def __init__(self, circ, poles, zeros, outfile):
solution.__init__(self, circ, outfile)
self.sol_type = "PZ"
self.poles = np.sort_complex(np.array(poles).reshape((-1, )))
self.zeros = np.sort_complex(np.array(zeros).reshape((-1, )))
data = np.vstack((self.poles.reshape(
(-1, 1)), self.zeros.reshape((-1, 1))))
if np.prod(self.poles.shape):
for i in range(self.poles.shape[0]):
self.variables += ['p%d' % i]
if np.prod(self.zeros.shape):
for i in range(self.zeros.shape[0]):
self.variables += ['z%d' % i]
for v in self.variables:
self.units.update({v: "rad/s"})
self.csv_headers = []
for i in range(len(self.variables)):
self.csv_headers.append("Re(%s)" % self.variables[i])
self.csv_headers.append("Im(%s)" % self.variables[i])
# save in Re/Im form
sdata = data.reshape(-1).view(np.float_).reshape((-1, 1))
self._add_data(sdata)
# store local data too:
self.data = case_insensitive_dict()
for i in range(len(self.variables)):
self.data.update({self.variables[i]: data[i, 0]})
def __init__(self, circ, poles, zeros, outfile):
solution.__init__(self, circ, outfile)
self.sol_type = "PZ"
self.poles = np.sort_complex(np.array(poles).reshape((-1,)))
self.zeros = np.sort_complex(np.array(zeros).reshape((-1,)))
data = np.vstack((self.poles.reshape((-1, 1)),
self.zeros.reshape((-1, 1))))
if np.prod(self.poles.shape):
for i in range(self.poles.shape[0]):
self.variables += ['p%d' % i]
if np.prod(self.zeros.shape):
for i in range(self.zeros.shape[0]):
self.variables += ['z%d' % i]
for v in self.variables:
self.units.update({v: "rad/s"})
self.csv_headers = []
for i in range(len(self.variables)):
self.csv_headers.append("Re(%s)" % self.variables[i])
self.csv_headers.append("Im(%s)" % self.variables[i])
# save in Re/Im form
sdata = data.reshape(-1).view(np.float_).reshape((-1, 1))
self._add_data(sdata)
# store local data too:
self.data = case_insensitive_dict()
for i in range(len(self.variables)):
self.data.update({self.variables[i]: data[i, 0]})
def myzpk2tf(self, z, p, k):
z = np.atleast_1d(z)
k = np.atleast_1d(k)
if len(z.shape) > 1:
temp = np.poly(z[0])
b = np.zeros((z.shape[0], z.shape[1] + 1), temp.dtype.char)
if len(k) == 1:
k = [k[0]] * z.shape[0]
for i in range(z.shape[0]):
b[i] = k[i] * poly(z[i])
else:
b = k * np.poly(z)
a = np.atleast_1d(np.poly(p))
# Use real output if possible. Copied from numpy.poly, since
# we can't depend on a specific version of numpy.
if issubclass(b.dtype.type, np.complexfloating):
# if complex roots are all complex conjugates, the roots are real.
roots = np.asarray(z, complex)
pos_roots = np.compress(roots.imag > 0, roots)
neg_roots = np.conjugate(np.compress(roots.imag < 0, roots))
if len(pos_roots) == len(neg_roots):
if np.all(np.sort_complex(neg_roots) == np.sort_complex(pos_roots)):
b = b.real.copy()
if issubclass(a.dtype.type, np.complexfloating):
