def test3(self):
print("===== formatter test3 ===========================")
arr = numpy.asarray([k for k in range(120)], dtype=float)
arr = arr.reshape((10, 12))
fmtinit(max=4)
print(fmt(arr))
print(fmt(arr, max=None))
print(fmt(arr, tail=1))
print(fmt(arr, tail=3))
print(fmt(arr[1, :], tail=1))
print(fmt(arr[1, :], tail=3))
print(fmt(arr[1, :], max=8, tail=2, linelength=200))
print(fmt(arr[1, :], max=8, tail=3, linelength=200))
print(fmt(arr[1, :], max=8, tail=4, linelength=200))
print(fmt(arr[1, :], max=8, tail=5, linelength=200))
def testcr2(self):
print("===== cyclic results 2 ================")
mdl = PhaseModel(cyclic=1.0)
x = numpy.array([0, 1, 9, 10], dtype=float)
y = numpy.array([1, 0, 10, 9], dtype=float)
problem = Problem(model=mdl, xdata=x, ydata=y)
Tools.printclass(problem)
r0 = numpy.linspace(-0.7, 1.5, 12)
print(fmt(r0, max=None))
r1 = problem.cyclicCorrection(r0)
print(fmt(r1, max=None))
self.assertTrue(all(numpy.abs(r1) <= 0.5))
def makeModel(self, nin, ndout, ndhid, npt):
x, y = self.makeData(nin, npt, ndout)
m1 = SoftMaxModel(ndim=nin, ndout=ndhid, offset=True)
m1.setLimits(-10, 10)
p1 = numpy.random.random(m1.npars) * 20 - 10
print(m1.npars, fmt(m1.result(x, p1)))
m2 = SoftMaxModel(ndim=ndhid, ndout=ndout, offset=True)
m2.setLimits(-10, 10)
m = m1 | m2
pp = numpy.random.random(m.npars) * 20 - 10
print(m.npars, fmt(m.result(x, pp)))
problem = ClassicProblem(model=m, xdata=x, ydata=y)
return problem
def testcr4(self):
print("===== cyclic results 4 ================")
mdl = PhaseModel(cyclic={0: 1.0})
x = numpy.array([0, 1, 9, 10], dtype=float)
y = numpy.array([1, 0, 10, 9], dtype=float)
problem = Problem(model=mdl, xdata=x, ydata=y)
Tools.printclass(problem)
r0 = numpy.linspace(-0.8, 1.5, 12).reshape(2, 6).transpose()
print(fmt(r0, max=None))
r1 = problem.cyclicCorrection(r0)
print(fmt(r1, max=None))
assertAAE(r0[:, 1], r1[:, 1])
self.assertTrue(all(numpy.abs(r1[:, 0]) < 0.5))
def dotest5(self, step=10):
print("====test5====step=%d========================" % step)
plot = self.doplot
nn = 10
x = numpy.linspace(0, 2, nn, dtype=float)
ym = 0.3 + 0.5 * x
nf = 0.1
numpy.random.seed(2345)
noise = numpy.random.randn(nn)
y = ym + nf * noise
limits = [-1, 2]
model = PolynomialModel(1)
model.setLimits(lowLimits=limits[0], highLimits=limits[1])
s = 0.0
s2 = 0.0
mr = 10
for k in range(mr):
dis = GaussErrorDistribution(scale=0.5)
ns = PhantomSampler(x,
model,
y,
distribution=dis,
verbose=0,
seed=k + 12 * step)
ns.step = step
logE = ns.sample()
par2 = ns.parameters
logz2 = ns.logZ
dlz2 = ns.logZprecision
s += logz2
s2 += logz2 * logz2
print("pars ", fmt(par2), " logZ ", fmt(logz2), " +- ",
fmt(dlz2))
logz = s / mr
dlz = math.sqrt(s2 / mr - logz * logz)
print("Average ", fmt(logz), " +- ", fmt(dlz))
def test2( self ):
print( " Test 2 DecisionTreeModel" )
ND = 10
NP = 100
m = DecisionTreeModel( ndim=ND, kdim=[0,1,2,3,4,5,6], depth=3 )
print( m )
numpy.random.seed( 12345 )
xdata = numpy.random.rand( NP, ND )
print( fmt( xdata ) )
