def testAccessors(self): """Test accessors.""" m1 = self.m p1 = Parameter("p1", 1) m1._addObject(p1, m1._parameters) m2 = RecipeContainer("m2") p2 = Parameter("p2", 2) m2._addObject(p2, m2._parameters) m1._addObject(m2, m1._containers) self.assertTrue(m1.m2 is m2) self.assertTrue(m1.p1 is p1) self.assertTrue(m2.p2 is p2) self.assertTrue(m1.m2.p2 is p2) self.assertTrue(m1[0] is p1) self.assertTrue(m1[0:] == [ p1, ]) self.assertTrue(m2[0] is p2) self.assertEqual(1, len(m1)) self.assertEqual(1, len(m2)) return
def testGetRestraints(self): """Test the _getRestraints method.""" m2 = RecipeOrganizer("m2") self.m._organizers = {} self.m._manage(self.m._organizers) self.m._addObject(m2, self.m._organizers) p1 = Parameter("p1", 1) p2 = Parameter("p2", 2) p3 = Parameter("p3", 3) p4 = Parameter("p4", 4) self.m._addParameter(p1) self.m._addParameter(p2) m2._addParameter(p3) m2._addParameter(p4) r1 = self.m.restrain("p1 + p2") r2 = m2.restrain("2*p3 + p4") res = self.m._getRestraints() self.assertTrue(r1 in res) self.assertTrue(r2 in res) self.assertEqual(2, len(res)) return
def testGetConstraints(self): """Test the _getConstraints method.""" m2 = RecipeOrganizer("m2") self.m._organizers = {} self.m._manage(self.m._organizers) self.m._addObject(m2, self.m._organizers) p1 = Parameter("p1", 1) p2 = Parameter("p2", 2) p3 = Parameter("p3", 3) p4 = Parameter("p4", 4) self.m._addParameter(p1) self.m._addParameter(p2) m2._addParameter(p3) m2._addParameter(p4) self.m.constrain(p1, "p2") m2.constrain(p3, "p4") cons = self.m._getConstraints() self.assertTrue(p1 in cons) self.assertTrue(p3 in cons) self.assertEqual(2, len(cons)) return
def testRestrain(self): """Test the restrain method.""" p1 = Parameter("p1", 1) p2 = Parameter("p2", 2) p3 = Parameter("p3", 3) self.m._eqfactory.registerArgument("p1", p1) self.m._eqfactory.registerArgument("p2", p2) self.assertEqual(0, len(self.m._restraints)) r = self.m.restrain("p1+p2", ub=10) self.assertEqual(1, len(self.m._restraints)) p2.setValue(10) self.assertEqual(1, r.penalty()) self.m.unrestrain(r) self.assertEqual(0, len(self.m._restraints)) r = self.m.restrain(p1, ub=10) self.assertEqual(1, len(self.m._restraints)) p1.setValue(11) self.assertEqual(1, r.penalty()) # Check errors on unregistered parameters self.assertRaises(ValueError, self.m.restrain, "2*p3") self.assertRaises(ValueError, self.m.restrain, "2*p2", ns={"p2": p3}) return
def bound_range(variable: Parameter, bound: tp.Union[tp.Tuple, tp.Dict], ratio: bool = False) -> Parameter: """Bound variable by range.""" value = variable.getValue() if isinstance(bound, dict): if ratio: for k, r in bound.items(): bound[k] = value * r variable.boundRange(**bound) else: if ratio: bound = tuple((r * value for r in bound)) variable.boundRange(*bound) return variable
def __init__(self, name, model, parname=None): """Create the Parameter. name -- Name of the Parameter model -- The BaseModel to which the underlying parameter belongs parname -- Name of parameter used by the model. If this is None (default), then name is used. """ self._parname = parname or name val = model.getParam(self._parname) self._model = model Parameter.__init__(self, name, val) return
def __init__(self, name, model, parname = None): """Create the Parameter. name -- Name of the Parameter model -- The BaseModel to which the underlying parameter belongs parname -- Name of parameter used by the model. If this is None (default), then name is used. """ self._parname = parname or name val = model.getParam(self._parname) self._model = model Parameter.__init__(self, name, val) return
def testAddParameterSet(self): """Test the addParameterSet method.""" parset2 = ParameterSet("parset2") p1 = Parameter("parset2", 1) self.parset.addParameterSet(parset2) self.assertTrue(self.parset.parset2 is parset2) self.assertRaises(ValueError, self.parset.addParameterSet, p1) p1.name = "p1" parset2.addParameter(p1) self.assertTrue(self.parset.parset2.p1 is p1) return
def __init__(self): """Initialize the attributes.""" Observable.__init__(self) self._xobs = None self._yobs = None self._dyobs = None self.xpar = Parameter("x") self.ypar = Parameter("y") self.dypar = Parameter("dy") self.ycpar = Parameter("ycalc") self.meta = {} # Observable self.xpar.addObserver(self._flush) self.ypar.addObserver(self._flush) self.dypar.addObserver(self._flush) return
def testEquationFromString(self): """Test the equationFromString method.""" p1 = Parameter("p1", 1) p2 = Parameter("p2", 2) p3 = Parameter("p3", 3) p4 = Parameter("p4", 4) factory = EquationFactory() factory.registerArgument("p1", p1) factory.registerArgument("p2", p2) # Check usage where all parameters are registered with the factory eq = equationFromString("p1+p2", factory) self.assertEqual(2, len(eq.args)) self.assertTrue(p1 in eq.args) self.assertTrue(p2 in eq.args) self.assertEqual(3, eq()) # Try to use a parameter that is not registered self.assertRaises(ValueError, equationFromString, "p1+p2+p3", factory) # Pass that argument in the ns dictionary eq = equationFromString("p1+p2+p3", factory, {"p3": p3}) self.assertEqual(3, len(eq.args)) self.assertTrue(p1 in eq.args) self.assertTrue(p2 in eq.args) self.assertTrue(p3 in eq.args) self.assertEqual(6, eq()) # Make sure that there are no remnants of p3 in the factory self.assertTrue("p3" not in factory.builders) # Pass and use an unregistered parameter self.assertRaises(ValueError, equationFromString, "p1+p2+p3+p4", factory, {"p3": p3}) # Try to overload a registered parameter self.assertRaises(ValueError, equationFromString, "p1+p2", factory, {"p2": p4}) return
def _newParameter(self, name, value, check=True): """Add a new Parameter to the container. This creates a new Parameter and adds it to the container using the _addParameter method. Returns the Parameter. """ p = Parameter(name, value) self._addParameter(p, check) return p
def testProxy(self): """Test the ParameterProxy class.""" l = Parameter("l", 3.14) # Try Accessor adaptation la = ParameterProxy("l2", l) self.assertEqual("l2", la.name) self.assertEqual(l.getValue(), la.getValue()) # Change the parameter l.value = 2.3 self.assertEqual(l.getValue(), la.getValue()) self.assertEqual(l.value, la.value) # Change the proxy la.value = 3.2 self.assertEqual(l.getValue(), la.getValue()) self.assertEqual(l.value, la.value) return
