[efa5e44] | 1 | """ |
---|
| 2 | Unit tests for fitting module |
---|
| 3 | @author Gervaise Alina |
---|
| 4 | """ |
---|
| 5 | import unittest |
---|
| 6 | import math |
---|
| 7 | |
---|
| 8 | from sas.fit.AbstractFitEngine import Model, FitHandler |
---|
| 9 | from sas.dataloader.loader import Loader |
---|
[acf8e4a5] | 10 | from sas.fit.BumpsFitting import BumpsFit as Fit |
---|
[efa5e44] | 11 | from sas.models.LineModel import LineModel |
---|
| 12 | from sas.models.Constant import Constant |
---|
| 13 | |
---|
[bc873053] | 14 | from bumps import fitters |
---|
| 15 | try: |
---|
| 16 | from bumps.options import FIT_CONFIG |
---|
| 17 | def set_fitter(alg, opts): |
---|
| 18 | FIT_CONFIG.selected_id = alg |
---|
| 19 | FIT_CONFIG.values[alg].update(opts, monitors=[]) |
---|
| 20 | except: |
---|
| 21 | # CRUFT: Bumps changed its handling of fit options around 0.7.5.6 |
---|
| 22 | def set_fitter(alg, opts): |
---|
| 23 | #print "fitting",alg,opts |
---|
| 24 | #print "options",fitters.FIT_OPTIONS[alg].__dict__ |
---|
| 25 | fitters.FIT_DEFAULT = alg |
---|
| 26 | fitters.FIT_OPTIONS[alg].options.update(opts, monitors=[]) |
---|
| 27 | |
---|
| 28 | |
---|
[efa5e44] | 29 | class testFitModule(unittest.TestCase): |
---|
| 30 | """ test fitting """ |
---|
| 31 | |
---|
| 32 | def test_bad_pars(self): |
---|
[acf8e4a5] | 33 | fitter = Fit() |
---|
[efa5e44] | 34 | |
---|
| 35 | data = Loader().load("testdata_line.txt") |
---|
| 36 | data.name = data.filename |
---|
| 37 | fitter.set_data(data,1) |
---|
| 38 | |
---|
| 39 | model1 = LineModel() |
---|
| 40 | model1.name = "M1" |
---|
| 41 | model = Model(model1, data) |
---|
| 42 | pars1= ['param1','param2'] |
---|
| 43 | try: |
---|
| 44 | fitter.set_model(model,1,pars1) |
---|
| 45 | except ValueError,exc: |
---|
| 46 | #print "ValueError was correctly raised: "+str(msg) |
---|
| 47 | assert str(exc).startswith('parameter param1') |
---|
| 48 | else: |
---|
| 49 | raise AssertionError("No error raised for fitting with wrong parameters name to fit") |
---|
| 50 | |
---|
[acf8e4a5] | 51 | def fit_single(self, isdream=False): |
---|
| 52 | fitter = Fit() |
---|
[efa5e44] | 53 | |
---|
| 54 | data = Loader().load("testdata_line.txt") |
---|
| 55 | data.name = data.filename |
---|
| 56 | fitter.set_data(data,1) |
---|
| 57 | |
---|
| 58 | # Receives the type of model for the fitting |
---|
| 59 | model1 = LineModel() |
---|
| 60 | model1.name = "M1" |
---|
| 61 | model = Model(model1,data) |
---|
| 62 | |
---|
| 63 | pars1= ['A','B'] |
---|
| 64 | fitter.set_model(model,1,pars1) |
---|
| 65 | fitter.select_problem_for_fit(id=1,value=1) |
---|
| 66 | result1, = fitter.fit(handler=FitHandler()) |
---|
| 67 | |
---|
| 68 | # The target values were generated from the following statements |
---|
| 69 | p,s,fx = result1.pvec, result1.stderr, result1.fitness |
---|
| 70 | #print "p0,p1,s0,s1,fx = %g, %g, %g, %g, %g"%(p[0],p[1],s[0],s[1],fx) |
---|
| 71 | p0,p1,s0,s1,fx_ = 3.68353, 2.61004, 0.336186, 0.105244, 1.20189 |
---|
| 72 | |
---|
| 73 | if isdream: |
---|
