def kaiser_bessel(kwidth, grid_mod, over_samp = 1.375):
# kernel = np.zeros(krad * grid_mod)
beta = np.pi * np.sqrt((kwidth * kwidth) / (over_samp * over_samp)
* (over_samp - 0.5) * (over_samp - 0.5) - 0.8)
print beta
x = np.linspace(0, kwidth, grid_mod)
x1 = np.sqrt(1 - (x / kwidth) ** 2)
y = np.i0(beta * x1)
y = y / y[0]
# pl.plot(x, y)
# pl.show()
fx = np.arange(192)
fx = fx - 192/2.0
fy = np.zeros(192, 'c16')
deapp = np.sqrt(((np.pi * fx / kwidth)**2 - beta**2).astype('c16'))
fy = np.sin(deapp)/deapp
return(x, y, fy)
def mk_kbd_window(alpha, N): n = numpy.arange(0., N/2) n = (n - N/4)/(N/4) n = n * n W = numpy.i0(numpy.pi*alpha*numpy.sqrt(1.0 - n)) W = numpy.cumsum(W) W = numpy.sqrt(W / W[-1]) return W
def tlsAndMBT(params, temps, data, eps = None):
"""Return residual of model, weighted by uncertainties.
Return value:
residual -- the weighted or unweighted vector of residuals
Arguments:
params -- an lmfit Parameters object containing df, Fd, fRef, alpha, delta0
temps -- a list/vector of temperatures at which to evaluate the model
data -- a list/vector of data to compare with model
eps -- a list/vector of uncertainty values for data
len(temps) == len(data) == len(eps)"""
#Unpack parameter values from params
df = params['df'].value
Fd = params['Fd'].value
fRef = params['fRef'].value
alpha = params['alpha'].value
delta0 = params['delta0'].value
tc = params['tc'].value
dtc = params['dtc'].value
zn = params['zn'].value
#Convert fRef to f0
f0 = fRef-df
#Calculate model from parameters
model = -df/(f0+df)+(f0/(f0+df))*(Fd/sc.pi* #TLS contribution
(np.real(digamma(0.5+sc.h*f0/(1j*2*sc.pi*sc.k*(temps))))
-np.log(sc.h*f0/(2*sc.pi*sc.k*(temps))))-
alpha/4.0* #MBD contribution
(np.sqrt((2*sc.pi*sc.k*temps)/delta0)*
np.exp(-delta0/(sc.k*temps))+
2*np.exp(-delta0/(sc.k*temps))*
np.exp(-sc.h*f0/(2*sc.k*temps))*
np.i0(sc.h*f0/(2*sc.k*temps)))+
0.5*zn*(1+np.tanh((temps-tc)/dtc)))
#Weight the residual if eps is supplied
if data is not None:
if eps is not None:
residual = (model-data)/eps
else:
residual = (model-data)
return residual
else:
return model
def kaiser_bessel2(kwidth, grid_mod, over_samp, imsize=128):
kw0 = kwidth/over_samp
kr = kwidth/2.0
beta = np.pi*np.sqrt((kw0*(over_samp-0.5))**2-0.8)
# print beta
kosf = np.floor(0.91/(over_samp*1e-3));
# print kr
# print kosf
# print kosf*kr
om = np.arange(kosf*kr)/(kosf*kr)
# print om
x = np.linspace(0, kr, grid_mod)
x_bess = np.sqrt(1 - (x / kr) ** 2)
# print x_bess
y = np.i0(beta * x_bess)
y = y / y[0]
# pl.plot(y)
# pl.show()
fx = np.arange(imsize*over_samp)
fx = fx - imsize*over_samp/2
fx = fx / imsize * 1.5
# print fx
sqa = np.sqrt((np.pi*np.pi*kw0*kw0*fx*fx-beta*beta).astype('c16'))
# print sqa
fy = np.sin(sqa)/sqa
fy = abs(fy)
fy = fy / fy.max()
# print fy
# pl.plot(fy)
# pl.show()
return (x,y,fx,fy)
def calc_kernel_kb(self, grid_params):
kw0 = 2.0 * grid_params.krad / grid_params.over_samp
kr = grid_params.krad
beta = np.pi * \
np.sqrt((kw0 * (grid_params.over_samp - 0.5)) ** 2 - 0.8)
x = np.linspace(0, kr, grid_params.grid_mod)
print x
x_bess = np.sqrt(1 - (x / kr) ** 2)
y = np.i0(beta * x_bess)
y = y / y[0]
x = np.concatenate((x, np.array([0.0, ])))
y = np.concatenate((y, np.array([0.0, ])))
self.kx = x
self.ky = y
pl.figure()
pl.plot(y)
def KBDWindow(dataSampleArray,alpha=4.):
"""
Returns a copy of the dataSampleArray KBD-windowed
KBD window is defined following pp. 108-109 and pp. 117-118 of
Bosi & Goldberg, "Introduction to Digital Audio..." book
"""
N = float(len(dataSampleArray))
t = arange(int(N/2 + 1))
input = dataSampleArray.copy()
# i0 --> 0th modified bessel function
kaiser = i0(alpha * pi * sqrt(1.0 - (4.0 * t / N - 1.0) ** 2)) / np.i0(np.pi * alpha)
denominator = sum(kaiser ** 2)
numerator = cumsum(kaiser[:-1] ** 2)
numerator = concatenate((numerator, numerator[::-1]), axis=0)
window = sqrt(numerator / denominator)
out = input * window
return out
def importanceSampling(self, K):
# Init variables
nx = self.sz
ny = self.sz
logZ = 0.