def testRestrain(self): """Test the restrain method.""" p1 = Parameter("p1", 1) p2 = Parameter("p2", 2) p3 = Parameter("p3", 3) self.m._eqfactory.registerArgument("p1", p1) self.m._eqfactory.registerArgument("p2", p2) self.assertEquals(0, len(self.m._restraints)) r = self.m.restrain("p1+p2", ub = 10) self.assertEquals(1, len(self.m._restraints)) p2.setValue(10) self.assertEquals(1, r.penalty()) self.m.unrestrain(r) self.assertEquals(0, len(self.m._restraints)) r = self.m.restrain(p1, ub = 10) self.assertEquals(1, len(self.m._restraints)) p1.setValue(11) self.assertEquals(1, r.penalty()) # Check errors on unregistered parameters self.assertRaises(ValueError, self.m.restrain, "2*p3") self.assertRaises(ValueError, self.m.restrain, "2*p2", ns = {"p2":p3}) return
def testLocateManagedObject(self): """Test the locateManagedObject method.""" m1 = self.m p1 = Parameter("p1", 1) m1._addObject(p1, m1._parameters) m2 = RecipeContainer("m2") p2 = Parameter("p2", 2) m2._addObject(p2, m2._parameters) m1._addObject(m2, m1._containers) p3 = Parameter("p3", 3) # Locate m2 in m1 (m1.m2) loc = m1._locateManagedObject(m2) self.assertEqual(loc, [m1, m2]) # Locate p1 (m1.p1) loc = m1._locateManagedObject(p1) self.assertEqual(loc, [m1, p1]) # Locate p2 in m2 (m2.p2) loc = m2._locateManagedObject(p2) self.assertEqual(loc, [m2, p2]) # Locate p2 in m1 (m1.m2.p2) loc = m1._locateManagedObject(p2) self.assertEqual(loc, [m1, m2, p2]) # Locate p3 in m1 (not there) loc = m1._locateManagedObject(p3) self.assertEqual(loc, []) # Locate p3 in m2 (not there) loc = m2._locateManagedObject(p3) self.assertEqual(loc, []) return
def testAddParameter(self): """Test the addParameter method.""" m = self.m p1 = Parameter("p1", 1) p2 = Parameter("p1", 2) # Check normal insert m._addParameter(p1) self.assertTrue(m.p1 is p1) self.assertTrue(p1.name in m._eqfactory.builders) # Try to insert another parameter with the same name self.assertRaises(ValueError, m._addParameter, p2) # Now allow this m._addParameter(p2, check=False) self.assertTrue(m.p1 is p2) self.assertTrue(p1.name in m._eqfactory.builders) # Try to insert a Parameter when a RecipeContainer with the same name # is already inside. c = RecipeContainer("test") m._addObject(c, m._containers) p3 = Parameter("test", 0) self.assertRaises(ValueError, m._addParameter, p3) p4 = Parameter("xyz", 0) m._addParameter(p4) # Check order self.assertEqual(m._parameters.keys(), ["p1", "xyz"]) self.assertEqual(m._parameters.values(), [p2, p4]) return
def testNewParameter(self): """Test the addParameter method.""" m = self.m p1 = Parameter("p1", 1) m._addParameter(p1) # Test duplication of Parameters self.assertRaises(ValueError, m._newParameter, "p1", 0) # Add a new Parameter p2 = m._newParameter("p2", 0) self.assertTrue(p2 is m.p2) return
def testRemoveParameter(self): """Test removeParameter method.""" m = self.m p1 = Parameter("p1", 1) p2 = Parameter("p1", 2) m._addParameter(p1) # Check for bad remove self.assertRaises(ValueError, m._removeParameter, p2) # Remove p1 m._removeParameter(p1) self.assertTrue(p1.name not in m._eqfactory.builders) # Try to remove it again self.assertRaises(ValueError, m._removeParameter, p1) # Try to remove a RecipeContainer c = RecipeContainer("test") self.assertRaises(ValueError, m._removeParameter, c) return
def test___call__(self): """check WeakBoundMethod.__call__() """ f = self.f self.assertEqual(7, f.evaluate()) self.assertEqual(7, f._eq._value) # verify f has the same effect as f._eq._flush self.w(()) self.assertTrue(None is f._eq._value) # check WeakBoundMethod behavior with no fallback x = Parameter('x', value=3) wgetx = weak_ref(x.getValue) self.assertEqual(3, wgetx()) del x self.assertRaises(ReferenceError, wgetx) return
def testVars(self): """Test to see if variables are added and removed properly.""" recipe = self.recipe con = self.fitcontribution recipe.addVar(con.A, 2) recipe.addVar(con.k, 1) recipe.addVar(con.c, 0) recipe.newVar("B", 0) names = recipe.getNames() self.assertEquals(names, ["A", "k", "c", "B"]) values = recipe.getValues() self.assertTrue((values == [2, 1, 0, 0]).all()) # Constrain a parameter to the B-variable to give it a value p = Parameter("Bpar", -1) recipe.constrain(recipe.B, p) values = recipe.getValues() self.assertTrue((values == [2, 1, 0]).all()) recipe.delVar(recipe.B) recipe.fix(recipe.k) names = recipe.getNames() self.assertEquals(names, ["A", "c"]) values = recipe.getValues() self.assertTrue((values == [2, 0]).all()) recipe.fix("all") names = recipe.getNames() self.assertEquals(names, []) values = recipe.getValues() self.assertTrue((values == []).all()) recipe.free("all") names = recipe.getNames() self.assertEquals(3, len(names)) self.assertTrue("A" in names) self.assertTrue("k" in names) self.assertTrue("c" in names) values = recipe.getValues() self.assertEquals(3, len(values)) self.assertTrue(0 in values) self.assertTrue(1 in values) self.assertTrue(2 in values) return
def bound_window( variable: Parameter, bound: tp.Union[float, tp.Tuple, tp.Dict], ratio: bool = False ) -> Parameter: """Bound variable by window.""" value = variable.getValue() if isinstance(bound, dict): if ratio: for k, r in bound.items(): bound[k] = value * r variable.boundWindow(**bound) elif isinstance(bound, float): if ratio: bound = bound * value variable.boundWindow(bound) else: if ratio: bound = tuple((r * value for r in bound)) variable.boundWindow(*bound) return variable
def testConstrain(self): """Test the constrain method.""" p1 = self.m._newParameter("p1", 1) p2 = self.m._newParameter("p2", 2) p3 = Parameter("p3", 3) self.assertFalse(p1.constrained) self.assertEqual(0, len(self.m._constraints)) self.m.constrain(p1, "2*p2") self.assertTrue(p1.constrained) self.assertTrue(p1 in self.m._constraints) self.assertEqual(1, len(self.m._constraints)) self.assertTrue(self.m.isConstrained(p1)) p2.setValue(10) self.m._constraints[p1].update() self.assertEqual(20, p1.getValue()) # Check errors on unregistered parameters self.assertRaises(ValueError, self.m.constrain, p1, "2*p3") self.assertRaises(ValueError, self.m.constrain, p1, "2*p2", {"p2": p3}) # Remove the constraint self.m.unconstrain(p1) self.assertFalse(p1.constrained) self.assertEqual(0, len(self.m._constraints)) self.assertFalse(self.m.isConstrained(p1)) # Try an straight constraint self.m.constrain(p1, p2) p2.setValue(7) self.m._constraints[p1].update() self.assertEqual(7, p1.getValue()) return