# if complex roots are all complex conjugates, the roots are real.
roots = np.asarray(p, complex)
pos_roots = np.compress(roots.imag > 0, roots)
neg_roots = np.conjugate(np.compress(roots.imag < 0, roots))
if len(pos_roots) == len(neg_roots):
if np.all(np.sort_complex(neg_roots) == np.sort_complex(pos_roots)):
a = a.real.copy()
return b, a
def zpk2tf(z, p, k):
""" Return polynomial transfer function representation from zeros and poles """
z = np.atleast_1d(z)
k = np.atleast_1d(k)
if len(z.shape) > 1:
temp = np.poly(z[0])
b = np.zeros((z.shape[0], z.shape[1] + 1), temp.dtype.char)
if len(k) == 1:
k = [k[0]] * z.shape[0]
for i in range(z.shape[0]):
b[i] = k[i] * np.poly(z[i])
else:
b = k * np.poly(z)
a = np.atleast_1d(np.poly(p))
# Use real output if possible
if issubclass(b.dtype.type, np.complexfloating):
# if complex roots are all complex conjugates, the roots are real.
roots = np.asarray(z, complex)
pos_roots = np.compress(roots.imag > 0, roots)
neg_roots = np.conjugate(np.compress(roots.imag < 0, roots))
if len(pos_roots) == len(neg_roots):
if np.all(
np.sort_complex(neg_roots) == np.sort_complex(pos_roots)):
b = b.real.copy()
if issubclass(a.dtype.type, np.complexfloating):
# if complex roots are all complex conjugates, the roots are real.
roots = np.asarray(p, complex)
pos_roots = np.compress(roots.imag > 0, roots)
neg_roots = np.conjugate(np.compress(roots.imag < 0, roots))
if len(pos_roots) == len(neg_roots):
if np.all(
np.sort_complex(neg_roots) == np.sort_complex(pos_roots)):
a = a.real.copy()
return b, a
def zpk2tf(z, p, k):
"""
Return polynomial transfer function representation from zeros and poles
Parameters
----------
z : array_like
Zeros of the transfer function.
p : array_like
Poles of the transfer function.
k : float
System gain.
Returns
-------
b : ndarray
Numerator polynomial coefficients.
a : ndarray
Denominator polynomial coefficients.
"""
z = atleast_1d(z)
k = atleast_1d(k)
if len(z.shape) > 1:
temp = poly(z[0])
b = zeros((z.shape[0], z.shape[1] + 1), temp.dtype.char)
if len(k) == 1:
k = [k[0]] * z.shape[0]
for i in range(z.shape[0]):
b[i] = k[i] * poly(z[i])
else:
b = k * poly(z)
a = atleast_1d(poly(p))
# Use real output if possible. Copied from numpy.poly, since
# we can't depend on a specific version of numpy.
if issubclass(b.dtype.type, numpy.complexfloating):
# if complex roots are all complex conjugates, the roots are real.
roots = numpy.asarray(z, complex)
pos_roots = numpy.compress(roots.imag > 0, roots)
neg_roots = numpy.conjugate(numpy.compress(roots.imag < 0, roots))
if len(pos_roots) == len(neg_roots):
if numpy.all(
numpy.sort_complex(neg_roots) == numpy.sort_complex(
pos_roots)):
b = b.real.copy()
if issubclass(a.dtype.type, numpy.complexfloating):
# if complex roots are all complex conjugates, the roots are real.
roots = numpy.asarray(p, complex)
pos_roots = numpy.compress(roots.imag > 0, roots)
neg_roots = numpy.conjugate(numpy.compress(roots.imag < 0, roots))
if len(pos_roots) == len(neg_roots):
if numpy.all(
numpy.sort_complex(neg_roots) == numpy.sort_complex(
pos_roots)):
a = a.real.copy()
return b, a
def allsortedclose(a, b, atol=1e-3, rtol=1e-3):
if np.iscomplex(a).any():
a = np.sort_complex(a)
else:
a = np.sort(a)
if np.iscomplex(b).any():
b = np.sort_complex(b)
else:
b = np.sort(b)
return np.allclose(a, b, rtol=rtol, atol=atol)
def allsortedclose(a, b, atol=1e-3, rtol=1e-3):
if np.iscomplex(a).any():
a = np.sort_complex(a)
else:
a = np.sort(a)
if np.iscomplex(b).any():
b = np.sort_complex(b)
else:
b = np.sort(b)
return np.allclose(a, b, rtol=rtol, atol=atol)
def test_anisotropic_xi_eigenvalues(rnd_data1, rnd_data2, rnd_data3, rnd_data4,
rnd_data5, rnd_data6):
"""
Comparison of eigenvalue calculation for xi from random complex material
data. Comparing polynomial calculation from determinant, from quadratic
eigenvalue problem and analytical calculation from sympy.