samples = np.zeros( (K, nx, ny) )
evalPQ = np.zeros( K )
for iX in range(nx):
for iY in range(ny):
kappa = 0.
logZ += np.log( 2 * np.pi * np.i0(kappa) )
samples[:, iX, iY] = np.random.vonmises(0, kappa, K)
#print logZ
for iSample in range(K):
for iX in range(nx):
for iY in range(ny):
curInd = hlp.ravel_multi_index( (iX, iY), (nx,ny) )
if iX > 0:
neiInd = hlp.ravel_multi_index( (iX-1, iY), (nx,ny) )
evalPQ[iSample] += self.J * np.cos( samples[iSample, iX, iY] - samples[iSample, iX-1, iY] )
else:
neiInd = hlp.ravel_multi_index( (nx-1, iY), (nx,ny) )
evalPQ[iSample] += self.J * np.cos( samples[iSample, iX, iY] - samples[iSample, nx-1, iY] )
if iY > 0:
neiInd = hlp.ravel_multi_index( (iX, iY-1), (nx,ny) )
evalPQ[iSample] += self.J * np.cos( samples[iSample, iX, iY] - samples[iSample, iX, iY-1] )
else:
neiInd = hlp.ravel_multi_index( (iX, ny-1), (nx,ny) )
evalPQ[iSample] += self.J * np.cos( samples[iSample, iX, iY] - samples[iSample, iX, ny-1] )
evalMax = np.max( evalPQ )
evalPQ = np.exp( evalPQ - evalMax )
logZ += evalMax - np.log(K) + np.log( np.sum( evalPQ ) )
return logZ
def sample(self):
if self.tr_cond == 'all_gains':
G = (1.0 / self.stim_dur) * np.random.choice(
[1.0], size=(self.n_loc, self.batch_size))
G = np.repeat(G, self.n_in, axis=0).T
G = np.tile(G, (self.stim_dur, 1, 1))
G = np.swapaxes(G, 0, 1)
else:
G = (0.5 / self.stim_dur) * np.random.choice(
[1.0], size=(1, self.batch_size))
G = np.repeat(G, self.n_in * self.n_loc, axis=0).T
G = np.tile(G, (self.stim_dur, 1, 1))
G = np.swapaxes(G, 0, 1)
H = (1.0 / self.resp_dur) * np.ones(
(self.batch_size, self.resp_dur, self.nneuron))
# Target presence/absence and stimuli
C = np.random.choice([0.0, 1.0], size=(self.batch_size, ))
C1ind = np.where(C == 1.0)[0] # change
S1 = np.pi * np.random.rand(self.n_loc, self.batch_size)
S2 = S1
S1 = np.repeat(S1, self.n_in, axis=0).T
S1 = np.tile(S1, (self.stim_dur, 1, 1))
S1 = np.swapaxes(S1, 0, 1)
S2[np.random.randint(0, self.n_loc, size=(len(C1ind), )),
C1ind] = np.pi * np.random.rand(len(C1ind))
S2 = np.repeat(S2, self.n_in, axis=0).T
S2 = np.tile(S2, (self.resp_dur, 1, 1))
S2 = np.swapaxes(S2, 0, 1)
# Noisy responses
L1 = G * np.exp(self.kappa * (np.cos(2.0 * (S1 - np.tile(
self.phi,
(self.batch_size, self.stim_dur, self.n_loc)))) - 1.0)) # stim 1
L2 = H * np.exp(self.kappa * (np.cos(2.0 * (S2 - np.tile(
self.phi,
(self.batch_size, self.resp_dur, self.n_loc)))) - 1.0)) # stim 2
Ld = (self.spon_rate / self.delay_dur) * np.ones(
(self.batch_size, self.delay_dur, self.nneuron)) # delay
R1 = np.random.poisson(L1)
R2 = np.random.poisson(L2)
Rd = np.random.poisson(Ld)
example_input = np.concatenate((R1, Rd, R2), axis=1)
example_output = np.repeat(C[:, np.newaxis], self.total_dur, axis=1)
example_output = np.repeat(example_output[:, :, np.newaxis], 1, axis=2)
cum_R1 = np.sum(R1, axis=1)
cum_R2 = np.sum(R2, axis=1)
mu_x = np.asarray([
np.arctan2(
np.dot(cum_R1[:, i * self.n_in:(i + 1) * self.n_in],
np.sin(2.0 * self.phi)),
np.dot(cum_R1[:, i * self.n_in:(i + 1) * self.n_in],
np.cos(2.0 * self.phi))) for i in range(self.n_loc)
])
mu_y = np.asarray([
np.arctan2(
np.dot(cum_R2[:, i * self.n_in:(i + 1) * self.n_in],
np.sin(2.0 * self.phi)),
np.dot(cum_R2[:, i * self.n_in:(i + 1) * self.n_in],
np.cos(2.0 * self.phi))) for i in range(self.n_loc)
])
temp_x = np.asarray([
np.swapaxes(np.multiply.outer(cum_R1, cum_R1), 1, 2)[i, i, :, :]
for i in range(self.batch_size)
])
temp_y = np.asarray([
