Source code for PySpice.Probe.WaveForm

####################################################################################################
#
# PySpice - A Spice Package for Python
# Copyright (C) 2014 Fabrice Salvaire
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program.  If not, see <http://www.gnu.org/licenses/>.
#
####################################################################################################

####################################################################################################

"""This module implements classes to handle analysis output.

"""

####################################################################################################

import logging
import os

import numpy as np

####################################################################################################

_module_logger = logging.getLogger(__name__)

####################################################################################################

from PySpice.Unit.Unit import UnitValues

####################################################################################################

[docs]class WaveForm(UnitValues): """This class implements waveform on top of a Numpy Array. Public Attributes: :attr:`name` :attr:`unit` :attr:`title` :attr:`abscissa` Numpy array of the analysis abscissa """ _logger = _module_logger.getChild('WaveForm') ##############################################
[docs] @classmethod def from_unit_values(cls, name, array, title=None, abscissa=None): obj = cls( name, array.prefixed_unit, array.shape, dtype=array.dtype, title=title, abscissa=abscissa, ) obj[...] = array[...] return obj
##############################################
[docs] @classmethod def from_array(cls, name, array, title=None, abscissa=None): # Fixme: ok ??? obj = cls(name, None, array.shape, title=title, abscissa=abscissa) obj[...] = array[...] return obj
############################################## def __new__(cls, name, prefixed_unit, shape, dtype=float, buffer=None, offset=0, strides=None, order=None, title=None, abscissa=None): # cls._logger.info(str((cls, prefixed_unit, shape, dtype, buffer, offset, strides, order))) obj = super(WaveForm, cls).__new__(cls, prefixed_unit, shape, dtype, buffer, offset, strides, order) # obj = np.asarray(data).view(cls) obj._name = str(name) obj._title = title # str(title) obj._abscissa = abscissa return obj ############################################## def __array_finalize__(self, obj): # self._logger.info('\n {}'.format(obj)) # Fixme: ??? else _prefixed_unit is not set super().__array_finalize__(obj) if obj is None: return self._name = getattr(obj, 'name', None) self._title = getattr(obj, 'title', None) self._abscissa = getattr(obj, 'abscissa', None) ############################################## def __array_ufunc__(self, ufunc, method, *inputs, **kwargs): # Fixme: check abscissa result = super().__array_ufunc__(ufunc, method, *inputs, **kwargs) # self._logger.info("result\n{}".format(result)) if isinstance(result, UnitValues): return self.from_unit_values(name='', array=result, title='', abscissa=self._abscissa) else: return result # e.g. foo <= 0 ############################################## @property def name(self): return self._name @property def abscissa(self): return self._abscissa @property def title(self): return self._title @title.setter def title(self, value): self._title = value ############################################## def __repr__(self): return '{0.__class__.__name__} {0._name} {1}'.format(self, super().__str__()) ############################################## def __str__(self): if self._title is not None: return self._title else: return self._name ##############################################
[docs] def str_data(self): # Fixme: ok ??? return repr(self.as_ndarray())
####################################################################################################
[docs]class Analysis: """Base class for the simulation output. Depending of the simulation type, the simulator will return waveforms as a function of * time * frequency * sweep * ... and corresponding to * a node's voltage * a source's current * ... The name of a waveform is * node's voltage: node's name * source's current: source'name * ... If the waveform name is a valid Python identifier, then you can get the corresponding waveform using:: analysis.waveforme_name else you have to use this fallback:: analysis['waveforme_name'] Examples of usages:: # Operating point analysis for node in analysis.nodes.values(): print('Node {}: {:5.2f} V'.format(str(node), float(node))) for node in analysis.branches.values(): print('Node {}: {:5.2f} A'.format(str(node), float(node))) # DC sensitivity analysis for element in analysis.elements.values(): print(element, float(element)) # Plot the voltage of the "out" node plt.plot(analysis.out.abscissa, analysis.out) Public Attributes: :attr:`nodes` Dictionary for node voltages indexed by node names :attr:`branches` Dictionary for branch currents indexed by source names :attr:`elements` Dictionary for elements ... """ ############################################## def __init__(self, simulation, nodes=(), branches=(), elements=(), internal_parameters=()): # Fixme: branches are elements in fact, and elements is not yet supported ... self._simulation = simulation # Fixme: to func? self._nodes = {waveform.name:waveform for waveform in nodes} self._branches = {waveform.name:waveform for waveform in branches} self._elements = {waveform.name:waveform for waveform in elements} self._internal_parameters = {waveform.name:waveform for waveform in internal_parameters} ############################################## @property def simulation(self): """Return the simulation instance""" return self._simulation @property def nodes(self): return self._nodes @property def branches(self): return self._branches @property def elements(self): return self._elements @property def internal_parameters(self): return self._internal_parameters ############################################## def _get_item(self, name): # Fixme: cache dict ??? if name in self._nodes: return self._nodes[name] elif name in self._branches: return self._branches[name] elif name in self._elements: return self._elements[name] elif name in self._internal_parameters: return self._internal_parameters[name] else: raise IndexError(name) ############################################## def __getitem__(self, name): try: return self._get_item(name) except IndexError: return self._get_item(name.lower()) ############################################## @staticmethod def _format_dict(d): return os.linesep.join([' '*2 + str(x) for x in d]) ############################################## def __getattr__(self, name): try: return self.__getitem__(name) except IndexError: raise AttributeError(name + os.linesep + 'Nodes :' + os.linesep + self._format_dict(self._nodes) + os.linesep + 'Branches :' + os.linesep + self._format_dict(self._branches) + os.linesep + 'Elements :' + os.linesep + self._format_dict(self._elements) + os.linesep + 'Internal Parameters :' + os.linesep + self._format_dict(self._internal_parameters) )
####################################################################################################
[docs]class OperatingPoint(Analysis): """This class implements an operating point analysis.""" pass
####################################################################################################
[docs]class SensitivityAnalysis(Analysis): """This class implements an sensitivity analysis.""" ############################################## def __init__(self, simulation, elements, internal_parameters): super().__init__(simulation=simulation, elements=elements, internal_parameters=internal_parameters)
####################################################################################################
[docs]class DcAnalysis(Analysis): """This class implements a DC analysis. When the DC analysis is performed with multiple sources, sweep is the last source. The loop scheme is:: for v1 in vsource1: for v2 in vsource2: ... """ ############################################## def __init__(self, simulation, sweep, nodes, branches, internal_parameters): super().__init__(simulation=simulation, nodes=nodes, branches=branches, internal_parameters=internal_parameters) self._sweep = sweep ############################################## @property def sweep(self): """Return an Numpy array for the sweep abscissa""" return self._sweep
####################################################################################################
[docs]class AcAnalysis(Analysis): """This class implements an AC analysis.""" ############################################## def __init__(self, simulation, frequency, nodes, branches, internal_parameters): super().__init__(simulation=simulation, nodes=nodes, branches=branches, internal_parameters=internal_parameters) self._frequency = frequency ############################################## @property def frequency(self): """Return an Numpy array for the frequency abscissa""" return self._frequency
####################################################################################################
[docs]class TransientAnalysis(Analysis): """This class implements a transient analysis.""" ############################################## def __init__(self, simulation, time, nodes, branches, internal_parameters): super().__init__(simulation=simulation, nodes=nodes, branches=branches, internal_parameters=internal_parameters) self._time = time ############################################## @property def time(self): """Return an Numpy array for the time abscissa""" return self._time