def test_parse_modelspace(self) : """ Test the _parse_modelspace command """ self.ss = Chou2006() self.ar = ARSolver(self.ss) modelspace = self.ss.constraint.modelspace # Test the alphas modelspace.alpha['defaultLowerBound'] = 0.5 modelspace.alpha['defaultUpperBound'] = 30.0 self.ar._parse_modelspace() assert_f = lambda t : eq_(t,(0.5,30.0)) map(assert_f,self.ar.modelspace['alpha']) modelspace.alpha['alpha_1'] = 'nonZero' modelspace.alpha['alpha_3'] = (1.0,20.0) self.ar._parse_modelspace() eq_(self.ar.modelspace['alpha'][0],'nonZero') eq_(self.ar.modelspace['alpha'][1],(0.5,30.0)) eq_(self.ar.modelspace['alpha'][2],(1.0,20.0)) eq_(self.ar.modelspace['alpha'][3],(0.5,30.0)) # Test the betas modelspace.beta['defaultLowerBound'] = 0.5 modelspace.beta['defaultUpperBound'] = 30.0 self.ar._parse_modelspace() assert_f = lambda t : eq_(t,(0.5,30.0)) map(assert_f,self.ar.modelspace['beta']) modelspace.beta['beta_1'] = 'nonZero' modelspace.beta['beta_3'] = 'nonZero' self.ar._parse_modelspace() eq_(self.ar.modelspace['beta'][0],'nonZero') eq_(self.ar.modelspace['beta'][1],(0.5,30.0)) eq_(self.ar.modelspace['beta'][2],'nonZero') eq_(self.ar.modelspace['beta'][3],(0.5,30.0)) # Test g modelspace.g['defaultLowerBound'] = 0.5 modelspace.g['defaultUpperBound'] = 30.0 self.ar._parse_modelspace() assert_f = lambda t : eq_(t,(0.5,30.0)) assert_f2 = lambda t : map(assert_f,t) map(assert_f2,self.ar.modelspace['g']) modelspace.g['g_1_1'] = 'nonZero' modelspace.g['g_3_2'] = (0.33,20.0) self.ar._parse_modelspace() eq_(self.ar.modelspace['g'][0][0],'nonZero') eq_(self.ar.modelspace['g'][1][0],(0.5,30.0)) eq_(self.ar.modelspace['g'][2][1],(0.33,20.0)) eq_(self.ar.modelspace['g'][3][1],(0.5,30.0))
def test_parse_initsol(self) : """ Test the _parse_initsol command """ self.ss = Chou2006() self.ar = ARSolver(self.ss) initsol = self.ss.constraint.initsol # Test the alphas initsol.alpha['defaultInitialValue'] = 0.5 self.ar._parse_initsol() assert_f = lambda t : assert_almost_equal(t,0.5,places=2) map(assert_f,self.ar.initsol['alpha']) initsol.alpha['alpha_1'] = 5.0 initsol.alpha['alpha_3'] = 20.0 self.ar._parse_initsol() assert_almost_equal(self.ar.initsol['alpha'][0],5.0,places=2) assert_almost_equal(self.ar.initsol['alpha'][1],0.5,places=2) assert_almost_equal(self.ar.initsol['alpha'][2],20.0,places=2) assert_almost_equal(self.ar.initsol['alpha'][3],0.5,places=2) # Test the betas initsol.beta['defaultInitialValue'] = 0.5 self.ar._parse_initsol() assert_f = lambda t : assert_almost_equal(t,0.5,places=2) map(assert_f,self.ar.initsol['beta']) initsol.beta['beta_1'] = 5.0 initsol.beta['beta_3'] = 20.0 