par0 = numpy.arange( m.npars, dtype=float ) + 1
res = m.result( xdata, par0 )
part = m.partial( xdata, par0 )
for k in range( NP ) :
print( fmt( res[k] ), fmt( part[k,:], max=None ) )
self.assertTrue( part[k,int(res[k]-0.99)] == 1 )
self.assertTrue( numpy.sum( part[k,:] ) == 1 )
def test1Model11( self ):
print( " Test SplinesDynamicModel shrink" )
xx = numpy.linspace( 0, 10, 501, dtype=float )
kn1 = numpy.linspace( 0, 10, 16, dtype=float )
m1 = BasicSplinesModel( knots=kn1 )
p1 = numpy.zeros( 18, dtype=float )
p1[7:9] = 1.0
kn2 = numpy.append( numpy.append( kn1[:7], [5.0] ), kn1[9:] )
m2 = BasicSplinesModel( knots=kn2 )
p2 = numpy.zeros( 17, dtype=float )
p2[7] = 1.4
print( fmt( kn1, max=None ) )
print( fmt( kn2, max=None ) )
if self.doplot :
plt.plot( xx, m1.result( xx, p1 ) )
plt.plot( xx, m2.result( xx, p2 ) )
plt.show()
def test4(self):
print("\n SalesmanProblem Test User distance\n")
np = 10
pars = numpy.arange(np * np, dtype=int)
problem = self.initProblem(np=np, distance=self.square)
print(problem)
mdis = problem.minimumDistance()
print("mindis ", mdis)
dis = problem.result(pars)
sdis = numpy.sum(dis)
print(fmt(dis, max=None), fmt(sdis))
self.assertAlmostEqual(sdis, 990, 6)
problem = self.initProblem(np=np, distance=self.square, scale=3)
dis = problem.result(pars)
sdis = numpy.sum(dis)
print(fmt(dis, max=None), fmt(sdis))
self.assertAlmostEqual(sdis, 990 / 3, 6)
def test1(self):
print("\n SalesmanProblem Test Manhattan\n")
np = 10
pars = numpy.arange(np * np, dtype=int)
problem = self.initProblem(np=np, distance="manhattan")
print(problem)
mdis = problem.minimumDistance()
print("mindis ", mdis)
dis = problem.result(pars)
sdis = numpy.sum(dis)
print(fmt(dis, max=None), fmt(sdis))
self.assertAlmostEqual(sdis, 198, 6)
problem = self.initProblem(np=np, distance="manhattan", scale=3)
dis = problem.result(pars)
sdis = numpy.sum(dis)
print(fmt(dis, max=None), fmt(sdis))
self.assertAlmostEqual(sdis, 198 / 3, 6)
def test6(self):
npt = 401
t, y = self.makeData()
t = numpy.linspace(0, 100, npt, dtype=float)
ym = 2 * numpy.sin(2 * math.pi * numpy.exp(t / 60) + 1) * numpy.exp(
-0.02 * t)
y = numpy.random.seed(12345)
y = ym + numpy.random.randn(npt) * 0.05
kn = [0, 25, 50, 60, 70, 80, 90, 100]
kn = [
0.000, 17.254, 33.456, 42.755, 61.243, 62.538, 74.369, 89.854,
93.388, 100.000
]
m = BasicSplinesModel(knots=kn)
ftr = Fitter(t, m, fixedScale=0.05)
par = ftr.fit(y)
lz0 = ftr.getLogZ(limits=[-10, +10])
print(fmt(par, max=None), fmt(lz0))
m = BasicSplinesModel(knots=kn)
ftr = Fitter(t, m)
par = ftr.fit(y)
lz0 = ftr.getLogZ(limits=[-10, +10])
print(fmt(par, max=None), fmt(lz0))
m = BasicSplinesModel(knots=kn)
ftr = Fitter(t, m)
par = ftr.fit(y)
lz0 = ftr.getLogZ(limits=[-10, +10], noiseLimits=[0.01, 1])
print(fmt(par, max=None), fmt(lz0))
def plot1(self):
x = numpy.linspace(0, 10, 101, dtype=float)
y = numpy.sin(x)
kn1 = numpy.linspace(0, 10, 11, dtype=float)
sm1 = BasicSplinesModel(knots=kn1)
ftr = Fitter(x, sm1)
par1 = ftr.fit(y)
print(fmt(par1, max=None))
plt.plot(x, y, 'k.')