np.swapaxes(np.multiply.outer(cum_R2, cum_R2), 1, 2)[i, i, :, :]
for i in range(self.batch_size)
])
kappa_x = np.asarray([
np.sqrt(
np.sum(temp_x[:, i * self.n_in:(i + 1) * self.n_in,
i * self.n_in:(i + 1) * self.n_in] *
np.repeat(np.cos(
np.subtract(
np.expand_dims(self.phi, axis=1),
np.expand_dims(self.phi,
axis=1).T))[np.newaxis, :, :],
self.batch_size,
axis=0),
axis=(1, 2))) for i in range(self.n_loc)
])
kappa_y = np.asarray([
np.sqrt(
np.sum(temp_y[:, i * self.n_in:(i + 1) * self.n_in,
i * self.n_in:(i + 1) * self.n_in] *
np.repeat(np.cos(
np.subtract(
np.expand_dims(self.phi, axis=1),
np.expand_dims(self.phi,
axis=1).T))[np.newaxis, :, :],
self.batch_size,
axis=0),
axis=(1, 2))) for i in range(self.n_loc)
])
if self.n_loc == 1:
d = np.i0(kappa_x) * np.i0(kappa_y) / np.i0(
np.sqrt(kappa_x**2 + kappa_y**2 +
2.0 * kappa_x * kappa_y * np.cos(mu_y - mu_x)))
else:
d = np.nanmean(np.i0(kappa_x) * np.i0(kappa_y) / np.i0(
np.sqrt(kappa_x**2 + kappa_y**2 +
2.0 * kappa_x * kappa_y * np.cos(mu_y - mu_x))),
axis=0)
P = d / (d + 1.0)
return example_input, example_output, C, P
import numpy as np from pylab import * x = np.linspace(0, 4, 100) vals = np.i0(x) plot(x, vals) show()
reason="NEP-18 support is not available in NumPy"),
),
pytest.param(
lambda x: np.imag(x),
marks=pytest.mark.skipif(
not IS_NEP18_ACTIVE,
reason="NEP-18 support is not available in NumPy"),
),
pytest.param(
lambda x: np.fix(x),
marks=pytest.mark.skipif(
not IS_NEP18_ACTIVE,
reason="NEP-18 support is not available in NumPy"),
),
pytest.param(
lambda x: np.i0(x.reshape((24, ))),
marks=pytest.mark.skipif(
not IS_NEP18_ACTIVE,
reason="NEP-18 support is not available in NumPy"),
),
pytest.param(
lambda x: np.sinc(x),
marks=pytest.mark.skipif(
not IS_NEP18_ACTIVE,
reason="NEP-18 support is not available in NumPy"),
),
pytest.param(
lambda x: np.nan_to_num(x),
marks=pytest.mark.skipif(
not IS_NEP18_ACTIVE,
reason="NEP-18 support is not available in NumPy"),
def vonmises(x, amp, cen, kappa):
# "1-d vonmises"
return (amp / (np.pi * 2 * np.i0(kappa))) * np.exp(kappa * np.cos(x - cen))
# print('kappa estimate with formula = ',k_est)
# use fsolve, to estimate 'kappa', when 'mu' has been estimated through
# equation (2.4) of Banerjee(2005):
#r_min = optimize.fsolve(myF_kappa, [0.0], args=r_bar)
## r_min = optimize.fmin_bfgs(myF, (0,0))
#print(r_min)
#
## use fsolve, to estimate 'kappa' and 'mu':
#initParams = [ 1, 1 ]
#results = optimize.minimize(vmf_log2, initParams, method='nelder-mead')
#print(results.x)
# ---------------------------------------------------------------------- #
# define the von Mises with an expression:
num = 1.0
denom = (2 * np.pi) * i0(kappa_)
conss = num / denom
y11 = stats.vonmises.logpdf(X_samples, kappa=kappa_, loc=loc_)
y22 = np.log(conss * np.exp(kappa_ * CS.dot(loc_cs).T))
# plot the logarithm of the pdf of the von Mises distribution sample:
fig, ax = plt.subplots(1, 1, figsize=(8, 4))
ax.plot(y11, 'r', label='yPred')
ax.plot(y22, 'b', label='yObs')
ax.legend()
sy1 = -np.sum(y11)
sy2 = -np.sum(y22)
#return X, X_samples
def bessel_(x):
bessel_.NumParam = 0
bessel_.NumVars = 1
return np.i0(x)
#width = pulseWidth(period, meanFref,randomNum)
count = 0
#1/width/width > 100
#exp(-pow(phase,2)/2.0/width/width)/width/sqrt(2.* 3.14159265) * randarr[phasenum]* exp(gasdev(&idum)) / e;
profile = np.zeros(phaseNum)
for i in range(phaseNum):
tval = i*tsamp
phase = tval/period % 1.