def testConstraint(self): """Test the Constraint class.""" p1 = Parameter("p1", 1) p2 = Parameter("p2", 2) factory = EquationFactory() factory.registerArgument("p1", p1) factory.registerArgument("p2", p2) c = Constraint() # Constrain p1 = 2*p2 eq = equationFromString("2*p2", factory) c.constrain(p1, eq) self.assertTrue(p1.constrained) self.assertFalse(p2.constrained) eq2 = equationFromString("2*p2+1", factory) c2 = Constraint() self.assertRaises(ValueError, c2.constrain, p1, eq2) p2.setConst() eq3 = equationFromString("p1", factory) self.assertRaises(ValueError, c2.constrain, p2, eq3) p2.setValue(2.5) c.update() self.assertEquals(5.0, p1.getValue()) p2.setValue(8.1) self.assertEquals(5.0, p1.getValue()) c.update() self.assertEquals(16.2, p1.getValue()) return
def testSetValue(self): """Test initialization.""" l = Parameter("l") l.setValue(3.14) self.assertAlmostEqual(3.14, l.getValue()) # Try array import numpy x = numpy.arange(0, 10, 0.1) l.setValue(x) self.assertTrue(l.getValue() is x) self.assertTrue(l.value is x) # Change the array y = numpy.arange(0, 10, 0.5) l.value = y self.assertTrue(l.getValue() is y) self.assertTrue(l.value is y) # Back to scalar l.setValue(1.01) self.assertAlmostEqual(1.01, l.getValue()) self.assertAlmostEqual(1.01, l.value) return
class Profile(Observable, Validatable): """Observed and calculated profile container. Profile is an Observable. The xpar, ypar and dypar attributes are observed by the Profile, which can in turn be observed by some other object. Attributes _xobs -- A numpy array of the observed independent variable (default None) xobs -- Read-only property of _xobs. _yobs -- A numpy array of the observed signal (default None) yobs -- Read-only property of _yobs. _dyobs -- A numpy array of the uncertainty of the observed signal (default None, optional). dyobs -- Read-only property of _dyobs. x -- A numpy array of the calculated independent variable (default None, property for xpar accessors). y -- The profile over the calculation range (default None, property for ypar accessors). dy -- The uncertainty in the profile over the calculation range (default None, property for dypar accessors). ycalc -- A numpy array of the calculated signal (default None). xpar -- A Parameter that stores x (named "x"). ypar -- A Parameter that stores y (named "y"). dypar -- A Parameter that stores dy (named "dy"). ycpar -- A Parameter that stores ycalc (named "ycalc"). This is not observed by the profile, but it is present so it can be constrained to. meta -- A dictionary of metadata. This is only set if provided by a parser. """ def __init__(self): """Initialize the attributes.""" Observable.__init__(self) self._xobs = None self._yobs = None self._dyobs = None self.xpar = Parameter("x") self.ypar = Parameter("y") self.dypar = Parameter("dy") self.ycpar = Parameter("ycalc") self.meta = {} # Observable self.xpar.addObserver(self._flush) self.ypar.addObserver(self._flush) self.dypar.addObserver(self._flush) return # We want x, y, ycalc and dy to stay in-sync with xpar, ypar and dypar x = property( lambda self : self.xpar.getValue(), lambda self, val : self.xpar.setValue(val) ) y = property( lambda self : self.ypar.getValue(), lambda self, val : self.ypar.setValue(val) ) dy = property( lambda self : self.dypar.getValue(), lambda self, val : self.dypar.setValue(val) ) ycalc = property( lambda self : self.ycpar.getValue(), lambda self, val : self.ycpar.setValue(val) ) # We want xobs, yobs and dyobs to be read-only xobs = property( lambda self: self._xobs ) yobs = property( lambda self: self._yobs ) dyobs = property( lambda self: self._dyobs ) def loadParsedData(self, parser): """Load parsed data from a ProfileParser. This sets the xobs, yobs, dyobs arrays as well as the metadata. """ x, y, junk, dy = parser.getData() self.meta = dict(parser.getMetaData()) self.setObservedProfile(x, y, dy) return def setObservedProfile(self, xobs, yobs, dyobs = None): """Set the observed profile. Arguments xobs -- Numpy array of the independent variable yobs -- Numpy array of the observed signal. dyobs -- Numpy array of the uncertainty in the observed signal. If dyobs is None (default), it will be set to 1 at each observed xobs. Raises ValueError if len(yobs) != len(xobs) Raises ValueError if dyobs != None and len(dyobs) != len(xobs) """ if len(yobs) != len(xobs): raise ValueError("xobs and yobs are different lengths") if dyobs is not None and len(dyobs) != len(xobs): raise ValueError("xobs and dyobs are different lengths") self._xobs = numpy.asarray(xobs, dtype=float) self._yobs = numpy.asarray(yobs, dtype=float) if dyobs is None: self._dyobs = numpy.ones_like(xobs) else: self._dyobs = numpy.asarray(dyobs, dtype=float) # Set the default calculation points if self.x is None: self.setCalculationPoints(self._xobs) else: self.setCalculationPoints(self.x) return def setCalculationRange(self, xmin = None, xmax = None, dx = None): """Set the calculation range Arguments xmin -- The minimum value of the independent variable. If xmin is None (default), the minimum observed value will be used. This is clipped to the minimum observed x. xmax -- The maximum value of the independent variable. If xmax is None (default), the maximum observed value will be used. This is clipped to the maximum observed x. dx -- The sample spacing in the independent variable. If dx is None (default), then the spacing in the observed points will be preserved. Note that xmin is always inclusive (unless clipped). xmax is inclusive if it is within the bounds of the observed data. raises AttributeError if there is no observed profile raises ValueError if xmin > xmax raises ValueError if dx > xmax-xmin raises ValueError if dx <= 0 """ clip = dx is None if