plt.plot(x, sm1.result(x, par1))
plt.show()
def bstest6(self, kn, order=3):
print("==== test 6 ====================")
x = numpy.linspace(min(kn), max(kn), 101, dtype=float)
n = 0
cc = ['k-', 'b-', 'r-', 'g-', 'c-', 'm-']
sm = BasicSplinesModel(knots=kn, order=order)
bm = BSplinesModel(knots=kn, order=order)
par = numpy.ones(bm.npars, dtype=float)
ysm = sm.result(x, par)
pts = sm.partial(x, par)
ybm = bm.result(x, par)
ptb = bm.partial(x, par)
print(fmt(numpy.sum(pts, 0), max=None))
print(fmt(numpy.sum(ptb, 0), max=None))
if self.doplot:
for k in kn:
plt.plot([k, k], [0, 1], 'k:')
plt.plot(x, ybm, 'b-')
plt.plot(x, ysm + 1.1, 'k-')
for i in range(bm.npars):
# plt.plot( x, ptb[:,i] - pts[:,i], cc[2] )
plt.plot(x, ptb[:, i], cc[2])
plt.plot(x, pts[:, i] + 1.1, cc[3])
plt.show()
assertAAE(ysm, ybm)
if order > 0:
assertAAE(pts, ptb)
def test4(self):
print("======= Neural Net Utilities test 4 ========================")
con = NeuralNetUtilities.ConnectWithBias(ndim=2, ndout=3)
atn = NeuralNetUtilities.Logistic()
print(con.ndim, con.ndout, con.nraster)
self.assertTrue(con.nraster == 6)
xx = numpy.array([[1, 2], [2, 3], [3, 4], [4, 5]], dtype=float)
par = numpy.arange(9, dtype=float) + 1
print(fmt(xx))
print(fmt(par, max=None))
res = atn.result(con.result(xx, par), par)
print("result ", res.shape)
print(fmt(res))
partial = atn.derivative(con.partial(xx, par), par)
print("partial ", partial.shape)
print(fmt(partial, max=None))
def stdstarttest(self, mystarteng, np=4, nwalker=10):
problem = self.initProblem(np=np)
errdis = DistanceCostFunction()
n2 = np * np
pars = numpy.arange(n2, dtype=int)
fi = numpy.arange(n2, dtype=int)
sl = WalkerList(problem, nwalker, pars, fi)
steng = mystarteng(sl, errdis, verbose=5)
for wlkr in steng.walkers:
logL = errdis.logLikelihood(problem, wlkr.allpars)
wid = wlkr.id
print("Walker :", wid, wlkr.allpars, fmt(logL))
steng.execute(wid, -math.inf)
nwlk = steng.walkers[wid]
print(" ", nwlk.allpars, fmt(nwlk.logL))
psort = numpy.sort(nwlk.allpars)
assertAE(psort, pars)
def test4(self):
print("******ASTROPY MODEL 4***********************")
gm = modeling.models.Gaussian1D()
# print( gm.__class__ )
# print( gm )
apm = AstropyModel(gm)
apm.setPrior(0, ExponentialPrior(scale=10))
apm.setPrior(1, UniformPrior(limits=[-10, 10]))
apm.setPrior(2, JeffreysPrior(limits=[0.1, 10]))
bfm = PolynomialModel(1)
bfm.setPrior(0, UniformPrior(limits=[-5, 5]))
m = apm + bfm
p = numpy.asarray([1.2, -0.1, 30, 1.0, 0.3], dtype=float)
N = 201
x = numpy.arange(N, dtype=float) / 25 - 2
y = (x - 0.70) / 0.4
y *= -0.5 * y
y = 8.0 * numpy.exp(y)
y += 0.2 * x + 1.0
numpy.random.seed(13456)
y += 0.5 * numpy.random.randn(N)
ns = NestedSampler(x, m, y)
ns.distribution.setLimits([0.01, 10])
ns.verbose = 2
evi = ns.sample(plot=self.doplot)
print("NS pars ", fmt(ns.parameters))
print("NS stdv ", fmt(ns.stdevs))
print("NS scal ", fmt(ns.scale))
def test2(self):
print("\n SalesmanProblem Test Euclidic\n")
np = 10
pars = numpy.arange(np * np, dtype=int)
problem = self.initProblem(np=np)
print(problem)
mdis = problem.minimumDistance()
print("mindis ", mdis)
dis = problem.result(pars)
sdis = numpy.sum(dis)
print(fmt(dis, max=None), fmt(sdis))
print(sdis, (90 + 9 * 9.05538514 + 12.72792206))
self.assertAlmostEqual(sdis, (90 + 9 * 9.05538514 + 12.72792206), 6)
problem = self.initProblem(np=np, weights=True)
dis = problem.result(pars)
sdis = numpy.sum(dis)
print(fmt(dis, max=None), fmt(sdis))
print(sdis, 2 * (90 + 9 * 9.05538514 + 12.72792206))
self.assertAlmostEqual(sdis, 2 * (90 + 9 * 9.05538514 + 12.72792206),
6)
def histo(self, x, pr, fun=None):
print(pr)
print(fmt(x))
print(fmt(fun[0]))
print(fmt(fun[1]))
if self.doplot:
num_bins = 80
# the histogram of the data
n, bins, patches = plt.hist(x,
num_bins,
facecolor='green',
alpha=0.5)
# add a 'best fit' line
if fun is not None:
plt.plot(fun[0], fun[1], 'r--')
plt.xlabel('Error')
plt.ylabel('Probability')
plt.title('Histogram of ' + str(pr))
# Tweak spacing to prevent clipping of ylabel
plt.subplots_adjust(left=0.15)
plt.show()
def testBernoulliErrorDistribution(self):
print(
"====== Test Bernoulli Error Distribution ======================")
poly = LogisticModel(fixed={0: 1})
param = numpy.asarray([0.5, 1.0], dtype=float)
data = numpy.asarray([0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1], dtype=int)
data = numpy.asarray([1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0], dtype=int)
print("x : ", fmt(self.x, max=None))
print("Data : ", fmt(data, max=None))
print("yfit : ", fmt(poly.result(self.x, param), max=None))