if (phase < 0.) : phase += 1.
phase -= 0.5;
#phasenum = (int)(tval/head->p0-fmod(tval/head->p0, 1.));
if 1/width/width > 100:
profile[i] = math.exp(-pow(phase,2)/2.0/width/width)/width/np.sqrt(2.* math.pi) #* randarr[phasenum]* exp(gasdev(&idum)) / e
else:
profile[i] = math.exp((1/width/width) * math.cos(phase))/2./math.pi/np.i0((1/width/width)) # * randarr[phasenum] * exp(gasdev(&idum)) / e;
for i in range(phaseNum):
if profile[i] >= max(profile)*0.7:
count += 1
#profile[i] = math.exp((1/width/width) * math.cos(phase))/2./math.pi/np.i0((1/width/width)) # * randarr[phasenum] * exp(gasdev(&idum)) / e;
print sum(profile)*1./phaseNum
profile = profile *0.001/ (sum(profile)*1./phaseNum)
print count,len(profile),count*1./float(len(profile))
print int(period / tsamp)
print 'width: %s' % (np.size(np.where(profile > 0.00001))/(phaseNum*1.))
print 'num of profile > 0: %s' %(np.size(np.where(profile > 0.00001)))
print 'width: %s, phaseNum: %s' %(width, phaseNum)
plt.plot(profile)
#plt.scatter(np.arange(len(profile)),profile)
plt.xlabel('phaseNum')
def smc(self, order, N, NT=1.1, resamp='mult', verbose=False):
"""
SMC algorithm to estimate (log)Z of the classical XY model with
free boundary conditions.
Parameters
----------
order : 1-D array_like
The order in which to add the random variables x, flat index.
N : int
The number of particles used to estimate Z.
NT : int
Threshold for ESS-based resampling (0,1] or 1.1. Resample if ESS < NT*N (NT=1.1, resample every time)
resamp : string
Type of resampling scheme {mult, res, strat, sys}.
verbose : bool
Output changed to logZk, xMean, ESS.
Output
------
logZ : float
Estimate of (log)Z in double precision.
"""
# Init variables
nx = self.sz
ny = self.sz
logZ = np.zeros( nx*ny )
indSorted = order.argsort()
orderSorted = order[indSorted]
# SMC specific
trajectory = np.zeros( (N, len(order)) )
ancestors = np.zeros( N, np.int )
nu = np.zeros( N )
tempNu = np.zeros( N )
ess = np.zeros( len(order)-1 )
iter = 0
# -------
# SMC
# -------
# First iteration
ix, iy = hlp.unravel_index( order[0], (nx,ny) )
tempMean = 0.
tempDispersion = 0.
trajectory[:,0] = np.random.vonmises(tempMean, tempDispersion, N)
# Log-trick update of adjustment multipliers and logZ
tempDispersion = np.zeros(N)
for iSMC in range( 1, len(order) ):
# Resampling with log-trick update
nu += np.log(2 * np.pi * np.i0(tempDispersion))
nuMax = np.max(nu)
tempNu = np.exp( nu - nuMax )
c = np.sum(tempNu)
tempNu /= c
ess[iSMC-1] = 1 / (np.sum(tempNu**2))
if ess[iSMC-1] < NT*float(N):
nu = np.exp( nu - nuMax )
if iter > 0:
logZ[iter] = logZ[iter-1] + nuMax + np.log( np.sum(nu) ) - np.log(N)
else:
logZ[iter] = nuMax + np.log( np.sum(nu) ) - np.log(N)
c = np.sum(nu)
nu /= c
ancestors = hlp.resampling( nu, scheme=resamp )
nu = np.zeros( N )
trajectory[:,:iSMC] = trajectory[ancestors, :iSMC]
iter += 1
# Calculate optimal proposal and adjustment multipliers
ix, iy = hlp.unravel_index( order[iSMC], (nx,ny) )
tempMean = np.zeros( N )
tempDispersion = np.zeros( N )
if ix > 0:
tempInd = hlp.ravel_multi_index( (ix-1,iy), (nx,ny) )
if tempInd in order[:iSMC]:
kappa = self.J
tempIndSMC = indSorted[orderSorted.searchsorted(tempInd)]
Y = tempDispersion*np.sin( -tempMean ) + kappa*np.sin( -trajectory[:,tempIndSMC] )
X = tempDispersion*np.cos( -tempMean ) + kappa*np.cos( -trajectory[:,tempIndSMC] )
tempDispersion = np.sqrt( tempDispersion**2 + kappa**2 + 2*kappa*tempDispersion*np.cos(-tempMean+trajectory[:,tempIndSMC] ) )
tempMean = -np.arctan2(Y ,X)
else:
tempInd = hlp.ravel_multi_index( (nx-1,iy), (nx,ny) )
if tempInd in order[:iSMC]:
kappa = self.J
tempIndSMC = indSorted[orderSorted.searchsorted(tempInd)]