"""
lc = LocalCoordinates("1")
myeps = np.zeros((3, 3), dtype=complex)
myeps[0:2, 0:2] = rnd_data1 + complex(0, 1) * rnd_data2
myeps[2, 2] = rnd_data3 + complex(0, 1) * rnd_data4
#np.random.random((3, 3)) + complex(0, 1)*np.random.random((3, 3))
((epsxx, epsxy, _), (epsyx, epsyy, _), (_, _, epszz)) = \
tuple(myeps)
m = AnisotropicMaterial(lc, myeps)
#n = np.random.random((3, 1))
#n = n/np.sqrt(np.sum(n*n, axis=0))
n = np.zeros((3, 1))
n[2, :] = 1
x = np.zeros((3, 1))
k = rnd_data5 + complex(0, 1) * rnd_data6
kpa = k - np.sum(n * k, axis=0) * n
(eigenvalues, _) = m.calcXiEigenvectorsNorm(x, n, kpa)
xiarray = m.calcXiNormZeros(x, n, kpa)
# sympy check with analytical solution
kx, ky, xi = sympy.symbols('k_x k_y xi')
exx, exy, _, eyx, eyy, _, _, _, ezz \
= sympy.symbols('e_xx e_xy e_xz e_yx e_yy e_yz e_zx e_zy e_zz')
#eps = Matrix([[exx, exy, exz], [eyx, eyy, eyz], [ezx, ezy, ezz]])
eps = sympy.Matrix([[exx, exy, 0], [eyx, eyy, 0], [0, 0, ezz]])
v = sympy.Matrix([[kx, ky, xi]])
m = -(v * v.T)[0] * sympy.eye(3) + v.T * v + eps
detm = m.det().collect(xi)
soldetm = sympy.solve(detm, xi)
subsdict = {
kx: kpa[0, 0],
ky: kpa[1, 0],
exx: epsxx,
exy: epsxy,
eyx: epsyx,
eyy: epsyy,
ezz: epszz,
sympy.I: complex(0, 1)
}
analytical_solution = np.sort_complex(
np.array([sol.evalf(subs=subsdict) for sol in soldetm], dtype=complex))
numerical_solution1 = np.sort_complex(xiarray[:, 0])
numerical_solution2 = np.sort_complex(eigenvalues[:, 0])
assert np.allclose(analytical_solution - numerical_solution1, 0)
assert np.allclose(analytical_solution - numerical_solution2, 0)
def test_ChebyshevII_transferfunction():
N = 15
Wn = 1e3
rs = 60
chebyshevII_worth_system = cbadc.analog_system.ChebyshevII(N, Wn, rs)
b, a = scipy.signal.cheby2(N, rs, Wn, btype='low', analog=True)
print("b, a\n")
print(b)
print(a)
z, p, k = scipy.signal.cheby2(N,
rs,
Wn,
btype='low',
analog=True,
output='zpk')
print("z,p,k")
print(z)
print(np.sort_complex(p))
print(k)
ss_z, ss_p, ss_k = chebyshevII_worth_system.zpk()
print("State Space model as zpk")
print(ss_z)
print(ss_p)
print(ss_k)
w, h = scipy.signal.freqs(b, a)
tf = chebyshevII_worth_system.transfer_function_matrix(w)
print(chebyshevII_worth_system)
if not np.allclose(h, tf):
print(h - tf)
raise BaseException("Filter mismatch")
def test_ChebyshevI_transferfunction():