problem = ClassicProblem(model=poly, xdata=self.x, ydata=data)
ped = BernoulliErrorDistribution()
print(str(ped))
self.assertFalse(ped.acceptWeight())
logL = ped.logLikelihood(problem, param)
print("logL = %8.3f" % (logL))
logL = ped.logLikelihood(problem, param)
mok = problem.result(param)
altL = ped.logLikelihood_alt(problem, param)
lLdata = ped.logLdata(problem, param)
print("Ldata: ", fmt(lLdata, max=None))
logL0 = numpy.sum(lLdata)
print("logL = %8.3f ldata %8.3f alt %8.3f" % (logL, logL0, altL))
assertAAE(logL, altL)
assertAAE(logL, logL0)
print("logL = %8.3f %8.3f" % (logL, altL))
assertAAE(logL, altL)
fitIndex = [0, 1]
dL = ped.partialLogL(problem, param, fitIndex)
nL = ped.numPartialLogL(problem, param, fitIndex)
aL = ped.partialLogL_alt(problem, param, fitIndex)
print("partial = ", dL)
print("altpart = ", aL)
print("numpart = ", nL)
assertAAE(dL, nL, 5)
print(" ", fmt([4.0 / (k + 1) for k in range(9)],
max=None))
for i in range(10):
param = numpy.asarray([i - 4, 1], dtype=float)
print(fmt(param[0]), " : ", end="")
for k in range(9):
param[1] = 4.0 / (k + 1)
print(" %8.3f" % ped.logLikelihood(problem, param), end="")
print("")
def dofit(self, ns, pp, plot=False):
logE = ns.sample(plot=plot)
par = ns.parameters
std = ns.standardDeviations
mlp = ns.samples.maxLikelihoodParameters
scale = ns.scale
scdev = ns.stdevScale
print("truth ", fma(pp))
print("par ", fma(par))
print("st dev ", fma(std))
print("ML par ", fma(mlp))
print("scale ", fmt(scale))
# ns.report( )
self.assertTrue(all(numpy.abs(par - mlp) < 2 * std))
self.assertTrue(all(numpy.abs(par - pp) < 2 * std))
def test2(self):
print("===== formatter test2 ===========================")
arr = numpy.asarray([k for k in range(36)], dtype=float)
arr = arr.reshape((3, 12))
fmtinit(max=None)
print(fmt(arr, max=None))
fmtinit(linelength=60)
print(fmt(arr, max=None))
fmtinit(linelength=80, format={"float64": " %7.2f"})
print(fmt(arr, max=None))
print("arr", fmt(arr, indent=4, format=" %7.2f"))
print(fmt(arr, max=2))
alist = [1, 2, 3, 4, 5]
print(fmt(alist))
self.assertTrue(isinstance(alist, list))
print(fmt(3), fmt(3.4))
def test1(self):
print("===== formatter test1 ===========================")
arr = numpy.asarray([k for k in range(120)], dtype=float)
arr = arr.reshape((3, 40))
print(fmt(arr, max=None))
print("arr", fmt(arr[1], indent=4, format=" %7.2f", max=20))
print(fmt(arr.reshape((3, 4, 10)), max=2))
alist = [1, 2, 3, 4, 5]
print(fmt(alist))
self.assertTrue(isinstance(alist, list))
print(fmt(3), fmt(3.4))
def XXXtest6(self):
print("====test6 CategoricalProblem============================")
x = numpy.array([[1, 4, 2, 1, 3, 2, 4, 2, 3, 2, 1, 2, 3]],
dtype=float).transpose()
y = numpy.array([1, 2, 2, 3, 3, 2, 1, 1, 2, 3, 3, 2, 2])
# w = numpy.array( [0.9,1.0,1.1,1.2], dtype=float )
mdl = DecisionTreeModel(ndim=1,
depth=2,
kdim=[0, 0, 0],
split=0.5,
itypes=[0])
p = [0.1, 0.5, 0.6, 0.4, 0.2, 0.3, 0.7, 0.3, 0.7, 0.2, 0.1, 0.3]
print(mdl.fullName())
problem = CategoricalProblem(model=mdl, xdata=x, ydata=y)
print(problem.ncateg, problem.npars, fmt(p, max=None))
print(fmt(problem.ydata, max=None))
print(fmt(x, max=None))
print(fmt(mdl.sortXdata(x), max=None))
lst = [k for k in range(len(p))]
ym = problem.result(p)
yl = problem.result(lst)
dy = problem.partial(p)
print(fmt(ym, max=None))
print(fmt(yl, max=None))
print(fmt(dy, max=None))
res = problem.residuals(p)
print(fmt(res, max=None))
def test5a(self):
print("====test5a==discard===================")
plot = self.doplot
nn = 10
x = numpy.linspace(0, 2, nn, dtype=float)
ym = 0.3 + 0.5 * x
nf = 0.1
numpy.random.seed(2345)
noise = numpy.random.randn(nn)
y = ym + nf * noise
limits = [-1, 2]
model = PolynomialModel(1)
model.setLimits(lowLimits=limits[0], highLimits=limits[1])
s = 0.0
s2 = 0.0
mr = 5
for k in [1, 5, 10, 20, 40]:
dis = GaussErrorDistribution(scale=0.5)
ns = NestedSampler(x,
model,
y,
distribution=dis,
verbose=0,
discard=k)
logE = ns.sample()
par2 = ns.parameters
logz2 = ns.logZ
dlz2 = ns.logZprecision
s += logz2
s2 += logz2 * logz2
print(fmt(k), " pars ", fmt(par2), " logZ ", fmt(logz2),
" +- ", fmt(dlz2))
logz = s / mr
dlz = math.sqrt(s2 / mr - logz * logz)
print("Average ", fmt(logz), " +- ", fmt(dlz))
def test2(self):
plot = self.doplot
ndata = 101
c0 = 3.2
c1 = -0.1
c2 = 0.3
c3 = 1.1
c4 = 2.1
ss = 0.2
x = numpy.linspace(-1, +1, ndata, dtype=float)
y = (x - c1) / c2
numpy.random.seed(12345)
y = c0 * numpy.exp(
-0.5 * y * y) + x * c3 + c4 + ss * numpy.random.randn(ndata)
print("Testing Nonlinear Fitters with RobustShell.")