self.ar._parse_initsol() assert_almost_equal(self.ar.initsol['beta'][0],5.0,places=2) assert_almost_equal(self.ar.initsol['beta'][1],0.5,places=2) assert_almost_equal(self.ar.initsol['beta'][2],20.0,places=2) assert_almost_equal(self.ar.initsol['beta'][3],0.5,places=2) # Test g initsol.g['defaultInitialValue'] = 0.5 self.ar._parse_initsol() assert_f = lambda t : assert_almost_equal(t,0.5,places=2) map(assert_f,self.ar.initsol['g'].ravel()) initsol.g['g_1_1'] = 0.45 initsol.g['g_3_2'] = 0.33 self.ar._parse_initsol() assert_almost_equal(self.ar.initsol['g'][0][0],0.45,places=2) assert_almost_equal(self.ar.initsol['g'][2][1],0.33,places=2) eq_(self.ar.initsol['g'][1][0],0.5) eq_(self.ar.initsol['g'][3][1],0.5)
def test_noinfo2(self) :
""" Run AR all equations under no info """
ss = Chou2006()
ss.exptype = "noinfo"
ss.equations = [1,2,3,4]
ar = ARSolver(ss)
ar.solve(maxiter=10000,tol=10e-7)
eq_(len(ar.all_exp_art),1)
print ar.all_exp_art[0]['eqns']
eq_(len(ar.all_exp_art[0]['eqns']),4)
art = ar.all_exp_art[0]['eqns'][0]
sol = {'alpha': 12,
'beta': 10,
'g': np.array([ 0,0,-0.8,0]),
'h': np.array([ 0.5,0,0,0])}
def test_noinfo1(self) :
""" Run AR eqn 1 under no info """
ss = Chou2006()
ss.exptype = "noinfo"
ss.equations = [1]
ar = ARSolver(ss)
ar.solve(maxiter=10000,tol=10e-7)
assert len(ar.all_exp_art) == 1
assert len(ar.all_exp_art[0]['eqns']) == 1
art = ar.all_exp_art[0]['eqns'][0]
sol = {'alpha': 12,
'beta': 10,
'g': np.array([ 0,0,-0.8,0]),
'h': np.array([ 0.5,0,0,0])}
params = art.params[-1]
assert np.allclose(sol['alpha'],params['alpha'],atol=0.5)
assert np.allclose(sol['beta'],params['beta'],atol=0.5)
assert np.allclose(sol['g'],params['g'],atol=0.2)
assert np.allclose(sol['h'],params['h'],atol=0.2)
def test_fullinfo1(self) :
""" RUN AR (slow) eqn 1 under full info """
ss = Chou2006()
ss.exptype = "fullinfo"
ss.equations = [1]
ar = ARSolver(ss)
ar.solve(maxiter=10000,tol=10e-6)
assert len(ar.all_exp_art) == 1
assert len(ar.all_exp_art[0]['eqns']) == 1
art = ar.all_exp_art[0]['eqns'][0]
sol = {'alpha': 11.177018643003306,
'beta': 8.7027543289527056,
'g': np.array([-1.0236953]),
'h': np.array([ 0.62268492])}
params = art.params[-1]
assert np.allclose(sol['alpha'],params['alpha'],atol=1)
assert np.allclose(sol['beta'],params['beta'],atol=1.5)
assert np.allclose(sol['g'],params['g'],atol=0.5)
assert np.allclose(sol['h'],params['h'],atol=0.5)
def test_fullinfo2(self) :
""" RUN AR (slow) eqn 2 under full info """
ss = Chou2006()
ss.exptype = "fullinfo"
ss.equations = [2]
ar = ARSolver(ss)
ar.solve(maxiter=10000,tol=10e-6)
assert len(ar.all_exp_art) == 1
assert len(ar.all_exp_art[0]['eqns']) == 1
art = ar.all_exp_art[0]['eqns'][0]
sol = {'alpha': 7.8801765469678156,
'beta': 2.8893110127454427,