Y = tempDispersion*np.sin( -tempMean ) + kappa*np.sin( -trajectory[:,tempIndSMC] )
X = tempDispersion*np.cos( -tempMean ) + kappa*np.cos( -trajectory[:,tempIndSMC] )
tempDispersion = np.sqrt( tempDispersion**2 + kappa**2 + 2*kappa*tempDispersion*np.cos(-tempMean+trajectory[:,tempIndSMC] ) )
tempMean = -np.arctan2(Y ,X)
if ix < nx-1:
tempInd = hlp.ravel_multi_index( (ix+1,iy), (nx,ny) )
if tempInd in order[:iSMC]:
kappa = self.J
tempIndSMC = indSorted[orderSorted.searchsorted(tempInd)]
Y = tempDispersion*np.sin( -tempMean ) + kappa*np.sin( -trajectory[:,tempIndSMC] )
X = tempDispersion*np.cos( -tempMean ) + kappa*np.cos( -trajectory[:,tempIndSMC] )
tempDispersion = np.sqrt( tempDispersion**2 + kappa**2 + 2*kappa*tempDispersion*np.cos(-tempMean+trajectory[:,tempIndSMC] ) )
tempMean = -np.arctan2(Y ,X)
else:
tempInd = hlp.ravel_multi_index( (0,iy), (nx,ny) )
if tempInd in order[:iSMC]:
kappa = self.J
tempIndSMC = indSorted[orderSorted.searchsorted(tempInd)]
Y = tempDispersion*np.sin( -tempMean ) + kappa*np.sin( -trajectory[:,tempIndSMC] )
X = tempDispersion*np.cos( -tempMean ) + kappa*np.cos( -trajectory[:,tempIndSMC] )
tempDispersion = np.sqrt( tempDispersion**2 + kappa**2 + 2*kappa*tempDispersion*np.cos(-tempMean+trajectory[:,tempIndSMC] ) )
tempMean = -np.arctan2(Y ,X)
if iy > 0:
tempInd = hlp.ravel_multi_index( (ix,iy-1), (nx,ny) )
if tempInd in order[:iSMC]:
kappa = self.J
tempIndSMC = indSorted[orderSorted.searchsorted(tempInd)]
Y = tempDispersion*np.sin( -tempMean ) + kappa*np.sin( -trajectory[:,tempIndSMC] )
X = tempDispersion*np.cos( -tempMean ) + kappa*np.cos( -trajectory[:,tempIndSMC] )
tempDispersion = np.sqrt( tempDispersion**2 + kappa**2 + 2*kappa*tempDispersion*np.cos(-tempMean+trajectory[:,tempIndSMC] ) )
tempMean = -np.arctan2(Y ,X)
else:
tempInd = hlp.ravel_multi_index( (ix,ny-1), (nx,ny) )
if tempInd in order[:iSMC]:
kappa = self.J
tempIndSMC = indSorted[orderSorted.searchsorted(tempInd)]
Y = tempDispersion*np.sin( -tempMean ) + kappa*np.sin( -trajectory[:,tempIndSMC] )
X = tempDispersion*np.cos( -tempMean ) + kappa*np.cos( -trajectory[:,tempIndSMC] )
tempDispersion = np.sqrt( tempDispersion**2 + kappa**2 + 2*kappa*tempDispersion*np.cos(-tempMean+trajectory[:,tempIndSMC] ) )
tempMean = -np.arctan2(Y ,X)
if iy < ny-1:
tempInd = hlp.ravel_multi_index( (ix,iy+1), (nx,ny) )
if tempInd in order[:iSMC]:
kappa = self.J
tempIndSMC = indSorted[orderSorted.searchsorted(tempInd)]
Y = tempDispersion*np.sin( -tempMean ) + kappa*np.sin( -trajectory[:,tempIndSMC] )
X = tempDispersion*np.cos( -tempMean ) + kappa*np.cos( -trajectory[:,tempIndSMC] )
tempDispersion = np.sqrt( tempDispersion**2 + kappa**2 + 2*kappa*tempDispersion*np.cos(-tempMean+trajectory[:,tempIndSMC] ) )
tempMean = -np.arctan2(Y ,X)
else:
tempInd = hlp.ravel_multi_index( (ix,0), (nx,ny) )
if tempInd in order[:iSMC]:
kappa = self.J
tempIndSMC = indSorted[orderSorted.searchsorted(tempInd)]
Y = tempDispersion*np.sin( -tempMean ) + kappa*np.sin( -trajectory[:,tempIndSMC] )
X = tempDispersion*np.cos( -tempMean ) + kappa*np.cos( -trajectory[:,tempIndSMC] )
tempDispersion = np.sqrt( tempDispersion**2 + kappa**2 + 2*kappa*tempDispersion*np.cos(-tempMean+trajectory[:,tempIndSMC] ) )
tempMean = -np.arctan2(Y ,X)
for iParticle in range(N):
trajectory[iParticle, iSMC] = hlp.vonmises(tempMean[iParticle], tempDispersion[iParticle])
nu += np.log(2 * np.pi * np.i0(tempDispersion))
nuMax = np.max(nu)
nu = np.exp( nu - nuMax )
logZ[iter] = logZ[iter-1] + nuMax + np.log( np.sum(nu) ) - np.log(N)
if verbose:
c = np.sum(nu)
nu /= c
trajMean = np.mean( (np.tile(nu, (len(order),1))).T*trajectory, axis=0 )
return logZ, trajMean[order].reshape( (nx,ny) ), ess
else:
return logZ[iter]