print(fmt([c0, c1, c2, c4, c3]))
modl2 = GaussModel()
modl2 += PolynomialModel(1)
# lmfit = CurveFitter( x, modl2 )
lmfit = LevenbergMarquardtFitter(x, modl2)
par2 = lmfit.fit(y)
std2 = lmfit.stdevs
print(fmt(par2, max=5))
print(fmt(std2, max=5))
# make some outliers
ko = [10 * k + 4 for k in range(10)]
ko = numpy.asarray(ko)
y[ko] += numpy.linspace(-5, 2, 10)
if plot:
plt.plot(x, y, 'k.')
plt.plot(x, modl2(x), 'k-')
# lmfit = CurveFitter( x, modl2 )
lmfit = LevenbergMarquardtFitter(x, modl2)
# lmfit.setParameters( initpar2 )
rf = RobustShell(lmfit)
# rf.setVerbose( 1 )
print(str(rf))
par1 = rf.fit(y)
std1 = rf.stdevs
print(fmt(par1, max=5))
print(fmt(std1, max=5))
print(fmt(par2, max=5))
print(fmt(rf.weights[ko], max=20))
if plot:
plt.plot(x, modl2(x), 'g-')
plt.plot(x, rf.weights, 'g-')
assertAAE(par2, par1, 1)
assertAAE(std2, std1, 1)
# lmfit = CurveFitter( x, modl2 )
lmfit = LevenbergMarquardtFitter(x, modl2)
# lmfit.setParameters( initpar2 )
rf = RobustShell(lmfit, onesided="negative")
# rf.setVerbose( 1 )
print(rf)
par3 = rf.fit(y)
std3 = rf.stdevs
print(fmt(par3, max=5))
print(fmt(std3, max=5))
print(fmt(par2, max=5))
print(fmt(rf.weights[ko], max=20))
if plot:
plt.plot(x, modl2(x), 'r-')
plt.plot(x, rf.weights, 'r-')
plt.show()
assertAAE(par2, par3, 1)
assertAAE(std2, std3, 1)
def test1(self):
""" # test slope fit """
print("\n Robust fitter Test 1 \n")
plot = self.doplot
ndata = 101
aa = 5.0 # average offset
bb = 2.0 # slope
ss = 0.3 # noise Normal distr.
x = numpy.linspace(-1, +1, ndata, dtype=float)
numpy.random.seed(12345)
y = aa + bb * x + ss * numpy.random.randn(ndata)
model = PolynomialModel(1)
fitter = Fitter(x, model)
print("Testing Straight Line fit")
par = fitter.fit(y)
std = fitter.stdevs
chisq = fitter.chisq
print("truth " + fmt(aa) + fmt(bb))
print("params " + fmt(par))
print("stdevs " + fmt(std))
print("chisq " + fmt(chisq))
assertAAE(par, numpy.asarray([aa, bb]), 2)
if plot:
plt.plot(x, y, 'k*')
plt.plot(x, model(x), 'k-')
# make some outliers
ko = [10 * k + 4 for k in range(10)]
ko = numpy.asarray(ko)
y[ko] += numpy.linspace(-5, 2, 10)
romod = PolynomialModel(1)
altfit = Fitter(x, romod)
alt = altfit.fit(y)
ast = altfit.stdevs
altch = altfit.chisq
print("params " + fmt(alt))
print("stdevs " + fmt(ast))
print("chisq " + fmt(altch))
if plot:
plt.plot(x, y, 'b.')