self.xobs is None: raise AttributeError("No observed profile") if xmin is None and xmax is None and dx is None: self.x = self.xobs self.y = self.yobs self.dy = self.dyobs return if xmin is None: xmin = self.xobs[0] else: xmin = float(xmin) if xmax is None: xmax = self.xobs[-1] else: xmax = float(xmax) if dx is None: dx = (self.xobs[-1] - self.xobs[0]) / len(self.xobs) else: dx = float(dx) if xmin > xmax: raise ValueError("xmax must be greater than xmin") if dx > xmax - xmin: raise ValueError("dx must be less than xmax-xmin") if dx <= 0: raise ValueError("dx must be positive") if clip: x = self.xobs indices = numpy.logical_and( xmin - epsilon <= x , x <= xmax + epsilon ) self.x = self.xobs[indices] self.y = self.yobs[indices] self.dy = self.dyobs[indices] else: self.setCalculationPoints(numpy.arange(xmin, xmax+0.5*dx, dx)) return def setCalculationPoints(self, x): """Set the calculation points. Arguments x -- A non-empty numpy array containing the calculation points. If xobs exists, the bounds of x will be limited to its bounds. This will create y and dy on the specified grid if xobs, yobs and dyobs exist. """ x = numpy.asarray(x) if self.xobs is not None: x = x[ x >= self.xobs[0] - epsilon ] x = x[ x <= self.xobs[-1] + epsilon ] self.x = x if self.yobs is not None: self.y = rebinArray(self.yobs, self.xobs, self.x) if self.dyobs is not None: # work around for interpolation issue making some of these non-1 if (self.dyobs == 1).all(): self.dy = numpy.ones_like(self.x) else: # FIXME - This does not follow error propogation rules and it # introduces (more) correlation between the data points. self.dy = rebinArray(self.dyobs, self.xobs, self.x) return def loadtxt(self, *args, **kw): """Use numpy.loadtxt to load data. Arguments are passed to numpy.loadtxt. unpack = True is enforced. The first two arrays returned by numpy.loadtxt are assumed to be x and y. If there is a third array, it is assumed to by dy. Any other arrays are ignored. These are passed to setObservedProfile. Raises ValueError if the call to numpy.loadtxt returns fewer than 2 arrays. Returns the x, y and dy arrays loaded from the file """ if len(args) == 8 and not args[-1]: args = list(args) args[-1] = True else: kw["unpack"] = True cols = numpy.loadtxt(*args, **kw) x = y = dy = None # Due to using 'unpack', a single column will come out as a single # array, thus the second check. if len(cols) < 2 or not isinstance(cols[0], numpy.ndarray): raise ValueError("numpy.loadtxt returned fewer than 2 arrays") x = cols[0] y = cols[1] if len(cols) > 2: dy = cols[2] self.setObservedProfile(x, y, dy) return x, y, dy def savetxt(self, fname, fmt='%.18e', delimiter=' '): """Call numpy.savetxt with x, ycalc, y, dy Arguments are passed to numpy.savetxt. """ x = self.x ycalc = self.ycalc if ycalc is None: raise AttributeError("ycalc is None") y = self.y dy = self.dy # Add the header if not hasattr(fname, 'write'): fname = file(fname, 'w') if fname.closed: raise ValueError("I/O operation on closed file") header = "# x ycalc y dy\n" fname.write(header) numpy.savetxt(fname, zip(x, ycalc, y, dy), fmt, delimiter) return def _flush(self, other): """Invalidate cached state. This will force any observer to invalidate its state. """ self.ycalc = None self.notify(other) return def _validate(self): """Validate my state. This validates that x, y, dy, xobx, yobs and dyobs are not None. This validates that x, y, and dy are the same length. Raises SrFitError if validation fails. """ datanotset = any(v is None for v in [self.x, self.y, self.dy, self.xobs, self.yobs, self.dyobs]) if datanotset: raise SrFitError("Missing data") if len(self.x) != len(self.y) or len(self.x) != len(self.dy): raise SrFitError("Data are different lengths") return
def testWrapper(self): """Test the adapter. This adapts a Parameter to the Parameter interface. :) """ l = Parameter("l", 3.14) # Try Accessor adaptation la = ParameterAdapter("l", l, getter=Parameter.getValue, setter=Parameter.setValue) self.assertEqual(l.name, la.name) self.assertEqual(l.getValue(), la.getValue()) # Change the parameter l.setValue(2.3) self.assertEqual(l.getValue(), la.getValue()) # Change the adapter la.setValue(3.2) self.assertEqual(l.getValue(), la.getValue()) # Try Attribute adaptation la = ParameterAdapter("l", l, attr="value") self.assertEqual(l.name, la.name) self.assertEqual("value", la.attr) self.assertEqual(l.getValue(), la.getValue()) # Change the parameter l.setValue(2.3) self.assertEqual(l.getValue(), la.getValue()) # Change the adapter la.setValue(3.2) self.assertEqual(l.getValue(), la.getValue()) return
def testSetValue(self): """Test initialization.""" l = Parameter("l") l.setValue(3.14) self.assertAlmostEqual(3.14, l.getValue()) # Try array import numpy x = numpy.arange(0, 10, 0.1) l.setValue(x) self.assertTrue( l.getValue() is x ) self.assertTrue( l.value is x ) # Change the array y = numpy.arange(0, 10, 0.5) l.value = y self.assertTrue( l.getValue() is y ) self.assertTrue( l.value is y ) # Back to scalar l.setValue(1.01) self.assertAlmostEqual(1.01, l.getValue()) self.assertAlmostEqual(1.01, l.value) return
def testRestraint(self): """Test the Restraint class.""" p1 = Parameter("p1", 1) p2 = Parameter("p2", 2) factory = EquationFactory() factory.registerArgument("p1", p1) factory.registerArgument("p2", p2) # Restrain 1 < p1 + p2 < 5 eq = equationFromString("p1 + p2", factory) r = Restraint(eq, 1, 5) # This should have no penalty p1.setValue(1) p2.setValue(1) self.assertEquals(0, r.penalty()) # Make p1 + p2 = 0 # This should have a penalty of 1*(1 - 0)**2 = 1 p1.setValue(-1) p2.setValue(1) self.assertEquals(1, r.penalty()) # Make p1 + p2 = 8 # This should have a penalty of 1*(8 - 5)**2 = 9 p1.setValue(4) p2.setValue(4) self.assertEquals(9, r.penalty()) # Set the chi^2 to get a dynamic penalty r.scaled = True self.assertEquals(13.5, r.penalty(1.5)) # Make a really large number to check the upper bound import numpy r.ub = numpy.inf p1.setValue(1e100) self.assertEquals(0, r.penalty()) return