N = 4
Wn = 1e3
rp = np.sqrt(2)
chebyshevI_worth_system = cbadc.analog_system.ChebyshevI(N, Wn, rp)
b, a = scipy.signal.cheby1(N, rp, Wn, btype='low', analog=True)
print("b, a\n")
print(b)
print(a)
z, p, k = scipy.signal.cheby1(N,
rp,
Wn,
btype='low',
analog=True,
output="zpk")
print("z,p,k")
print(z)
print(np.sort_complex(p))
print(k)
w, h = scipy.signal.freqs(b, a)
tf = chebyshevI_worth_system.transfer_function_matrix(w)
if not np.allclose(h, tf):
print(h - tf)
raise BaseException("Filter mismatch")
def test_ButterWorth_transferfunction():
N = 7
Wn = 1e3
butter_worth_system = cbadc.analog_system.ButterWorth(N, Wn)
print(butter_worth_system)
b, a = scipy.signal.butter(N, Wn, btype='low', analog=True)
print("b, a\n")
print(b)
print(a)
z, p, k = scipy.signal.butter(N,
Wn,
btype='low',
output="zpk",
analog=True)
print("z,p,k")
print(z)
print(np.sort_complex(p))
print(k)
print(zpk2abcd(z, p, k))
w, h = scipy.signal.freqs(b, a)
tf = butter_worth_system.transfer_function_matrix(w)[:, 0, :].flatten()
if not np.allclose(h, tf):
# print(h, tf)
print(h - tf)
raise BaseException("Filter mismatch")
def test_Cauer_transferfunction():
N = 11
Wn = 1e3
rp = np.sqrt(2)
rs = 60
cauer_worth_system = cbadc.analog_system.Cauer(N, Wn, rp, rs)
b, a = scipy.signal.ellip(N, rp, rs, Wn, btype='low', analog=True)
print("b, a\n")
print(b)
print(a)
print(cauer_worth_system)
w, h = scipy.signal.freqs(b, a)
z, p, k = scipy.signal.ellip(N,
rp,
rs,
Wn,
btype='low',
analog=True,
output='zpk')
print("z,p,k")
print(z)
print(np.sort_complex(p))
print(k)
tf = cauer_worth_system.transfer_function_matrix(w)
print(cauer_worth_system)
if not np.allclose(h, tf):
print(h - tf)
raise BaseException("Filter mismatch")
def test_sort_complex(self):
ma = Masked(np.array([1 + 2j, 0 + 4j, 3 + 0j, -1 - 1j]),
mask=[True, False, False, False])
o = np.sort_complex(ma)
indx = np.lexsort((ma.unmasked.imag, ma.unmasked.real, ma.mask))
expected = ma[indx]
assert_masked_equal(o, expected)
def sort_and_output(self, E0):
n = OPT_numdgt()
self._out.setfloatwidth(n, n)
Q = []
for i in E0.transpose():
if (i[1]):
q = i[0] / i[1]
im = np.imag(q)
im = fixzero(im, 1e8)
re = np.real(q)
if (abs(re) > 1e8):
continue
re = fixzero(re, 1e8)
Q.append(re + 1j * im)
Q = np.array(Q)
Q = np.sort_complex(1j * Q) / 1j
# Q = np.sort_complex(Q)
for i in Q:
# print(' {:.6e}'.format(i))
# use outdata...
self._out << np.real(i)
if (np.imag(i) < 0):
self._out << "- j*"
else:
self._out << "+ j*"
self._out << abs(np.imag(i))
self._out << "\n"
def compute_quadratic_state_integral_ALDS(H, x0, T, dt=None):
''' Assuming the state x(t) evolves in time according to a linear dynamic:
dx(t)/dt = H * x(t)
Compute the following integral:
int_{0}^{T} x(t)^T * x(t) dt
'''
if (dt is None):
w, V = eig(H) # H = V*matlib.diagflat(w)*V^{-1}
print "Eigenvalues H:", np.sort_complex(w).T
Lambda_inv = matlib.diagflat(1.0 / w)
e_2T_Lambda = matlib.diagflat(np.exp(2 * T * w))
int_e_2T_Lambda = 0.5 * Lambda_inv * (e_2T_Lambda - matlib.eye(n))
# V_inv = np.linalg.inv(V)
# cost = x0.T*(V_inv.T*(int_e_2T_Lambda*(V_inv*x0)))
V_inv_x0 = np.linalg.solve(V, x0)
cost = V_inv_x0.T * int_e_2T_Lambda * V_inv_x0
return cost[0, 0]
N = int(T / dt)
x = simulate_ALDS(H, x0, dt, N)
cost = 0.0
not_finite_warning_printed = False
for i in range(N):
if (np.all(np.isfinite(x[:, i]))):
cost += dt * (x[:, i].T * x[:, i])[0, 0]
elif (not not_finite_warning_printed):
print 'WARNING: x is not finite at time step %d' % (i) #, x[:,i].T
not_finite_warning_printed = True
return cost
def t3_sort_complex(complex_array):
if len(complex_array) == 0:
return (np.array([]), np.array([]))
data = [[abs(i), i] for i in complex_array]
data.sort(key=lambda data: data[0])
return (np.array(data)[:,
1], np.flip(np.sort_complex(complex_array), axis=0))
def map2k(maps, field_deg, k_max):
field_deg = np.float64(field_deg)
# k_max = k_max.astype(np.int)
nside = maps.shape[0]
# print(nside)
ft_map = np.fft.rfft2(maps) * nside
P = abs(ft_map)**2
l_min = 360.0 / field_deg #ell minimum ell=2pi/theta
l_max = nside * l_min
ly = np.fft.fftfreq(nside) * l_max #* l_max
lx = np.fft.rfftfreq(nside) * l_max
l = np.sqrt(lx[np.newaxis, :]**2 + ly[:, np.newaxis]**2)
p_l = 1.0j * P + l
pl = np.zeros(nside * (nside // 2 + 1), dtype='complex')
p_l = np.sort_complex(p_l)
p_l = p_l.reshape((1, len(pl)))
pl[:] = p_l[0, :]
# print(pl.shape)
p_ll = interpolate.interp1d(pl.real,
pl.imag,
bounds_error=False,
kind='linear',
fill_value=0.0)
ell = np.arange(k_max)
C_l = p_ll(ell)
return ell, C_l
def sortColumns(matrix):
new_matrix = []
print("wymair macierzy: ", np.shape(matrix))
for i in range(len(matrix[0, :])):
column = matrix[:, i]
new_matrix.append(np.sort_complex(column))
return np.array(new_matrix).transpose()
def test_sort_complex_1(self):
IntTestData = [5, 3, 6, 2, 1]
ComplexTextData = [1 + 2j, 2 - 1j, 3 - 2j, 3 - 3j, 3 + 5j]
a = np.sort(IntTestData)
print(a)
b = np.sort_complex(IntTestData)
print(b)
c = np.sort(ComplexTextData)
print(c)
d = np.sort_complex(ComplexTextData)
print(d)
def compute_closed_loop_eigenvales(robot, gains_array):
''' Compute eigenvalues of linear part of closed-loop system:
d3f = -Kd_bar*d2f - (K*Upsilon+Kp_bar)*df - Kp_bar*K*A*Kf*e_f + ...