plt.plot(x, romod(x), 'b-')
rf = RobustShell(altfit)
Tools.printclass(rf)
alt = rf.fit(y)
ast = altfit.stdevs
altch = altfit.chisq
print(rf)
print("params " + fmt(alt))
print("stdevs " + fmt(ast))
print("chisq " + fmt(altch) + fmt(rf.scale))
print("weight " + fmt(rf.weights[ko], max=None))
if plot:
plt.plot(x, romod(x), 'g-')
plt.plot(x, rf.weights, 'g-')
assertAAE(par, alt, 1)
assertAAE(std, ast, 1)
rf = RobustShell(altfit, kernel=Cosine(), domain=4.0)
alt = rf.fit(y)
ast = altfit.stdevs
altch = altfit.chisq
print(rf)
print("params " + fmt(alt))
print("stdevs " + fmt(ast))
print("chisq " + fmt(altch) + fmt(rf.scale))
print("weight " + fmt(rf.weights[ko], max=None))
if plot:
plt.plot(x, romod(x), 'r-')
plt.plot(x, rf.weights, 'r-')
assertAAE(par, alt, 1)
assertAAE(std, ast, 1)
rf = RobustShell(altfit, kernel=Huber, domain=1.0)
# rf.setNoiseScale( 0.3 )
alt = rf.fit(y)
ast = altfit.stdevs
altch = altfit.chisq
print(rf)
print("params " + fmt(alt))
print("stdevs " + fmt(ast))
print("chisq " + fmt(altch) + fmt(rf.scale))
print("weight " + fmt(rf.weights[ko], max=None))
if plot:
plt.plot(x, romod(x), 'c-')
plt.plot(x, rf.weights, 'c-')
assertAAE(par, alt, 1)
assertAAE(std, ast, 1)
rf = RobustShell(altfit, kernel=Uniform)
alt = rf.fit(y)
ast = altfit.stdevs
altch = altfit.chisq
print(rf)
print("params " + fmt(alt))
print("stdevs " + fmt(ast))
print("chisq " + fmt(altch) + fmt(rf.scale))
print("weight " + fmt(rf.weights[ko], max=None))
if plot:
plt.plot(x, romod(x), 'm-')
plt.plot(x, rf.weights, 'm-')
plt.show()
assertAAE(par, alt, 1)
assertAAE(std, ast, 1)
def test5(self):
print("5 Test Dynamic Modifiable Splines: evidence with fixed scale ")
t, y = self.makeData()
# npt = 201
# t = numpy.linspace( 0, 100, npt, dtype=float )
# ym = 2 * numpy.sin( 2 * math.pi * numpy.exp( t / 80 ) +1 ) * numpy.exp( -0.02 * t )
# y = numpy.random.seed( 12345 )
# y = ym + numpy.random.randn( npt ) * 0.05
knots = [0, 100]
mxk = 15
mdl = SplinesDynamicModel(knots=knots,
dynamic=True,
maxKnots=mxk,
minDistance=0.04)
mdl.setLimits(lowLimits=[-10.0], highLimits=[+10.0])
ep = EvidenceProblem(model=mdl, xdata=t, ydata=y)
distr = ModelDistribution(scale=0.1)
ns = NestedSampler(problem=ep, distribution=distr)
ns.verbose = 2
ns.minimumIterations = 2000
## Comment next if-statement out for a full run of NestedSampler
if not self.dofull:
ns.ensemble = 10
ns.minimumIterations = 500
evid = ns.sample(plot=self.doplot)
if not self.doplot: return
## Plot the evolutie of knots and sample weights
cc = ['k,', 'b,', 'r,', 'g,', 'c,', 'm,']
sl = ns.samples
ka = numpy.zeros((mxk, len(sl)), dtype=float)
plt.figure(1, figsize=[14, 8])
for k, s in enumerate(sl):
n = len(s.model.knots)
ka[:n, k] = s.model.knots
for j in range(mxk):
plt.plot(ka[j, :], cc[j % 6])
wgts = sl.getWeightEvolution()
mw = max(wgts)
plt.plot(100 * wgts / mw, 'k-')
plt.xlabel("Iteration number")
plt.ylabel("Knot position cq. sample weight")
plt.show()
plt.figure(2, figsize=[14, 8])
plt.plot(t, y, 'k.')