class Profile(Observable, Validatable): """Observed and calculated profile container. Profile is an Observable. The xpar, ypar and dypar attributes are observed by the Profile, which can in turn be observed by some other object. Attributes _xobs -- A numpy array of the observed independent variable (default None) xobs -- Read-only property of _xobs. _yobs -- A numpy array of the observed signal (default None) yobs -- Read-only property of _yobs. _dyobs -- A numpy array of the uncertainty of the observed signal (default None, optional). dyobs -- Read-only property of _dyobs. x -- A numpy array of the calculated independent variable (default None, property for xpar accessors). y -- The profile over the calculation range (default None, property for ypar accessors). dy -- The uncertainty in the profile over the calculation range (default None, property for dypar accessors). ycalc -- A numpy array of the calculated signal (default None). xpar -- A Parameter that stores x (named "x"). ypar -- A Parameter that stores y (named "y"). dypar -- A Parameter that stores dy (named "dy"). ycpar -- A Parameter that stores ycalc (named "ycalc"). This is not observed by the profile, but it is present so it can be constrained to. meta -- A dictionary of metadata. This is only set if provided by a parser. """ def __init__(self): """Initialize the attributes.""" Observable.__init__(self) self._xobs = None self._yobs = None self._dyobs = None self.xpar = Parameter("x") self.ypar = Parameter("y") self.dypar = Parameter("dy") self.ycpar = Parameter("ycalc") self.meta = {} # Observable self.xpar.addObserver(self._flush) self.ypar.addObserver(self._flush) self.dypar.addObserver(self._flush) return # We want x, y, ycalc and dy to stay in-sync with xpar, ypar and dypar x = property( lambda self : self.xpar.getValue(), lambda self, val : self.xpar.setValue(val) ) y = property( lambda self : self.ypar.getValue(), lambda self, val : self.ypar.setValue(val) ) dy = property( lambda self : self.dypar.getValue(), lambda self, val : self.dypar.setValue(val) ) ycalc = property( lambda self : self.ycpar.getValue(), lambda self, val : self.ycpar.setValue(val) ) # We want xobs, yobs and dyobs to be read-only xobs = property( lambda self: self._xobs ) yobs = property( lambda self: self._yobs ) dyobs = property( lambda self: self._dyobs ) def loadParsedData(self, parser): """Load parsed data from a ProfileParser. This sets the xobs, yobs, dyobs arrays as well as the metadata. """ x, y, junk, dy = parser.getData() self.meta = dict(parser.getMetaData()) self.setObservedProfile(x, y, dy) return def setObservedProfile(self, xobs, yobs, dyobs = None): """Set the observed profile. Arguments xobs -- Numpy array of the independent variable yobs -- Numpy array of the observed signal. dyobs -- Numpy array of the uncertainty in the observed signal. If dyobs is None (default), it will be set to 1 at each observed xobs. Raises ValueError if len(yobs) != len(xobs) Raises ValueError if dyobs != None and len(dyobs) != len(xobs) """ if len(yobs) != len(xobs): raise ValueError("xobs and yobs are different lengths") if dyobs is not None and len(dyobs) != len(xobs): raise ValueError("xobs and dyobs are different lengths") self._xobs = numpy.asarray(xobs, dtype=float) self._yobs = numpy.asarray(yobs, dtype=float) if dyobs is None: self._dyobs = numpy.ones_like(xobs) else: self._dyobs = numpy.asarray(dyobs, dtype=float) # Set the default calculation points if self.x is None: self.setCalculationPoints(self._xobs) else: self.setCalculationPoints(self.x) return def setCalculationRange(self, xmin=None, xmax=None, dx=None): """Set epsilon-inclusive calculation range. Adhere to the observed ``xobs`` points when ``dx`` is the same as in the data. ``xmin`` and ``xmax`` are clipped at the bounds of the observed data. Parameters ---------- xmin : float or "obs", optional The minimum value of the independent variable. Keep the current minimum when not specified. If specified as "obs" reset to the minimum observed value. xmax : float or "obs", optional The maximum value of the independent variable. Keep the current maximum when not specified. If specified as "obs" reset to the maximum observed value. dx : float or "obs", optional The sample spacing in the independent variable. When different from the data, resample the ``x`` as anchored at ``xmin``. Note that xmin is always inclusive (unless clipped). xmax is inclusive if it is within the bounds of the observed data. Raises ------ AttributeError If there is no observed data. ValueError When xmin > xmax or if dx <= 0. Also if dx > xmax - xmin. """ if self.xobs is None: raise AttributeError("No observed profile") # local helper function def _isobs(a): if not isinstance(a, six.string_types): return False if a != 'obs': raise ValueError('Must be either float or "obs".') return True # resolve new low and high bounds for x lo = (self.x[0] if xmin is None else self.xobs[0] if _isobs(xmin) else float(xmin)) lo = max(lo, self.xobs[0]) hi = (self.x[-1] if xmax is None else self.xobs[-1] if _isobs(xmax) else float(xmax)) hi = min(hi, self.xobs[-1]) # determine if we need to clip the original grid clip = True step = None ncur = len(self.x) stepcur = (1 if ncur < 2 else (self.x[-1] - self.x[0]) / (ncur - 1.0)) nobs = len(self.xobs) stepobs = (1 if nobs < 2 else (self.xobs[-1] - self.xobs[0]) / (nobs - 1.0)) if dx is None: # check if xobs overlaps with x i0 = numpy.fabs(self.xobs - self.x[0]).argmin() n0 = min(len(self.x), len(self.xobs) - i0) if not numpy.allclose(self.xobs[i0 : i0 + n0], self.x[:n0]): clip = False step = stepcur if ncur > 1 else stepobs elif _isobs(dx): assert clip and step is None elif numpy.allclose(stepobs, dx): assert clip and step is None else: clip = False step = float(dx) # verify that we either clip or have the step defined. assert clip or step is not None # hi, lo, step, clip all resolved here. # validate arguments if lo > hi: raise ValueError("xmax must be greater than xmin.") if not clip: if step > hi - lo: raise ValueError("dx must be less than (xmax - xmin).") if step <= 0: raise ValueError("dx must be positive.") # determine epsilon extensions to the lower and upper bounds. epslo = abs(lo) * epsilon + epsilon epshi = abs(hi) * epsilon + epsilon # process the new grid. if clip: indices = (lo - epslo <= self.xobs) & (self.xobs <= hi + epshi) self.x = self.xobs[indices] self.y = self.yobs[indices] self.dy = self.dyobs[indices] else: x1 = numpy.arange(lo, hi + epshi, step) self.setCalculationPoints(x1) return def setCalculationPoints(self, x): """Set the calculation points. Arguments x -- A non-empty numpy array containing the calculation points. If xobs exists, the bounds of x will be limited to its bounds. This will create y and dy on the specified grid if xobs, yobs and dyobs exist. """ x = numpy.asarray(x) if self.xobs is not None: x = x[ x >= self.xobs[0] - epsilon ] x = x[ x <= self.xobs[-1] + epsilon ] self.x = x if self.yobs is not None: self.y = rebinArray(self.yobs, self.xobs, self.x) if self.dyobs is not None: # work around for interpolation issue making some of these non-1 if (self.dyobs == 1).all(): self.dy = numpy.ones_like(self.x) else: # FIXME - This does not follow error propogation rules and it # introduces (more) correlation between the data points. self.dy = rebinArray(self.dyobs, self.xobs, self.x) return def loadtxt(self, *args, **kw): """Use numpy.loadtxt to load data. Arguments are passed to numpy.loadtxt. unpack = True is enforced. The first two arrays returned by numpy.loadtxt are assumed to be x and y. If there is a third array, it is assumed to by dy. Any other arrays are ignored. These are passed to setObservedProfile. Raises ValueError if the call to numpy.loadtxt returns fewer than 2 arrays. Returns the x, y and dy arrays loaded from the file """ if len(args) == 8 and not args[-1]: args = list(args) args[-1] = True else: kw["unpack"] = True cols = numpy.loadtxt(*args, **kw) x = y = dy = None # Due to using 'unpack', a single column will come out as a single # array, thus the second check. if len(cols) < 2 or not isinstance(cols[0], numpy.ndarray): raise ValueError("numpy.loadtxt returned fewer than 2 arrays") x = cols[0] y = cols[1] if len(cols) > 2: dy = cols[2] self.setObservedProfile(x, y, dy) return x, y, dy def savetxt(self, fname, **kwargs): """Call `numpy.savetxt` with x, ycalc, y, dy. Parameters ---------- fname : filename or file handle This is passed to `numpy.savetxt`. **kwargs The keyword arguments that are passed to `numpy.savetxt`. We preset file header "x ycalc y dy". Use ``header=''`` to save data without any header. Raises ------ SrFitError When `self.ycalc` has not been set. See also -------- numpy.savetxt """ x = self.x ycalc = self.ycalc if ycalc is None: raise SrFitError("ycalc is None") y = self.y dy = self.dy kwargs.setdefault('header', 'x ycalc y dy') data = numpy.transpose([x, ycalc, y, dy]) numpy.savetxt(fname, data, **kwargs) return def _flush(self, other): """Invalidate cached state. This will force any observer to invalidate its state. """ self.ycalc = None self.notify(other) return def _validate(self): """Validate my state. This validates that x, y, dy, xobx, yobs and dyobs are not None. This validates that x, y, and dy are the same length. Raises SrFitError if validation fails. """ datanotset = any(v is None for v in [self.x, self.y, self.dy, self.xobs, self.yobs, self.dyobs]) if datanotset: raise SrFitError("Missing data") if len(self.x) != len(self.y) or len(self.x) != len(self.dy): raise SrFitError("Data are different lengths") return
def testRestraint(self): """Test the Restraint class.""" p1 = Parameter("p1", 1) p2 = Parameter("p2", 2) factory = EquationFactory() factory.registerArgument("p1", p1) factory.registerArgument("p2", p2) # Restrain 1 < p1 + p2 < 5 eq = equationFromString("p1 + p2", factory) r = Restraint(eq, 1, 5) # This should have no penalty p1.setValue(1) p2.setValue(1) self.assertEqual(0, r.penalty()) # Make p1 + p2 = 0 # This should have a penalty of 1*(1 - 0)**2 = 1 p1.setValue(-1) p2.setValue(1) self.assertEqual(1, r.penalty()) # Make p1 + p2 = 8 # This should have a penalty of 1*(8 - 5)**2 = 9 p1.setValue(4) p2.setValue(4) self.assertEqual(9, r.penalty()) # Set the chi^2 to get a dynamic penalty r.scaled = True self.assertEqual(13.5, r.penalty(1.5)) # Make a really large number to check the upper bound import numpy r.ub = numpy.inf p1.setValue(1e100) self.assertEqual(0, r.penalty()) return
def testResidual(self): """Test the residual, which requires all other methods.""" fc = self.fitcontribution profile = self.profile gen = self.gen # Add the calculator and profile fc.setProfile(profile) self.assertTrue(fc.profile is profile) fc.addProfileGenerator(gen, "I") self.assertTrue(fc._eq._value is None) self.assertTrue(fc._reseq._value is None) self.assertEquals(1, len(fc._generators)) self.assertTrue(gen.name in fc._generators) # Let's create some data xobs = arange(0, 10, 0.5) yobs = xobs profile.setObservedProfile(xobs, yobs) # Check our fitting equation. self.assertTrue(array_equal(fc._eq(), gen(xobs))) # Now calculate the residual chiv = fc.residual() self.assertAlmostEqual(0, dot(chiv, chiv)) # Now change the equation fc.setEquation("2*I") self.assertTrue(fc._eq._value is None) self.assertTrue(fc._reseq._value is None) chiv = fc.residual() self.assertAlmostEqual(dot(yobs, yobs), dot(chiv, chiv)) # Try to add a parameter c = Parameter("c", 2) fc._addParameter(c) fc.setEquation("c*I") self.assertTrue(fc._eq._value is None) self.assertTrue(fc._reseq._value is None) chiv = fc.residual() self.assertAlmostEqual(dot(yobs, yobs), dot(chiv, chiv)) # Try something more complex c.setValue(3) fc.setEquation("c**2*sin(I)") self.assertTrue(fc._eq._value is None) self.assertTrue(fc._reseq._value is None) from numpy import sin xobs = arange(0, 10, 0.5) from numpy import sin yobs = 9 * sin(xobs) profile.setObservedProfile(xobs, yobs) self.assertTrue(fc._eq._value is None) self.assertTrue(fc._reseq._value is None) chiv = fc.residual() self.assertAlmostEqual(0, dot(chiv, chiv)) # Choose a new residual. fc.setEquation("2*I") fc.setResidualEquation("resv") chiv = fc.residual() self.assertAlmostEqual(sum((2 * xobs - yobs) ** 2) / sum(yobs ** 2), dot(chiv, chiv)) # Make a custom residual. fc.setResidualEquation("abs(eq-y)**0.5") chiv = fc.residual() self.assertEqual(sum(abs(2 * xobs - yobs)), dot(chiv, chiv)) return