'''
ny=3
ei_cls_f = matlib.empty((3*nf,T), dtype=complex)*np.nan
ei_cls = matlib.empty((3*nf+2*ny,T), dtype=complex)*np.nan
for t in range(T):
H = compute_closed_loop_transition_matrix(gains_array, robot, q[:,t], v[:,t])
H_f = H[2*ny:, 2*ny:]
ei_cls_f[:,t] = np.sort_complex(eigvals(H_f)).reshape((3*nf,1))
ei_cls[:,t] = np.sort_complex(eigvals(H)).reshape((3*nf+2*ny,1))
plot_stats_eigenvalues(ei_cls_f, name='Closed-loop force tracking')
plot_stats_eigenvalues(ei_cls, name='Closed-loop momentum tracking')
return ei_cls
def t3_sort_complex(complex_array):
if len(complex_array) == 0:
return (np.array([]), np.array([]))
tup = np.sort_complex(complex_array)
module_array = np.array([abs(i) for i in complex_array])
x = module_array.argsort()
return (complex_array[x], np.flip(tup, axis=0))
def compare_eigenvalues(values_test, values_answer, ds_name):
# first sort values: first on real part, then on imaginary part
values_test = np.sort_complex(values_test)
values_answer = np.sort_complex(values_answer)
fig, ax = plt.subplots(1, figsize=(12, 8))
fail_idxs = []
for i, val_test in enumerate(values_test):
# calculate distance from this point to all points
distances = np.sqrt((values_answer.real - val_test.real)**2 +
(values_answer.imag - val_test.imag)**2)
# find point with closest distance
match = values_answer[distances.argmin()]
# answer + test in pixels
x_answ, y_answ = ax.transData.transform((match.real, match.imag))
x_test, y_test = ax.transData.transform((val_test.real, val_test.imag))
# get distance between both in pixels
dist_pix = np.sqrt((x_answ - x_test)**2 + (y_answ - y_test)**2)
try:
# distance in pixels should lie below tolerance
assert dist_pix < PIX_TOL
except AssertionError:
fail_idxs.append(i)
ax.clear()
# this should be empty if all values match. If not, raise error and save log
if fail_idxs:
for i in fail_idxs:
val_test = values_test[i]
val_answ = values_answer[i]
print(f"FAIL: {val_test} >> {val_answ} (test >> answer)")
ax.plot(val_test.real, val_test.imag, ".g", markersize=3)
ax.plot(val_answ.real, val_answ.imag, "xr", markersize=3)
ax.set_title(ds_name, fontsize=15)
ax.set_xlabel(r"Re($\omega$)", fontsize=15)
ax.set_ylabel(r"Im($\omega$)", fontsize=15)
ax.tick_params(which="both", labelsize=15)
ax.legend(["test results", "answer results"], loc="best", fontsize=13)
print(f">>> FAILED EIGENVALUES: {len(fail_idxs)}/{len(values_answer)}")
filename = (output / f"FAILED_{ds_name}.png").resolve()
fig.savefig(filename, dpi=400)
plt.close(fig)
raise AssertionError
plt.close(fig)
def sort_func():
x = np.array([3, 1, 2])
print("x ", x)
# 数组排序
print("np.sort(a) ", np.sort(x))
# 多维数组排序
a = np.array([[1, 5, 4], [3, 2, 1]])
print("x ", a)
# 数组排序 sort along the last axis
print("np.sort(a) ", np.sort(a))
# sort the flattened array
print(" np.sort(a, axis=None) ", np.sort(a, axis=None))
# sort along the first axis
print(" np.sort(a, axis=0) ", np.sort(a, axis=0))
# 使用键序列执行间接排序。
surnames = ('Hertz', 'Galilei', 'Hertz')
first_names = ('Heinrich', 'Galileo', 'Gustav')
ind = np.lexsort((first_names, surnames))
print("ind", ind)
rs = [surnames[i] + ", " + first_names[i] for i in ind]
print("rs", rs)
# 返回将数组分类的索引 返回索引
x = np.array([3, 1, 2])
print("np.argsort(x)", np.argsort(x))
# 排序后重新赋值到新的数组中
rs = [x[i] for i in ind]
print("rs", rs)
# 返回沿第一个轴排序的数组的副本。
x = np.array([3, 1, 2])
print("np.msort(x)", np.msort(x))
