source code for openmdao.test_suite.components.exec_comp_for_test

from __future__ import print_function

import time
import os

from openmdao.components.exec_comp import ExecComp
from openmdao.core.analysis_error import AnalysisError
from openmdao.utils.mpi import MPI

[docs]class ExecComp4Test(ExecComp): """ A version of ExecComp for benchmarking and testing. Parameters ---------- exprs : str or list of str The expressions that determine the inputs and outputs of this component. nl_delay : float(0.01) The sleep time in seconds that will occur when solve_nonlinear is called. lin_delay : float(0.01) The sleep time in seconds that will occur when apply_linear is called. rec_procs : tuple of the form (minprocs, maxprocs) Minimum and maximun MPI processes usable by this component. fail_rank : int or collection of int (0) Rank (if running under MPI) or worker number (if running under multiprocessing) where failures will be initiated. fails : list or tuple of int If the current self.num_nl_solves matches any of these, then this component will raise an exception. fail_hard : bool(False) If True and fails is not empty, this component will raise a RuntimeError when a failure is induced. Otherwise, an AnalysisError will be raised. """
[docs] def __init__(self, exprs, nl_delay=0.01, lin_delay=0.01, req_procs=(1,1), fail_rank=-1, fails=(), fail_hard=False, **kwargs): super(ExecComp4Test, self).__init__(exprs, **kwargs) self.nl_delay = nl_delay self.lin_delay = lin_delay self.num_nl_solves = 0 self.num_compute_partials = 0 self.req_procs = req_procs self.fail_rank = fail_rank if isinstance(fail_rank, int): self.fail_rank = (self.fail_rank,) self.fails = fails self.fail_hard = fail_hard
[docs] def compute(self, inputs, outputs): """ Execute this component's assignment statements. Parameters ---------- inputs : `Vector` `Vector` containing inputs. outputs : `Vector` `Vector` containing outputs. """ try: if self.comm.rank in self.fail_rank and self.num_nl_solves in self.fails: if self.fail_hard: raise RuntimeError("OMG, a critical error!") else: raise AnalysisError("just an analysis error") super(ExecComp4Test, self).compute(inputs, outputs) time.sleep(self.nl_delay) finally: self.num_nl_solves += 1
[docs] def compute_partials(self, inputs, partials): """ Use complex step method to update the given Jacobian. Parameters ---------- inputs : `Vector` `Vector` containing parameters. (p) partials : `Jacobian` Contains sub-jacobians. """ super(ExecComp4Test, self).compute_partials(inputs, partials) time.sleep(self.lin_delay) self.num_compute_partials += 1