'g': np.array([ 0.51119746]),
'h': np.array([ 0.76771705])}
params = art.params[-1]
assert np.allclose(sol['alpha'],params['alpha'],atol=1)
assert np.allclose(sol['beta'],params['beta'],atol=1.5)
assert np.allclose(sol['g'],params['g'],atol=0.5)
assert np.allclose(sol['h'],params['h'],atol=0.5)
def test_partialinfo1(self) : """ Run AR (slow) on eqn 1 under partial info """ ss = Chou2006() ss.exptype = "partialinfo" ss.equations = [1] initsol = ss.constraint.initsol initsol.beta['beta_1'] = 20.0 initsol.h['h_1_1'] = 0.1 ar = ARSolver(ss) eq_(ar.initsol['beta'][0],20.0) eq_(ar.initsol['h'][0][0],0.1)
def test_core_init_params(self) : """ Test _core_init_params""" self.ss = Chou2006() self.ss.exptype= "fullinfo" self.ss.equations = [1,3] self.ss.constraint.initsol.beta['beta_1'] = 7.0 self.ss.constraint.initsol.h['h_1_1'] = 1.2 self.ar = ARSolver(self.ss) (a_list,b_list,g_list,h_list) = self.ar._core_init_params() print a_list print b_list print g_list print h_list #assert False eq_(a_list,[5.0,5.0]) eq_(b_list,[7.0,5.0]) ass_f = lambda t : eq_(np.allclose(t[0],t[1]),True) map(ass_f,zip(h_list,[np.array([1.2]),np.array([1.0,1.0])])) map(ass_f,zip(g_list,[np.array([1.0]),np.array([1.0])]))
def setUp(self):
self.ss = Chou2006()
self.ar = ARSolver(self.ss)
def setUp(self) : self.ss = Chou2006() self.ar = ARSolver(self.ss)
class TestARSolverChou2006(object) :
""" Test cases for the AR solver """
def setUp(self) :
self.ss = Chou2006()
self.ar = ARSolver(self.ss)
def tearDown(self) :
pass
def test_parse_initsol(self) :
""" Test the _parse_initsol command """
self.ss = Chou2006()
self.ar = ARSolver(self.ss)
initsol = self.ss.constraint.initsol
# Test the alphas
initsol.alpha['defaultInitialValue'] = 0.5
self.ar._parse_initsol()
assert_f = lambda t : assert_almost_equal(t,0.5,places=2)
map(assert_f,self.ar.initsol['alpha'])
initsol.alpha['alpha_1'] = 5.0
initsol.alpha['alpha_3'] = 20.0
self.ar._parse_initsol()
assert_almost_equal(self.ar.initsol['alpha'][0],5.0,places=2)
assert_almost_equal(self.ar.initsol['alpha'][1],0.5,places=2)
assert_almost_equal(self.ar.initsol['alpha'][2],20.0,places=2)
assert_almost_equal(self.ar.initsol['alpha'][3],0.5,places=2)
# Test the betas
initsol.beta['defaultInitialValue'] = 0.5
self.ar._parse_initsol()
assert_f = lambda t : assert_almost_equal(t,0.5,places=2)
map(assert_f,self.ar.initsol['beta'])
initsol.beta['beta_1'] = 5.0
initsol.beta['beta_3'] = 20.0
self.ar._parse_initsol()
assert_almost_equal(self.ar.initsol['beta'][0],5.0,places=2)
assert_almost_equal(self.ar.initsol['beta'][1],0.5,places=2)
assert_almost_equal(self.ar.initsol['beta'][2],20.0,places=2)
assert_almost_equal(self.ar.initsol['beta'][3],0.5,places=2)