# General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with BASILISK. If not, see <http://www.gnu.org/licenses/>.
"""
Von Mises sampler and density.
"""
from math import pi, acos, cos, sin
from numpy import log, exp, i0 # i0 is modified Bessel function of first kind
from random import vonmisesvariate
i0_float = lambda x: float(i0(float(x)))
def to_radian(coords):
if coords[1]>0:
return acos(coords[0])
else:
return 2*pi-acos(coords[0])
def to_coords(angle):
return array([cos(angle), sin(angle)])
def mod2pi(x):
if x >= 2*pi:
return mod2pi(x - 2*pi)
if x < 0:
lambda x: x.reshape((x.shape[0] * x.shape[1], x.shape[2])),
lambda x: abs(x),
lambda x: x > 0.5,
lambda x: x.rechunk((4, 4, 4)),
lambda x: x.rechunk((2, 2, 1)),
pytest.param(lambda x: da.einsum("ijk,ijk", x, x),
marks=pytest.mark.xfail(
reason='depends on resolution of https://github.com/numpy/numpy/issues/12974')),
lambda x: np.isreal(x),
lambda x: np.iscomplex(x),
lambda x: np.isneginf(x),
lambda x: np.isposinf(x),
lambda x: np.real(x),
lambda x: np.imag(x),
lambda x: np.fix(x),
lambda x: np.i0(x.reshape((24,))),
lambda x: np.sinc(x),
lambda x: np.nan_to_num(x),
]
@pytest.mark.parametrize('func', functions)
def test_basic(func):
c = cupy.random.random((2, 3, 4))
n = c.get()
dc = da.from_array(c, chunks=(1, 2, 2), asarray=False)
dn = da.from_array(n, chunks=(1, 2, 2))
ddc = func(dc)
ddn = func(dn)
def get_angle_biased(x, k):
random_angle = np.random.uniform(-np.pi, np.pi)
angle = (1/(2*np.pi*np.i0(k)))*(np.e**(k*(x-random_angle)))
return angle
def forward(ctx, inp, nu):
ctx._nu = nu
ctx.save_for_backward(inp)
return torch.from_numpy(np.i0(inp.detach().numpy()))
import numpy as np
import matplotlib.pyplot as mp
x = np.linspace(-5, 5, 1001)
y = np.i0(x)
mp.gcf().set_facecolor(np.ones(3) * 240 / 255)
mp.title('Bessel Function', fontsize=20)
mp.xlabel('x', fontsize=14)
mp.ylabel('y', fontsize=14)
mp.tick_params(labelsize=10)
mp.grid(linestyle=':')
mp.plot(x, y, c='dodgerblue', label='Bessel')
mp.legend()
mp.show()
def vonmises(x, amp, cen, kappa):
# "1-d vonmises"
top = (amp / (np.pi * 2 * np.i0(kappa)))
bot = np.exp(kappa * np.cos(x / 360.0 * 2.0 * np.pi - cen))
return top * bot
def tune(plotfreq=False, plottime=False, input_device_index=None):
# Set up the Kaiser window
n = np.arange(CHUNK_SIZE) + 0.5 # Assuming CHUNK_SIZE is even
x = (n - CHUNK_SIZE / 2) / (CHUNK_SIZE / 2)
window = np.i0(KAISER_BETA * np.sqrt(1 - x ** 2)) / np.i0(KAISER_BETA)
# Get audio data
p = pyaudio.PyAudio()
#device_info = p.get_device_info_by_index(input_device_index)
#print(device_info)
stream = p.open(format=FORMAT, channels=1, rate=RATE, input=True,
input_device_index=input_device_index,
frames_per_buffer=CHUNK_SIZE)
if plotfreq or plottime:
# Set up plotting paraphernalia
plt.ion()
if plottime:
figtime = plt.figure()
axtime = figtime.gca()
if plotfreq:
figfreq = plt.figure()
axfreq = figfreq.gca()
print 'Press return to stop...'