y = np.sort_complex([5, 3, 6, 2, 1])
print("y", y)
# numpy.argmax 返回沿轴的最大值的索引
a = np.arange(6).reshape(2, 3)
print("a", a)
print("np.argmax(a)", np.argmax(a))
# numpy.argpartition
a = np.array([[np.nan, 4], [2, 3]])
print("np.nanargmax(a)", np.nanargmax(a))
a = np.arange(6).reshape(2, 3)
print("a", a)
print("np.argmin(a)", np.argmin(a))
# 找到非零的数组元素的索引，按元素分组
x = np.argwhere(x > 1)
print("x", x)
x = np.where(x > 1)
print("x", x)
# 查找要插入元素以维持顺序的索引
x = np.searchsorted([1, 2, 3, 4, 5], 3)
print("x", x)
def quartic_roots(k, x1, x2, x3):
K = complex128(ellipk(k**2))
e0 = complex128((x1*j - x2)**2 + .25 * K**2)
e1 = complex128(4*(x1*j-x2)*x3)
e2 = complex128(4*(x3**2) - 2 * (x1**2) - 2 * (x2**2) + (K**2) * (k**2 - 0.5))
e3 = complex128(4*x3*(x2 + j*x1))
e4 = complex128(x2**2 - x1**2 + 2*j*x1*x2 + 0.25*K**2)
return sort_complex(roots([e4, e3, e2, e1, e0])) # I put the sort_complex to have a canonical form, so that when we order them they will vary continuously
def quartic_roots(k, x1, x2, x3):
K = complex128(ellipk(k**2))
e0 = complex128((x1*j - x2)**2 + .25 * K**2)
e1 = complex128(4*(x1*j-x2)*x3)
e2 = complex128(4*(x3**2) - 2 * (x1**2) - 2 * (x2**2) + (K**2) * (k**2 - 0.5))
e3 = complex128(4*x3*(x2 + j*x1))
e4 = complex128(x2**2 - x1**2 + 2*j*x1*x2 + 0.25*K**2)
return sort_complex(roots([e4, e3, e2, e1, e0])) # I put the sort_complex to have a canonical form, so that when we order them they will vary continuously
def get_highest_val(a):
arr = []
for v in a:
arr.append(v)
arr = np.sort_complex(np.array(arr))
if len(arr) != 0:
ret = arr[-1]
else:
ret = 0
return ret
def compute_cl_poles(ctl_num, ctl_den):
ctl_tf = control.tf(ctl_num, ctl_den, dt)
#print(ctl_tf)
cl_tf = control.feedback(dt_tf, ctl_tf, sign=-1)
#print control.damp(cl_tf)
#print(cl_tf)
cl_poles_d = control.pole(cl_tf)
#print(cl_polesd)
cl_poles_c = np.sort_complex(np.log(cl_poles_d) / dt)
print('cl_poles_c\n{}'.format(cl_poles_c))
return cl_poles_d
def cplxpair(x, tol=100):
"""
Sort complex numbers into complex conjugate pairs.
This function replaces MATLAB's cplxpair for vectors.
"""
x = carray(x)
x = np.atleast_1d(x.squeeze())
x = x.tolist()
x = [np.real_if_close(i, tol) for i in x]
xreal = np.array(list(filter(np.isreal, x)))
xcomplex = np.array(list(filter(np.iscomplex, x)))
xreal = np.sort_complex(xreal)
xcomplex = np.sort_complex(xcomplex)
xcomplex_ipos = xcomplex[xcomplex.imag > 0.]
xcomplex_ineg = xcomplex[xcomplex.imag <= 0.]
if len(xcomplex_ipos) != len(xcomplex_ineg):
raise ValueError("Complex numbers can't be paired.")
res = []
for i, j in zip(xcomplex_ipos, xcomplex_ineg):
if not abs(i - np.conj(j)) < tol * eps:
raise ValueError("Complex numbers can't be paired.")
res += [j, i]
return np.hstack((np.array(res), xreal))
def cplxpair(x, tol=100):
"""
Sort complex numbers into complex conjugate pairs.
This function replaces MATLAB's cplxpair for vectors.
"""
x = carray(x)
x = np.atleast_1d(x.squeeze())
x = x.tolist()
x = [np.real_if_close(i, tol) for i in x]
xreal = np.array(list(filter(np.isreal, x)))
xcomplex = np.array(list(filter(np.iscomplex, x)))
xreal = np.sort_complex(xreal)
xcomplex = np.sort_complex(xcomplex)
xcomplex_ipos = xcomplex[xcomplex.imag > 0.]
xcomplex_ineg = xcomplex[xcomplex.imag <= 0.]
if len(xcomplex_ipos) != len(xcomplex_ineg):
raise ValueError("Complex numbers can't be paired.")