# Test g
initsol.g['defaultInitialValue'] = 0.5
self.ar._parse_initsol()
assert_f = lambda t : assert_almost_equal(t,0.5,places=2)
map(assert_f,self.ar.initsol['g'].ravel())
initsol.g['g_1_1'] = 0.45
initsol.g['g_3_2'] = 0.33
self.ar._parse_initsol()
assert_almost_equal(self.ar.initsol['g'][0][0],0.45,places=2)
assert_almost_equal(self.ar.initsol['g'][2][1],0.33,places=2)
eq_(self.ar.initsol['g'][1][0],0.5)
eq_(self.ar.initsol['g'][3][1],0.5)
def test_parse_modelspace(self) :
""" Test the _parse_modelspace command """
self.ss = Chou2006()
self.ar = ARSolver(self.ss)
modelspace = self.ss.constraint.modelspace
# Test the alphas
modelspace.alpha['defaultLowerBound'] = 0.5
modelspace.alpha['defaultUpperBound'] = 30.0
self.ar._parse_modelspace()
assert_f = lambda t : eq_(t,(0.5,30.0))
map(assert_f,self.ar.modelspace['alpha'])
modelspace.alpha['alpha_1'] = 'nonZero'
modelspace.alpha['alpha_3'] = (1.0,20.0)
self.ar._parse_modelspace()
eq_(self.ar.modelspace['alpha'][0],'nonZero')
eq_(self.ar.modelspace['alpha'][1],(0.5,30.0))
eq_(self.ar.modelspace['alpha'][2],(1.0,20.0))
eq_(self.ar.modelspace['alpha'][3],(0.5,30.0))
# Test the betas
modelspace.beta['defaultLowerBound'] = 0.5
modelspace.beta['defaultUpperBound'] = 30.0
self.ar._parse_modelspace()
assert_f = lambda t : eq_(t,(0.5,30.0))
map(assert_f,self.ar.modelspace['beta'])
modelspace.beta['beta_1'] = 'nonZero'
modelspace.beta['beta_3'] = 'nonZero'
self.ar._parse_modelspace()
eq_(self.ar.modelspace['beta'][0],'nonZero')
eq_(self.ar.modelspace['beta'][1],(0.5,30.0))
eq_(self.ar.modelspace['beta'][2],'nonZero')
eq_(self.ar.modelspace['beta'][3],(0.5,30.0))
# Test g
modelspace.g['defaultLowerBound'] = 0.5
modelspace.g['defaultUpperBound'] = 30.0
self.ar._parse_modelspace()
assert_f = lambda t : eq_(t,(0.5,30.0))
assert_f2 = lambda t : map(assert_f,t)
map(assert_f2,self.ar.modelspace['g'])
modelspace.g['g_1_1'] = 'nonZero'
modelspace.g['g_3_2'] = (0.33,20.0)
self.ar._parse_modelspace()
eq_(self.ar.modelspace['g'][0][0],'nonZero')
eq_(self.ar.modelspace['g'][1][0],(0.5,30.0))
eq_(self.ar.modelspace['g'][2][1],(0.33,20.0))
eq_(self.ar.modelspace['g'][3][1],(0.5,30.0))
def test_regressors(self) :
""" Test _set_regressors code """
# Test for fullinfo and 1st and 3rd eqn
self.ss = Chou2006()
self.ss.exptype = "fullinfo"
self.ss.equations = [1,3]
self.ar = ARSolver(self.ss)
eq_(len(self.ar.regressors),2)
eq_(self.ar.regressors[0]['degrad'],[1])
eq_(self.ar.regressors[0]['prod'],[3])
eq_(self.ar.regressors[1]['degrad'],[3,4])
eq_(self.ar.regressors[1]['prod'],[2])
# Test for fullinfo and all eqns
self.ss = Chou2006()
self.ss.exptype = "fullinfo"
self.ar = ARSolver(self.ss)
eq_(self.ar.regressors[0]['degrad'],[1])