cc = ['k-', 'b-', 'r-', 'g-', 'c-', 'm-', 'y-']
cp = ['k*', 'b*', 'r*', 'g*', 'c*', 'm*', 'y*']
for k in range(0, 2100, 200):
kc = k % 7
s = sl[k]
mdl = s.model
kn = mdl.knots
plt.plot(kn, [-1 + k / 1000] * len(kn), cp[kc])
plt.plot(t, mdl(t), cc[kc])
print(fmt(k))
print(fmt(kn, max=None))
print(fmt(mdl.parameters, max=None), fmt(s.logL))
m = BasicSplinesModel(knots=kn)
ftr = Fitter(t, m, fixedScale=0.05)
par = ftr.fit(y)
lz0 = ftr.getLogZ(limits=[-10, +10])
print(fmt(par, max=None), fmt(lz0))
m = BasicSplinesModel(knots=kn)
ftr = Fitter(t, m)
par = ftr.fit(y)
lz0 = ftr.getLogZ(limits=[-10, +10])
print(fmt(par, max=None), fmt(lz0))
m = BasicSplinesModel(knots=kn)
ftr = Fitter(t, m)
par = ftr.fit(y)
lz0 = ftr.getLogZ(limits=[-10, +10], noiseLimits=[0.01, 1])
print(fmt(par, max=None), fmt(lz0))
plt.show()
def XXXtestNestedSampler2( self ) :
print( "========================" )
print( " Test Nested Sampler 2" )
print( "========================" )
ND = 10
NP = 100
numpy.random.seed( 12345 )
xdata = numpy.random.rand( NP, ND ) * 10
print( fmt( xdata ) )
y = 0.5 * xdata[:,1]
print( xdata.shape, y.shape )
y += numpy.where( xdata[:,4] > 3, 5, 0 )
y += numpy.where( xdata[:,6] > 8, 0, 3 )
y += numpy.where( xdata[:,4] > 7, 2, 0 )
noise = numpy.random.randn( NP )
y += 0.02 * noise
pm = DecisionTreeModel( ndim=ND, kdim=[0], depth=0 )
print( pm )
lolim = [-10]
hilim = [+10]
pm.setLimits( lowLimits=lolim, highLimits=hilim )
Tools.printclass( pm )
ed = ModelLikelihood( limits=[0.01,100] )
engs = ["birth", "death", "struct"]
# engs = ["birth", "struct"]
ns = NestedSampler( xdata, pm, y, ensemble=100, distribution=ed,
engines=engs, seed=2031967 )
# ns.minimumIterations = 4000
ns.verbose = 2
# ns.weed = 10000
logE = ns.sample( )
sl = ns.samples
yfit = sl.average( xdata )
ksrt = numpy.argsort( yfit )
xxx = numpy.arange( NP, dtype=float )
mi = sl.medianIndex
print( mi, sl.modusIndex, len( sl ) )
if not self.doplot : return
plt.plot( xxx, y[ksrt], 'k.' )
plt.plot( xxx, yfit[ksrt], 'r-' )
for k in range( mi-3, mi+4 ) :
mdl = sl[k].model
print( mdl.fullName() )
ymod = mdl.result( xdata, sl[k].parameters )
plt.plot( xxx, ymod[ksrt], 'g-' )
print( ymod )
print( ksrt )
plt.show()
print( ymod[ksrt] )
def testNestedSampler( self ) :
print( "========================" )
print( " Test Nested Sampler" )
print( "========================" )
ND = 10
NP = 100
numpy.random.seed( 12345 )
xdata = numpy.random.rand( NP, ND ) * 10
print( fmt( xdata ) )
y = 0.5 * xdata[:,1]
print( xdata.shape, y.shape )
y += numpy.where( xdata[:,4] > 3, 5, 0 )
y += numpy.where( xdata[:,6] > 8, 0, 3 )
y += numpy.where( xdata[:,4] > 7, 2, 0 )
noise = numpy.random.randn( NP )
y += 0.02 * noise
pm = DecisionTreeModel( ndim=ND, kdim=[0], depth=0 )
print( pm )
lolim = [-10]
hilim = [+10]
pm.setLimits( lowLimits=lolim, highLimits=hilim )
Tools.printclass( pm )
ns = NestedSampler( xdata, pm, y, ensemble=100, seed=2031967 )
ns.verbose = 2
ns.weed = 10000
ns.distribution.setLimits( [0.01, 100] )
if not self.dofull :
ns.ensemble = 10
logE = ns.sample( )
sl = ns.samples
yfit = sl.average( xdata )
# ksrt = numpy.argsort( yfit )
xxx = numpy.arange( NP, dtype=float )
mi = sl.medianIndex
print( mi, sl.modusIndex, len( sl ) )
if not self.doplot : return
mdl = sl[mi].model
ksrt = mdl.sortXdata( xdata )
plt.plot( xxx, y[ksrt], 'k.' )
plt.plot( xxx, yfit[ksrt], 'r-' )
for k in range( mi-3, mi+4 ) :
mdl = sl[k].model
print( mdl.fullName() )
ymod = mdl.result( xdata, sl[k].parameters )
plt.plot( xxx, ymod[ksrt], 'g-' )
print( ymod )
print( ksrt )
print( ymod[ksrt] )
plt.show()
def test6( self ):
print( " Test 6 DecisionTreeModel" )
dtm = DecisionTreeModel( ndim=20, kdim=[0,1,2,3,4,5,6],
split=[0.1*k for k in range(2,8)], depth=3 )