| 74 | # Dream is not a minimizer: just check that the fit is within |
---|
| 75 | # uncertainty |
---|
| 76 | self.assertTrue( abs(p[0]-p0) <= s0 ) |
---|
| 77 | self.assertTrue( abs(p[1]-p1) <= s1 ) |
---|
| 78 | else: |
---|
| 79 | self.assertTrue( abs(p[0]-p0) <= 1e-5 ) |
---|
| 80 | self.assertTrue( abs(p[1]-p1) <= 1e-5 ) |
---|
| 81 | self.assertTrue( abs(fx-fx_) <= 1e-5 ) |
---|
| 82 | |
---|
| 83 | def fit_bumps(self, alg, **opts): |
---|
[bc873053] | 84 | set_fitter(alg, opts) |
---|
[acf8e4a5] | 85 | self.fit_single(isdream=(alg=='dream')) |
---|
[efa5e44] | 86 | |
---|
| 87 | def test_bumps_de(self): |
---|
| 88 | self.fit_bumps('de') |
---|
| 89 | |
---|
| 90 | def test_bumps_dream(self): |
---|
| 91 | self.fit_bumps('dream', burn=500, steps=100) |
---|
| 92 | |
---|
| 93 | def test_bumps_amoeba(self): |
---|
| 94 | self.fit_bumps('amoeba') |
---|
| 95 | |
---|
| 96 | def test_bumps_newton(self): |
---|
| 97 | self.fit_bumps('newton') |
---|
| 98 | |
---|
| 99 | def test_bumps_lm(self): |
---|
| 100 | self.fit_bumps('lm') |
---|
| 101 | |
---|
| 102 | def test2(self): |
---|
| 103 | """ fit 2 data and 2 model with no constrainst""" |
---|
| 104 | #load data |
---|
| 105 | l = Loader() |
---|
| 106 | data1=l.load("testdata_line.txt") |
---|
| 107 | data1.name = data1.filename |
---|
| 108 | |
---|
| 109 | data2=l.load("testdata_line1.txt") |
---|
| 110 | data2.name = data2.filename |
---|
| 111 | |
---|
| 112 | #Importing the Fit module |
---|
[acf8e4a5] | 113 | fitter = Fit() |
---|
[efa5e44] | 114 | # Receives the type of model for the fitting |
---|
| 115 | model11 = LineModel() |
---|
| 116 | model11.name= "M1" |
---|
| 117 | model22 = LineModel() |
---|
| 118 | model11.name= "M2" |
---|
| 119 | |
---|
| 120 | model1 = Model(model11,data1) |
---|
| 121 | model2 = Model(model22,data2) |
---|
| 122 | pars1= ['A','B'] |
---|
| 123 | fitter.set_data(data1,1) |
---|
| 124 | fitter.set_model(model1,1,pars1) |
---|
| 125 | fitter.select_problem_for_fit(id=1,value=0) |
---|
| 126 | fitter.set_data(data2,2) |
---|
| 127 | fitter.set_model(model2,2,pars1) |
---|
| 128 | fitter.select_problem_for_fit(id=2,value=0) |
---|
| 129 | |
---|
| 130 | try: result1, = fitter.fit(handler=FitHandler()) |
---|
| 131 | except RuntimeError,msg: |
---|
| 132 | assert str(msg)=="Nothing to fit" |
---|
| 133 | else: raise AssertionError,"No error raised for fitting with no model" |
---|
| 134 | fitter.select_problem_for_fit(id=1,value=1) |
---|
| 135 | fitter.select_problem_for_fit(id=2,value=1) |
---|
| 136 | R1,R2 = fitter.fit(handler=FitHandler()) |
---|
| 137 | |
---|
| 138 | self.assertTrue( math.fabs(R1.pvec[0]-4)/3 <= R1.stderr[0] ) |
---|
| 139 | self.assertTrue( math.fabs(R1.pvec[1]-2.5)/3 <= R1.stderr[1] ) |
---|
| 140 | self.assertTrue( R1.fitness/(len(data1.x)+len(data2.x)) < 2) |
---|
| 141 | |
---|
| 142 | |
---|
| 143 | def test_constraints(self): |
---|
| 144 | """ fit 2 data and 2 model with 1 constrainst""" |
---|
| 145 | #load data |
---|
| 146 | l = Loader() |
---|
| 147 | data1= l.load("testdata_line.txt") |
---|
| 148 | data1.name = data1.filename |