eq_(self.ar.regressors[0]['prod'],[3])
eq_(self.ar.regressors[1]['degrad'],[2])
eq_(self.ar.regressors[1]['prod'],[1])
eq_(self.ar.regressors[2]['degrad'],[3,4])
eq_(self.ar.regressors[2]['prod'],[2])
eq_(self.ar.regressors[3]['degrad'],[4])
eq_(self.ar.regressors[3]['prod'],[1])
# Test for partial info and all equations
self.ss = Chou2006()
self.ss.exptype = "partialinfo"
self.ar = ARSolver(self.ss)
eq_(self.ar.regressors[0]['degrad'],[1,3])
eq_(self.ar.regressors[0]['prod'],[1,3])
eq_(self.ar.regressors[1]['degrad'],[1,2])
eq_(self.ar.regressors[1]['prod'],[1,2])
eq_(self.ar.regressors[2]['degrad'],[2,3,4])
eq_(self.ar.regressors[2]['prod'],[2,3,4])
eq_(self.ar.regressors[3]['degrad'],[1,4])
eq_(self.ar.regressors[3]['prod'],[1,4])
# Test for no info and all equations
self.ss = Chou2006()
self.ss.exptype = "noinfo"
self.ar = ARSolver(self.ss)
for ii in range(4) :
eq_(self.ar.regressors[ii]['degrad'],[1,2,3,4])
eq_(self.ar.regressors[ii]['prod'],[1,2,3,4])
# Test for no info and 1st and 3rd equations
self.ss = Chou2006()
self.ss.exptype = "noinfo"
self.ss.equations = [1,3]
self.ar = ARSolver(self.ss)
eq_(len(self.ar.regressors),2)
for ii in range(2) :
eq_(self.ar.regressors[ii]['degrad'],[1,2,3,4])
eq_(self.ar.regressors[ii]['prod'],[1,2,3,4])
def test_core_init_params(self) :
""" Test _core_init_params"""
self.ss = Chou2006()
self.ss.exptype= "fullinfo"
self.ss.equations = [1,3]
self.ss.constraint.initsol.beta['beta_1'] = 7.0
self.ss.constraint.initsol.h['h_1_1'] = 1.2
self.ar = ARSolver(self.ss)
(a_list,b_list,g_list,h_list) = self.ar._core_init_params()
print a_list
print b_list
print g_list
print h_list
#assert False
eq_(a_list,[5.0,5.0])
eq_(b_list,[7.0,5.0])
ass_f = lambda t : eq_(np.allclose(t[0],t[1]),True)
map(ass_f,zip(h_list,[np.array([1.2]),np.array([1.0,1.0])]))
map(ass_f,zip(g_list,[np.array([1.0]),np.array([1.0])]))
def test_core_calc_design(self) :
""" Test the design matrix"""
# Test under the case of fullinfo
# Generate dummy data
ss = Chou2006()
ss.exptype = "fullinfo"
ss.equations = [1]
ar = ARSolver(ss)
prof = ss.experiments[0].profile
reg_p = [3]
reg_d = [1]
Lp = np.ones(prof.n_sample)
Ld = np.ones(prof.n_sample)
X_p = [np.log(prof.var[:,2])]
X_d = [np.log(prof.var[:,0])]
Lp = np.vstack((Lp,np.array(X_p))).T
Ld = np.vstack((Ld,np.array(X_d))).T
Cp = np.dot(LA.inv(np.dot(Lp.T,Lp)),Lp.T)
Cd = np.dot(LA.inv(np.dot(Ld.T,Ld)),Ld.T)
lp_list1,ld_list1 = [Lp],[Ld]
cp_list1,cd_list1 = [Cp],[Cd]
# Generate code results
lp_list2,ld_list2,cp_list2,cd_list2 = ar._core_calc_design(prof)
eq_(len(lp_list2),1)
eq_(len(ld_list2),1)
eq_(len(cp_list2),1)
eq_(len(cd_list2),1)
assert np.allclose(Lp,lp_list2[0])