def testResidual(self): """Test the residual, which requires all other methods.""" fc = self.fitcontribution profile = self.profile gen = self.gen # Add the calculator and profile fc.setProfile(profile) self.assertTrue(fc.profile is profile) fc.addProfileGenerator(gen, "I") self.assertTrue(fc._eq._value is None) self.assertTrue(fc._reseq._value is None) self.assertEquals(1, len(fc._generators)) self.assertTrue(gen.name in fc._generators) # Let's create some data xobs = arange(0, 10, 0.5) yobs = xobs profile.setObservedProfile(xobs, yobs) # Check our fitting equation. self.assertTrue(array_equal(fc._eq(), gen(xobs))) # Now calculate the residual chiv = fc.residual() self.assertAlmostEqual(0, dot(chiv, chiv)) # Now change the equation fc.setEquation("2*I") self.assertTrue(fc._eq._value is None) self.assertTrue(fc._reseq._value is None) chiv = fc.residual() self.assertAlmostEqual(dot(yobs, yobs), dot(chiv, chiv)) # Try to add a parameter c = Parameter("c", 2) fc._addParameter(c) fc.setEquation("c*I") self.assertTrue(fc._eq._value is None) self.assertTrue(fc._reseq._value is None) chiv = fc.residual() self.assertAlmostEqual(dot(yobs, yobs), dot(chiv, chiv)) # Try something more complex c.setValue(3) fc.setEquation("c**2*sin(I)") self.assertTrue(fc._eq._value is None) self.assertTrue(fc._reseq._value is None) from numpy import sin xobs = arange(0, 10, 0.5) from numpy import sin yobs = 9*sin(xobs) profile.setObservedProfile(xobs, yobs) self.assertTrue(fc._eq._value is None) self.assertTrue(fc._reseq._value is None) chiv = fc.residual() self.assertAlmostEqual(0, dot(chiv, chiv)) # Choose a new residual. fc.setEquation("2*I") fc.setResidualEquation("resv") chiv = fc.residual() self.assertAlmostEqual(sum((2*xobs-yobs)**2)/sum(yobs**2), dot(chiv, chiv)) # Make a custom residual. fc.setResidualEquation("abs(eq-y)**0.5") chiv = fc.residual() self.assertEqual(sum(abs(2*xobs-yobs)), dot(chiv, chiv)) return
class Profile(Observable, Validatable): """Observed and calculated profile container. Profile is an Observable. The xpar, ypar and dypar attributes are observed by the Profile, which can in turn be observed by some other object. Attributes _xobs -- A numpy array of the observed independent variable (default None) xobs -- Read-only property of _xobs. _yobs -- A numpy array of the observed signal (default None) yobs -- Read-only property of _yobs. _dyobs -- A numpy array of the uncertainty of the observed signal (default None, optional). dyobs -- Read-only property of _dyobs. x -- A numpy array of the calculated independent variable (default None, property for xpar accessors). y -- The profile over the calculation range (default None, property for ypar accessors). dy -- The uncertainty in the profile over the calculation range (default None, property for dypar accessors). ycalc -- A numpy array of the calculated signal (default None). xpar -- A Parameter that stores x (named "x"). ypar -- A Parameter that stores y (named "y"). dypar -- A Parameter that stores dy (named "dy"). ycpar -- A Parameter that stores ycalc (named "ycalc"). This is not observed by the profile, but it is present so it can be constrained to. meta -- A dictionary of metadata. This is only set if provided by a parser. """ def __init__(self): """Initialize the attributes.""" Observable.__init__(self) self._xobs = None self._yobs = None self._dyobs = None self.xpar = Parameter("x") self.ypar = Parameter("y") self.dypar = Parameter("dy") self.ycpar = Parameter("ycalc") self.meta = {} # Observable self.xpar.addObserver(self._flush) self.ypar.addObserver(self._flush) self.dypar.addObserver(self._flush) return # We want x, y, ycalc and dy to stay in-sync with xpar, ypar and dypar x = property(lambda self: self.xpar.getValue(), lambda self, val: self.xpar.setValue(val)) y = property(lambda self: self.ypar.getValue(), lambda self, val: self.ypar.setValue(val)) dy = property(lambda self: self.dypar.getValue(), lambda self, val: self.dypar.setValue(val)) ycalc = property(lambda self: self.ycpar.getValue(), lambda self, val: self.ycpar.setValue(val)) # We want xobs, yobs and dyobs to be read-only xobs = property(lambda self: self._xobs) yobs = property(lambda self: self._yobs) dyobs = property(lambda self: self._dyobs) def loadParsedData(self, parser): """Load parsed data from a ProfileParser. This sets the xobs, yobs, dyobs arrays as well as the metadata. """ x, y, junk, dy = parser.getData() self.meta = dict(parser.getMetaData()) self.setObservedProfile(x, y, dy) return def setObservedProfile(self, xobs, yobs, dyobs=None): """Set the observed profile. Arguments xobs -- Numpy array of the independent variable yobs -- Numpy array of the observed signal. dyobs -- Numpy array of the uncertainty in the observed signal. If dyobs is None (default), it will be set to 1 at each observed xobs. Raises ValueError if len(yobs) != len(xobs) Raises ValueError if dyobs != None and len(dyobs) != len(xobs) """ if len(yobs) != len(xobs): raise ValueError("xobs and yobs are different lengths") if dyobs is not None and len(dyobs) != len(xobs): raise ValueError("xobs and dyobs are different lengths") self._xobs = numpy.asarray(xobs, dtype=float) self._yobs = numpy.asarray(yobs, dtype=float) if dyobs is None: self._dyobs = numpy.ones_like(xobs) else: self._dyobs = numpy.asarray(dyobs, dtype=float) # Set the default calculation points if self.x is None: self.setCalculationPoints(self._xobs) else: self.setCalculationPoints(self.x) return def setCalculationRange(self, xmin=None, xmax=None, dx=None): """Set epsilon-inclusive calculation range. Adhere to the observed ``xobs`` points when ``dx`` is the same as in the data. ``xmin`` and ``xmax`` are clipped at the bounds of the observed data. Parameters ---------- xmin : float or "obs", optional The minimum value of the independent variable. Keep the current minimum when not specified. If specified as "obs" reset to the minimum observed value. xmax : float or "obs", optional The maximum value of the independent variable. Keep the current maximum when not specified. If specified as "obs" reset to the maximum observed value. dx : float or "obs", optional The sample spacing in the independent