dtm.parameters = numpy.arange( dtm.npars, dtype=float ) + 1
print( dtm.fullName() )
print( fmt( dtm.parameters, max=None ) )
self.assertTrue( dtm.npbase == dtm.npars )
self.assertTrue( dtm.npars == dtm.countLeaf() )
self.assertTrue( dtm.ncomp == dtm.countBranch() )
print( "GROW at location 2 dim 3 split 0.44" )
dtm.grow( offset=0, location=2, kdim=3, split=0.44 )
print( dtm )
print( fmt( dtm.parameters, max=None ) )
self.assertTrue( dtm.npbase == dtm.npars )
self.assertTrue( dtm.npars == dtm.countLeaf() )
self.assertTrue( dtm.ncomp == dtm.countBranch() )
print( "VARY at location 4, dim=5, split=0.55" )
dtm.vary( location=4, kdim=5, split=0.55 )
print( dtm )
print( fmt( dtm.parameters, max=None ) )
self.assertTrue( dtm.npbase == dtm.npars )
self.assertTrue( dtm.npars == dtm.countLeaf() )
self.assertTrue( dtm.ncomp == dtm.countBranch() )
print( "GROW at location 8" )
dtm.grow( offset=0, location=8 )
print( dtm )
print( fmt( dtm.parameters, max=None ) )
self.assertTrue( dtm.npbase == dtm.npars )
self.assertTrue( dtm.npars == dtm.countLeaf() )
self.assertTrue( dtm.ncomp == dtm.countBranch() )
self.assertTrue( len( dtm.parNames ) == dtm.npars )
print( "SHRINK at location 7" )
dtm.shrink( offset=0, location=7 )
print( dtm )
print( fmt( dtm.parameters, max=None ) )
self.assertTrue( dtm.npbase == dtm.npars )
self.assertTrue( dtm.npars == dtm.countLeaf() )
self.assertTrue( dtm.ncomp == dtm.countBranch() )
print( "SHRINK at location 3" )
dtm.shrink( offset=0, location=3 )
print( dtm )
print( fmt( dtm.parameters, max=None ) )
self.assertTrue( dtm.npbase == dtm.npars )
self.assertTrue( dtm.npars == dtm.countLeaf() )
self.assertTrue( dtm.ncomp == dtm.countBranch() )
print( "SHRINK at location 0" )
dtm.shrink( offset=0, location=0 )
print( dtm.fullName() )
print( fmt( dtm.parameters, max=None ) )
self.assertTrue( dtm.npbase == dtm.npars )
self.assertTrue( dtm.npars == dtm.countLeaf() )
self.assertTrue( dtm.ncomp == dtm.countBranch() )
printclass( dtm )
def test7(self):
print("7 Test Dynamic Modifiable Splines: evidence unknown scale")
t, y = self.makeData()
knots = [0, 100]
mxk = 15
mdl = SplinesDynamicModel(knots=knots,
dynamic=True,
maxKnots=mxk,
minDistance=0.01)
mdl.setLimits(lowLimits=[-10.0], highLimits=[+10.0])
ep = EvidenceProblem(model=mdl, xdata=t, ydata=y)
distr = ModelDistribution(limits=[0.01, 1])
#ns = NestedSampler( t, mdl, y, seed=1235, engines=eng, distribution=distr )
#print( ns.problem )
#print( ns.model )
#print( fmt( ns.xdata ) )
#print( fmt( ns.ydata ) )
ns = NestedSampler(problem=ep, distribution=distr)
print(ns.problem)
print(ns.model)
print(fmt(ns.xdata))
print(fmt(ns.ydata))
ns.verbose = 2
## Comment next if-statement out for a full run of NestedSampler
if not self.dofull:
ns.ensemble = 10
ns.minimumIterations = 500
evid = ns.sample(plot=self.doplot)
if not self.doplot: return
## Plot the evolutie of knots and sample weights
cc = ['k,', 'b,', 'r,', 'g,', 'c,', 'm,']
sl = ns.samples
ka = numpy.zeros((mxk, len(sl)), dtype=float)
plt.figure(1, figsize=[14, 8])
for k, s in enumerate(sl):
n = len(s.model.knots)
ka[:n, k] = s.model.knots
for j in range(mxk):
plt.plot(ka[j, :], cc[j % 6])
wgts = sl.getWeightEvolution()
mw = max(wgts)
plt.plot(100 * wgts / mw, 'k-')
plt.xlabel("Iteration number")
plt.ylabel("Knot position cq. sample weight")
plt.show()
plt.figure(2, figsize=[14, 8])
plt.plot(t, y, 'k.')
cc = ['k-', 'b-', 'r-', 'g-', 'c-', 'm-', 'y-']
cp = ['k*', 'b*', 'r*', 'g*', 'c*', 'm*', 'y*']
for k in range(0, 2100, 200):
kc = k % 7
s = sl[k]
mdl = s.model
kn = mdl.knots
plt.plot(kn, [-1 + k / 1000] * len(kn), cp[kc])
plt.plot(t, mdl(t), cc[kc])
print(fmt(k))
print(fmt(kn, max=None))
print(fmt(mdl.parameters, max=None), fmt(s.logL))
plt.show()