assert np.allclose(Ld,ld_list2[0])
assert np.allclose(Cp,cp_list2[0])
assert np.allclose(Cd,cd_list2[0])
# Test under the case of no info
ss = Chou2006()
ss.exptype = "noinfo"
ss.equations = [1,2]
ar = ARSolver(ss)
prof = ss.experiments[0].profile
lp_list1,ld_list1 = [],[]
cp_list1,cd_list1 = [],[]
reg_p = [1,2,3,4]
reg_d = [1,2,3,4]
Lp = np.ones(prof.n_sample)
Ld = np.ones(prof.n_sample)
X_p = np.log(prof.var.T)
X_d = np.log(prof.var.T)
print "X_p.shape", X_p.shape
print "X_d.shape", X_d.shape
Lp = np.vstack((Lp,X_p)).T
Ld = np.vstack((Ld,X_d)).T
Cp = np.dot(LA.inv(np.dot(Lp.T,Lp)),Lp.T)
Cd = np.dot(LA.inv(np.dot(Ld.T,Ld)),Ld.T)
lp_list1,ld_list1 = [Lp,Lp],[Ld,Ld]
cp_list1,cd_list1 = [Cp,Cp],[Cd,Cd]
# Generate code results
lp_list2,ld_list2,cp_list2,cd_list2 = ar._core_calc_design(prof)
eq_(len(lp_list2),2)
eq_(len(ld_list2),2)
eq_(len(cp_list2),2)
eq_(len(cd_list2),2)
assert np.allclose(Lp,lp_list2[0])
assert np.allclose(Ld,ld_list2[0])
assert np.allclose(Cp,cp_list2[0])
assert np.allclose(Cd,cd_list2[0])
assert np.allclose(Lp,lp_list2[1])
assert np.allclose(Ld,ld_list2[1])
assert np.allclose(Cp,cp_list2[1])
assert np.allclose(Cd,cd_list2[1])
def test_core_calc_sse(self) :
""" Test the sse """
calc_sse = self.ar._core_calc_sse
L = np.array([[1,2,3],[2,3,4]])
b = np.array([1,2,3])
y = np.array([12,15])
assert_almost_equal(calc_sse(y,L,b),29)
def test_core_phase1(self) :
""" The phase1"""
pass
def test_core_phase2(self) :
""" Test phase 2 """
pass
def test_core_monitor(self) :
""" Test monitor """
pass
@attr('slow')
def test_fullinfo1(self) :
""" RUN AR (slow) eqn 1 under full info """
ss = Chou2006()
ss.exptype = "fullinfo"
ss.equations = [1]
ar = ARSolver(ss)
ar.solve(maxiter=10000,tol=10e-6)
assert len(ar.all_exp_art) == 1
assert len(ar.all_exp_art[0]['eqns']) == 1
art = ar.all_exp_art[0]['eqns'][0]
sol = {'alpha': 11.177018643003306,
'beta': 8.7027543289527056,
'g': np.array([-1.0236953]),
'h': np.array([ 0.62268492])}
params = art.params[-1]
assert np.allclose(sol['alpha'],params['alpha'],atol=1)
assert np.allclose(sol['beta'],params['beta'],atol=1.5)
assert np.allclose(sol['g'],params['g'],atol=0.5)
assert np.allclose(sol['h'],params['h'],atol=0.5)
@attr('slow')
def test_fullinfo2(self) :
""" RUN AR (slow) eqn 2 under full info """
ss = Chou2006()
ss.exptype = "fullinfo"
ss.equations = [2]
ar = ARSolver(ss)
ar.solve(maxiter=10000,tol=10e-6)
assert len(ar.all_exp_art) == 1
assert len(ar.all_exp_art[0]['eqns']) == 1
art = ar.all_exp_art[0]['eqns'][0]
sol = {'alpha': 7.8801765469678156,
'beta': 2.8893110127454427,
'g': np.array([ 0.51119746]),
'h': np.array([ 0.76771705])}
params = art.params[-1]
assert np.allclose(sol['alpha'],params['alpha'],atol=1)