i = 0
while 1:
i += 1
# Check if something has been input. If so, exit.
if sys.stdin in select([sys.stdin, ], [], [], 0)[0]:
# Absorb the input and break
sys.stdin.readline()
break
# Acquire sound data
snd_data = array('h', stream.read(CHUNK_SIZE))
signal = np.array(snd_data)
#if sys.byteorder == 'big':
#snd_data.byteswap()
if plottime:
if i > 1:
axtime.lines.remove(timeline)
[timeline, ] = axtime.plot(signal, 'b-')
figtime.canvas.draw()
# Apply a Kaiser window on the signal before taking the FFT. This
# makes the signal look better if it is periodic. Derivatives at the
# edges of the signal match better, which means that the frequency
# domain will have fewer side-lobes. However, it does cause each spike
# to grow a bit broader.
# One can change the value of beta to tradeoff between side-lobe height
# and main lobe width.
signal *= window
spectrum = np.fft.rfft(signal, int(RATE / RESOLUTION))
peak = np.argmax(abs(spectrum)) # peak directly gives the
# desired frequency
# Threshold on the maximum peak present in the signal (meaning we
# expect the signal to be approximately unimodal)
if spectrum[peak] > THRESHOLD:
# Put a band-pass filter in place to look at only those frequencies
# we want. The desired peak is the harmonic located in the
# frequency region of interest.
desired_peak = np.argmax(abs(spectrum[90:550]))
print desired_peak
if plotfreq:
try:
axfreq.lines.remove(freqline)
except UnboundLocalError:
pass
[freqline, ] = axfreq.plot(abs(spectrum), 'b-')
figfreq.canvas.draw()
stream.stop_stream()
stream.close()
p.terminate()
def myF_2vM1U(theta, *params):
"""
The first derivative of the log-likelihood function 'l' to be set equal to
zero in order to estimate the mixture parameters:
p1, p2: the weights of the two von Mises distributions on semi-circle
kappa2, kappa2: the concentrations of the two von Mises distributions
mu1, mu2: the locations of the two von Mises distributions
theta:= [ p1, kappa1, mu1, p2, kappa2, mu2 ]
params:= X_samples: the observations sample
The function returns a vector F with the derivatives of 'l' wrt the
components of theta.
This function is to be called with optimize.fsolve function of scipy:
roots_ = optimize.fsolve( myF_2vM1U, in_guess, args=my_par )
"""
# scal_ = 0.5
ll = 2.0
x_ = np.array(params).T
# print(type(x_))
# print('x_=', x_.shape)
# the unknown parameters:
p1_ = theta[0]
kappa1_ = theta[1]
m1_ = theta[2]
p2_ = theta[3]
kappa2_ = theta[4]
m2_ = theta[5]
pu_ = 1.0 - p1_ - p2_
# the 1st von Mises distribution on semi-circle:
fvm1_ = _vmf_pdf(x_, kappa1_, m1_, ll)
# fvm1_ = stats.vonmises.pdf( x_, kappa1_, m1_, scal_ )
# the 2nd von Mises distribution on semi-circle:
fvm2_ = _vmf_pdf(x_, kappa2_, m2_, ll)
# fvm2_ = stats.vonmises.pdf( x_, kappa2_, m2_, scal_ )
# the uniform distribution:
xu_1 = min(x_)
xu_2 = (max(x_) - min(x_))
fu_ = stats.uniform.pdf(x_, loc=xu_1, scale=xu_2)
# mixture distribution:
fm_ = p1_ * fvm1_ + p2_ * fvm2_ + pu_ * fu_
# first derivative wrt weight p1:
dldp1 = sum(np.divide(np.subtract(fvm1_, fu_), fm_))
# first derivative wrt weight p2:
dldp2 = sum(np.divide(np.subtract(fvm2_, fu_), fm_))
# first derivative wrt location mu1:=
dldm1 = (ll*kappa1_*p1_)*sum( np.multiply( np.divide( fvm1_, fm_ ), \
np.sin(ll*(x_ - m1_)) ) )
# first derivative wrt location mu2:=
dldm2 = (ll*kappa2_*p2_)*sum( np.multiply( np.divide( fvm2_, fm_ ), \
np.sin(ll*(x_ - m2_)) ) )
# first derivative wrt concentration kappa1:
Ak1 = (iv(1.0, kappa1_) / i0(kappa1_))
dldk1 = p1_*sum( np.multiply( np.divide( fvm1_, fm_ ), \
( np.cos(ll*(x_ - m1_)) - Ak1 ) ) )
# first derivative wrt concentration kappa1:
Ak2 = (iv(1.0, kappa2_) / i0(kappa2_))
dldk2 = p2_*sum( np.multiply( np.divide( fvm2_, fm_ ), \
( np.cos(ll*(x_ - m2_)) - Ak2 ) ) )
F = [dldp1[0], dldk1[0], dldm1[0], dldp2[0], dldk2[0], dldm2[0]]
return F
def c1(cls, th, k1, a, b):
"""First component of the fitted function.