---|
| 149 | data2= l.load("testdata_cst.txt") |
---|
| 150 | data2.name = data2.filename |
---|
| 151 | |
---|
| 152 | # Receives the type of model for the fitting |
---|
| 153 | model11 = LineModel() |
---|
| 154 | model11.name= "line" |
---|
| 155 | model11.setParam("A", 1.0) |
---|
| 156 | model11.setParam("B",1.0) |
---|
| 157 | |
---|
| 158 | model22 = Constant() |
---|
| 159 | model22.name= "cst" |
---|
| 160 | model22.setParam("value", 1.0) |
---|
| 161 | |
---|
| 162 | model1 = Model(model11,data1) |
---|
| 163 | model2 = Model(model22,data2) |
---|
| 164 | model1.set(A=4) |
---|
| 165 | model1.set(B=3) |
---|
| 166 | # Constraint the constant value to be equal to parameter B (the real value is 2.5) |
---|
| 167 | #model2.set(value='line.B') |
---|
| 168 | pars1= ['A','B'] |
---|
| 169 | pars2= ['value'] |
---|
| 170 | |
---|
| 171 | #Importing the Fit module |
---|
[acf8e4a5] | 172 | fitter = Fit() |
---|
[efa5e44] | 173 | fitter.set_data(data1,1) |
---|
| 174 | fitter.set_model(model1,1,pars1) |
---|
| 175 | fitter.set_data(data2,2,smearer=None) |
---|
| 176 | fitter.set_model(model2,2,pars2,constraints=[("value","line.B")]) |
---|
| 177 | fitter.select_problem_for_fit(id=1,value=1) |
---|
| 178 | fitter.select_problem_for_fit(id=2,value=1) |
---|
| 179 | |
---|
| 180 | R1,R2 = fitter.fit(handler=FitHandler()) |
---|
| 181 | self.assertTrue( math.fabs(R1.pvec[0]-4.0)/3. <= R1.stderr[0]) |
---|
| 182 | self.assertTrue( math.fabs(R1.pvec[1]-2.5)/3. <= R1.stderr[1]) |
---|
| 183 | self.assertTrue( R1.fitness/(len(data1.x)+len(data2.x)) < 2) |
---|
| 184 | |
---|
| 185 | |
---|
| 186 | def test4(self): |
---|
| 187 | """ fit 2 data concatenates with limited range of x and one model """ |
---|
| 188 | #load data |
---|
| 189 | l = Loader() |
---|
| 190 | data1 = l.load("testdata_line.txt") |
---|
| 191 | data1.name = data1.filename |
---|
| 192 | data2 = l.load("testdata_line1.txt") |
---|
| 193 | data2.name = data2.filename |
---|
| 194 | |
---|
| 195 | # Receives the type of model for the fitting |
---|
| 196 | model1 = LineModel() |
---|
| 197 | model1.name= "M1" |
---|
| 198 | model1.setParam("A", 1.0) |
---|
| 199 | model1.setParam("B",1.0) |
---|
| 200 | model = Model(model1,data1) |
---|
| 201 | |
---|
| 202 | pars1= ['A','B'] |
---|
| 203 | #Importing the Fit module |
---|
| 204 | |
---|
[acf8e4a5] | 205 | fitter = Fit() |
---|
[efa5e44] | 206 | fitter.set_data(data1,1,qmin=0, qmax=7) |
---|
| 207 | fitter.set_model(model,1,pars1) |
---|
| 208 | fitter.set_data(data2,1,qmin=1,qmax=10) |
---|
| 209 | fitter.select_problem_for_fit(id=1,value=1) |
---|
| 210 | result2, = fitter.fit(handler=FitHandler()) |
---|
| 211 | |
---|
| 212 | self.assert_(result2) |
---|
| 213 | self.assertTrue( math.fabs(result2.pvec[0]-4)/3 <= result2.stderr[0] ) |
---|
| 214 | self.assertTrue( math.fabs(result2.pvec[1]-2.5)/3 <= result2.stderr[1] ) |
---|
| 215 | self.assertTrue( result2.fitness/len(data1.x) < 2) |
---|
| 216 | |
---|
| 217 | |
---|
| 218 | if __name__ == "__main__": |
---|
| 219 | unittest.main() |
---|
| 220 | |
---|