assert np.allclose(sol['beta'],params['beta'],atol=1.5)
assert np.allclose(sol['g'],params['g'],atol=0.5)
assert np.allclose(sol['h'],params['h'],atol=0.5)
#assert_equal_dict(art.params,sol)
@attr('mediocre')
def test_noinfo1(self) :
""" Run AR eqn 1 under no info """
ss = Chou2006()
ss.exptype = "noinfo"
ss.equations = [1]
ar = ARSolver(ss)
ar.solve(maxiter=10000,tol=10e-7)
assert len(ar.all_exp_art) == 1
assert len(ar.all_exp_art[0]['eqns']) == 1
art = ar.all_exp_art[0]['eqns'][0]
sol = {'alpha': 12,
'beta': 10,
'g': np.array([ 0,0,-0.8,0]),
'h': np.array([ 0.5,0,0,0])}
params = art.params[-1]
assert np.allclose(sol['alpha'],params['alpha'],atol=0.5)
assert np.allclose(sol['beta'],params['beta'],atol=0.5)
assert np.allclose(sol['g'],params['g'],atol=0.2)
assert np.allclose(sol['h'],params['h'],atol=0.2)
@attr('mediocre')
def test_noinfo2(self) :
""" Run AR all equations under no info """
ss = Chou2006()
ss.exptype = "noinfo"
ss.equations = [1,2,3,4]
ar = ARSolver(ss)
ar.solve(maxiter=10000,tol=10e-7)
eq_(len(ar.all_exp_art),1)
print ar.all_exp_art[0]['eqns']
eq_(len(ar.all_exp_art[0]['eqns']),4)
art = ar.all_exp_art[0]['eqns'][0]
sol = {'alpha': 12,
'beta': 10,
'g': np.array([ 0,0,-0.8,0]),
'h': np.array([ 0.5,0,0,0])}
@attr('slow')
def test_partialinfo1(self) :
""" Run AR (slow) on eqn 1 under partial info """
ss = Chou2006()
ss.exptype = "partialinfo"
ss.equations = [1]
initsol = ss.constraint.initsol
initsol.beta['beta_1'] = 20.0
initsol.h['h_1_1'] = 0.1
ar = ARSolver(ss)
eq_(ar.initsol['beta'][0],20.0)
eq_(ar.initsol['h'][0][0],0.1)
def test_core_calc_design(self) : """ Test the design matrix""" # Test under the case of fullinfo # Generate dummy data ss = Chou2006() ss.exptype = "fullinfo" ss.equations = [1] ar = ARSolver(ss) prof = ss.experiments[0].profile reg_p = [3] reg_d = [1] Lp = np.ones(prof.n_sample) Ld = np.ones(prof.n_sample) X_p = [np.log(prof.var[:,2])] X_d = [np.log(prof.var[:,0])] Lp = np.vstack((Lp,np.array(X_p))).T Ld = np.vstack((Ld,np.array(X_d))).T Cp = np.dot(LA.inv(np.dot(Lp.T,Lp)),Lp.T) Cd = np.dot(LA.inv(np.dot(Ld.T,Ld)),Ld.T) lp_list1,ld_list1 = [Lp],[Ld] cp_list1,cd_list1 = [Cp],[Cd] # Generate code results lp_list2,ld_list2,cp_list2,cd_list2 = ar._core_calc_design(prof) eq_(len(lp_list2),1) eq_(len(ld_list2),1) eq_(len(cp_list2),1) eq_(len(cd_list2),1) assert np.allclose(Lp,lp_list2[0]) assert np.allclose(Ld,ld_list2[0]) assert np.allclose(Cp,cp_list2[0]) assert np.allclose(Cd,cd_list2[0]) # Test under the case of no info ss = Chou2006() ss.exptype = "noinfo" ss.equations = [1,2] ar = ARSolver(ss) prof = ss.experiments[0].profile lp_list1,ld_list1 = [],[] cp_list1,cd_list1 = [],[] reg_p = [1,2,3,4] reg_d = [1,2,3,4] Lp = np.ones(prof.n_sample) Ld = np.ones(prof.n_sample) X_p = np.log(prof.var.T) X_d = np.log(prof.var.T) print "X_p.shape", X_p.shape print "X_d.shape", X_d.shape Lp = np.vstack((Lp,X_p)).T Ld = np.vstack((Ld,X_d)).T Cp = np.dot(LA.inv(np.dot(Lp.T,Lp)),Lp.T) Cd = np.dot(LA.inv(np.dot(Ld.T,Ld)),Ld.T) lp_list1,ld_list1 = [Lp,Lp],[Ld,Ld] cp_list1,cd_list1 = [Cp,Cp],[Cd,Cd] # Generate code results lp_list2,ld_list2,cp_list2,cd_list2 = ar._core_calc_design(prof) eq_(len(lp_list2),2) eq_(len(ld_list2),2) eq_(len(cp_list2),2) eq_(len(cd_list2),2) assert np.allclose(Lp,lp_list2[0]) assert np.allclose(Ld,ld_list2[0]) assert np.allclose(Cp,cp_list2[0]) assert np.allclose(Cd,cd_list2[0]) assert np.allclose(Lp,lp_list2[1]) assert np.allclose(Ld,ld_list2[1]) assert np.allclose(Cp,cp_list2[1]) assert np.allclose(Cd,cd_list2[1])
def test_regressors(self) : """ Test _set_regressors code """ # Test for fullinfo and 1st and 3rd eqn self.ss = Chou2006() self.ss.exptype = "fullinfo" self.ss.equations = [1,3] self.ar = ARSolver(self.ss) eq_(len(self.ar.regressors),2) eq_(self.ar.regressors[0]['degrad'],[1]) eq_(self.ar.regressors[0]['prod'],[3]) eq_(self.ar.regressors[1]['degrad'],[3,4]) eq_(self.ar.regressors[1]['prod'],[2]) # Test for fullinfo and all eqns self.ss = Chou2006() self.ss.exptype = "fullinfo" self.ar = ARSolver(self.ss) eq_(self.ar.regressors[0]['degrad'],[1]) eq_(self.ar.regressors[0]['prod'],[3]) eq_(self.ar.regressors[1]['degrad'],[2]) eq_(self.ar.regressors[1]['prod'],[1]) eq_(self.ar.regressors[2]['degrad'],[3,4]) eq_(self.ar.regressors[2]['prod'],[2]) eq_(self.ar.regressors[3]['degrad'],[4]) eq_(self.ar.regressors[3]['prod'],[1]) # Test for partial info and all equations self.ss = Chou2006() self.ss.exptype = "partialinfo" self.ar = ARSolver(self.ss) eq_(self.ar.regressors[0]['degrad'],[1,3]) eq_(self.ar.regressors[0]['prod'],[1,3]) eq_(self.ar.regressors[1]['degrad'],[1,2]) eq_(self.ar.regressors[1]['prod'],[1,2]) eq_(self.ar.regressors[2]['degrad'],[2,3,4]) eq_(self.ar.regressors[2]['prod'],[2,3,4]) eq_(self.ar.regressors[3]['degrad'],[1,4]) eq_(self.ar.regressors[3]['prod'],[1,4]) # Test for no info and all equations self.ss = Chou2006() self.ss.exptype = "noinfo" self.ar = ARSolver(self.ss) for ii in range(4) : eq_(self.ar.regressors[ii]['degrad'],[1,2,3,4]) eq_(self.ar.regressors[ii]['prod'],[1,2,3,4]) # Test for no info and 1st and 3rd equations self.ss = Chou2006() self.ss.exptype = "noinfo" self.ss.equations = [1,3] self.ar = ARSolver(self.ss) eq_(len(self.ar.regressors),2) for ii in range(2) : eq_(self.ar.regressors[ii]['degrad'],[1,2,3,4]) eq_(self.ar.regressors[ii]['prod'],[1,2,3,4])