"""
return a * b * np.exp(k1 * np.cos(th)) / (2. * np.pi * np.i0(k1))
left=cutoff1d[0]
right=cutoff1d[1]
if left==0:
scalefreq=0.0
elif right==1:
scalefreq=1
else:
scalefreq=0.5*(left+right)
s=0
kais=[]
for i in range(N):
# print sp.bessel(0,beta*np.sqrt(1-np.power((i-alpha)/alpha,2)),analog=True)
# print beta
c.append(np.cos(np.pi*m[i]*scalefreq))
kais.append(np.i0(beta*np.sqrt(1-np.power((i-alpha)/alpha,2)))/np.i0(beta))
FIR.append(h[i]*kais[i])
s=s+FIR[i]*c[i]
pl.plot(c,label='c')
pl.plot(h,label='h')
pl.plot(kais,label='kais')
pl.plot(FIR,label='FIR')
pl.legend()
pl.figure()
Fh=np.fft.fftshift(np.fft.fft(h))
Fc=np.fft.fftshift(np.fft.fft(c))
Fkais=np.fft.fftshift(np.fft.fft(kais))
#!/usr/bin/python import numpy from pylab import * x = numpy.linspace(0, 4, 100) vals = numpy.i0(x) plot(x, vals) show()
"""
#Script to evaluation the response of each angle relative to the imaging angle
import matplotlib.pyplot as plt
import numpy as np
from smart_rotation import smart_rotation
filepath = "/mnt/fileserver/Henry-SPIM/smart_rotation/06142018/sample1/merged/workspace_final3/angularcount_final/"
sr = smart_rotation(24,10)
sr.evaluate_angles(filepath,24,10,0,False)
savepath = "/mnt/fileserver/Henry-SPIM/smart_rotation/06142018/sample1/merged/workspace_final3/figures/"
alpha = 0.2
fig,(a0,a1,a2,a3) = plt.subplots(4,1,sharex = True,figsize=(4.5,3),gridspec_kw={'height_ratios':[12,1,1,1]})
plt.subplots_adjust(left=0,right=1,top=1,bottom=0,hspace=0.03)
linewidth = 1
markersize=2
a0.plot(np.arange(0,360,10),np.divide(sr.a[:,0]/np.pi/2,np.i0(sr.k))*np.exp(np.abs(sr.k))[:,0],'g-o',label="max",lw =linewidth,markersize=markersize)
a0.plot(np.arange(0,360,10),np.divide(sr.a[:,0]/np.pi/2,np.i0(sr.k))*np.exp(-np.abs(sr.k))[:,0],'r-o',label="min",lw=linewidth,markersize=markersize)
a0.grid(False)
a0.legend(loc="upper right")
a0.set_xlim(0,360)
a0.set_ylim(0,15000)
a0.xaxis.tick_top()
#a0.set_xlabel('imaging angle')
a0.set_xlabel('Angle within sample')
axis = [a1,a2]
for i in axis:
i.set_yticklabels([])
i.xaxis.set_ticks_position('none')
i.yaxis.set_ticks_position('none')
i.set_xlim(0,360)
mini = 1
print u"分期付款" #pmt输入为利率和期数,总价，输出每期钱数 print "Payment",np.pmt(0.10/12,12*30,100000) #nper参数为贷款利率，固定的月供和贷款额，输出付款期数 print "Number of payments", np.nper(0.10/12,-100,9000) #rate参数为付款期数，每期付款资金，现值和终值计算利率 print "Interest rate",12*np.rate(167,-100,9000,0) print u"窗函数" #bartlett函数可以计算巴特利特窗 window = np.bartlett(42) print "bartlett",window #blackman函数返回布莱克曼窗。该函数唯一的参数为输出点的数量。如果数量 #为0或小于0，则返回一个空数组。 window = np.blackman(10) print "blackman",window # hamming函数返回汉明窗。该函数唯一的参数为输出点的数量。如果数量为0或 # 小于0，则返回一个空数组。 window = np.hamming(42) print "hamming",window # kaiser函数返回凯泽窗。该函数的第一个参 # 数为输出点的数量。如果数量为0或小于0，则返回一个空数组。第二个参数为β值。 window = np.kaiser(42,14) print "kaiser",window # 以i0 表示第一类修正的零阶贝塞尔函数。 x= np.linspace(0,4,100) vals = np.i0(x) print "i0",vals val = np.sinc(x) print "sinc",val
def likelyhood(self,theta, r):
"""likelyhood
computes the likelyhood that an ellipticity with angle theta at distance r is recorded"""
br = self.b0 * r**self.alpha
expar = (1./(2.*numpy.pi * numpy.i0(br)))
return numpy.exp(-(br*cos(2*theta))) * expar