variable. When different from the data, resample the ``x`` as anchored at ``xmin``. Note that xmin is always inclusive (unless clipped). xmax is inclusive if it is within the bounds of the observed data. Raises ------ AttributeError If there is no observed data. ValueError When xmin > xmax or if dx <= 0. Also if dx > xmax - xmin. """ if self.xobs is None: raise AttributeError("No observed profile") # local helper function def _isobs(a): if not isinstance(a, six.string_types): return False if a != 'obs': raise ValueError('Must be either float or "obs".') return True # resolve new low and high bounds for x lo = (self.x[0] if xmin is None else self.xobs[0] if _isobs(xmin) else float(xmin)) lo = max(lo, self.xobs[0]) hi = (self.x[-1] if xmax is None else self.xobs[-1] if _isobs(xmax) else float(xmax)) hi = min(hi, self.xobs[-1]) # determine if we need to clip the original grid clip = True step = None ncur = len(self.x) stepcur = (1 if ncur < 2 else (self.x[-1] - self.x[0]) / (ncur - 1.0)) nobs = len(self.xobs) stepobs = (1 if nobs < 2 else (self.xobs[-1] - self.xobs[0]) / (nobs - 1.0)) if dx is None: # check if xobs overlaps with x i0 = numpy.fabs(self.xobs - self.x[0]).argmin() n0 = min(len(self.x), len(self.xobs) - i0) if not numpy.allclose(self.xobs[i0:i0 + n0], self.x[:n0]): clip = False step = stepcur if ncur > 1 else stepobs elif _isobs(dx): assert clip and step is None elif numpy.allclose(stepobs, dx): assert clip and step is None else: clip = False step = float(dx) # verify that we either clip or have the step defined. assert clip or step is not None # hi, lo, step, clip all resolved here. # validate arguments if lo > hi: raise ValueError("xmax must be greater than xmin.") if not clip: if step > hi - lo: raise ValueError("dx must be less than (xmax - xmin).") if step <= 0: raise ValueError("dx must be positive.") # determine epsilon extensions to the lower and upper bounds. epslo = abs(lo) * epsilon + epsilon epshi = abs(hi) * epsilon + epsilon # process the new grid. if clip: indices = (lo - epslo <= self.xobs) & (self.xobs <= hi + epshi) self.x = self.xobs[indices] self.y = self.yobs[indices] self.dy = self.dyobs[indices] else: x1 = numpy.arange(lo, hi + epshi, step) self.setCalculationPoints(x1) return def setCalculationPoints(self, x): """Set the calculation points. Arguments x -- A non-empty numpy array containing the calculation points. If xobs exists, the bounds of x will be limited to its bounds. This will create y and dy on the specified grid if xobs, yobs and dyobs exist. """ x = numpy.asarray(x) if self.xobs is not None: x = x[x >= self.xobs[0] - epsilon] x = x[x <= self.xobs[-1] + epsilon] self.x = x if self.yobs is not None: self.y = rebinArray(self.yobs, self.xobs, self.x) if self.dyobs is not None: # work around for interpolation issue making some of these non-1 if (self.dyobs == 1).all(): self.dy = numpy.ones_like(self.x) else: # FIXME - This does not follow error propogation rules and it # introduces (more) correlation between the data points. self.dy = rebinArray(self.dyobs, self.xobs, self.x) return def loadtxt(self, *args, **kw): """Use numpy.loadtxt to load data. Arguments are passed to numpy.loadtxt. unpack = True is enforced. The first two arrays returned by numpy.loadtxt are assumed to be x and y. If there is a third array, it is assumed to by dy. Any other arrays are ignored. These are passed to setObservedProfile. Raises ValueError if the call to numpy.loadtxt returns fewer than 2 arrays. Returns the x, y and dy arrays loaded from the file """ if len(args) == 8 and not args[-1]: args = list(args) args[-1] = True else: kw["unpack"] = True cols = numpy.loadtxt(*args, **kw) x = y = dy = None # Due to using 'unpack', a single column will come out as a single # array, thus the second check. if len(cols) < 2 or not isinstance(cols[0], numpy.ndarray): raise ValueError("numpy.loadtxt returned fewer than 2 arrays") x = cols[0] y = cols[1] if len(cols) > 2: dy = cols[2] self.setObservedProfile(x, y, dy) return x, y, dy def savetxt(self, fname, **kwargs): """Call `numpy.savetxt` with x, ycalc, y, dy. Parameters ---------- fname : filename or file handle This is passed to `numpy.savetxt`. **kwargs The keyword arguments that are passed to `numpy.savetxt`. We preset file header "x ycalc y dy". Use ``header=''`` to save data without any header. Raises ------ SrFitError When `self.ycalc` has not been set. See also -------- numpy.savetxt """ x = self.x ycalc = self.ycalc if ycalc is None: raise SrFitError("ycalc is None") y = self.y dy = self.dy kwargs.setdefault('header', 'x ycalc y dy') data = numpy.transpose([x, ycalc, y, dy]) numpy.savetxt(fname, data, **kwargs) return def _flush(self, other): """Invalidate cached state. This will force any observer to invalidate its state. """ self.ycalc = None self.notify(other) return def _validate(self): """Validate my state. This validates that x, y, dy, xobx, yobs and dyobs are not None. This validates that x, y, and dy are the same length. Raises SrFitError if validation fails. """ datanotset = any( v is None for v in [self.x, self.y, self.dy, self.xobs, self.yobs, self.dyobs]) if datanotset: raise SrFitError("Missing data") if len(self.x) != len(self.y) or len(self.x) != len(self.dy): raise SrFitError("Data are different lengths") return
def testWrapper(self): """Test the adapter. This adapts a Parameter to the Parameter interface. :) """ l = Parameter("l", 3.14) # Try Accessor adaptation la = ParameterAdapter("l", l, getter = Parameter.getValue, setter = Parameter.setValue) self.assertEqual(l.name, la.name) self.assertEqual(l.getValue(), la.getValue()) # Change the parameter l.setValue(2.3) self.assertEqual(l.getValue(), la.getValue()) # Change the adapter la.setValue(3.2) self.assertEqual(l.getValue(), la.getValue()) # Try Attribute adaptation la = ParameterAdapter("l", l, attr = "value") self.assertEqual(l.name, la.name) self.assertEqual("value", la.attr) self.assertEqual(l.getValue(), la.getValue()) # Change the parameter l.setValue(2.3) self.assertEqual(l.getValue(), la.getValue()) # Change the adapter la.setValue(3.2) self.assertEqual(l.getValue(), la.getValue()) return