res = []
for i, j in zip(xcomplex_ipos, xcomplex_ineg):
if not abs(i - np.conj(j)) < tol * eps:
raise ValueError("Complex numbers can't be paired.")
res += [j, i]
return np.hstack((np.array(res), xreal))
def optimize_gains(robot, gains_array, q0, v0, niter=10):
nominal_gains_array = copy.deepcopy(gains_array)
H = compute_closed_loop_transition_matrix(gains_array, robot, q0, v0)
# step_response(H, N=10000, plot=0)
print "Initial gains:", gains_array.T
# print "Initial normalized gains:", normalize_gains_array(gains_array, nominal_gains_array).T
print "Initial eigenvalues:\n", np.sort_complex(eigvals(H)).T;
print "Initial cost", cost_function(np.ones_like(gains_array), robot, q0, v0, nominal_gains_array)
#optimize gain
global nit
nit = 0
opt_res = basinhopping(cost_function, np.ones_like(gains_array), niter, disp=False, T=0.1, stepsize=.01,
minimizer_kwargs={'args':(robot, q0, v0, nominal_gains_array)}, callback=callback)
opt_gains = denormalize_gains_array(opt_res.x, nominal_gains_array);
# print opt_res, '\n'
#Plot step response
H = compute_closed_loop_transition_matrix(opt_gains, robot, q0, v0)
# step_response(H, N=10000, plot=0)
print "Optimal gains: ", opt_gains
print "Optimal normalized gains:", normalize_gains_array(opt_gains, nominal_gains_array).T
print "Optimal eigenvalues:\n", list(np.sort_complex(eigvals(H)).T)
return opt_gains
def StateConcurrence(mat_4x4):
'''
Method that calculates the concurrence of any 2 qubit state, returns 0 if the
state is pure and returns max(0, E_1 - E_2 - E_3 - E_4) where E are the eigenvalues
in descending order of the matrix R = sqrt(sqrt(rho) rho_tilde sqrt(rho))
and rho_tilde is worked out using pauli_y outer pauli_y rho* pauli_y outer pauli_y
'''
pauli_y = np.zeros(shape=(2, 2), dtype=complex)
pauli_y[0, 1] = -1j
pauli_y[1, 0] = 1j
transform = np.outer(pauli_y, pauli_y)
buf = np.matmul(transform,
np.conj(mat_4x4)) # buffer variable for 3 matrix mult
rho_tilde = np.matmul(buf, transform)
root_rho = sqrtm(
mat_4x4
) # we now have all the matrices to make the argument in sqrt() for R
buf2 = np.matmul(root_rho, rho_tilde) # bufer variable for 3 matrix mult
root_arg = np.matmul(buf2, root_rho)
big_R = sqrtm(root_arg) # now we have R we work out its eigenvalues
try:
eigenvalue_array = np.linalg.eigvals(big_R)
except np.linalg.LinAlgError:
print(big_R)
print(mat_4x4)
return 0
#now we sort the eigenvalues into ascending order, taking their moduli
np.sort_complex(eigenvalue_array)
#print(eigenvalue_array)
#print(eigenvalue_array, 'EIGENS', len(eigenvalue_array))
concurrence_complex = 2 * eigenvalue_array[0] - np.sum(eigenvalue_array)
#floating point error means it will never be 0 so we return the mixed state
#concurrence and if it falls below a certain threshold we can call it "0"
return np.abs(concurrence_complex)
def test_dense_ising_T_real(seed_rng):
T = 3
J = np.random.normal()
g = np.random.normal()
h = np.random.normal()
init = np.random.normal(size=(2, 2)) + 1.0j * np.random.normal(size=(2, 2))
init = init.T.conj() + init
init = init @ init
init /= np.trace(init) #init is a proper density matrix
diT = dense.ising.ising_T(T, J, g, h, init)
assert diT.dtype == np.complex_
assert diT.shape == (4**T, 4**T)
ditev = la.eigvals(diT)
ditevc = np.zeros((4**T))
ditevc[-1] = 1.0
assert np.sort_complex(ditev) == pytest.approx(ditevc, rel=5e-4, abs=5e-4)
def roots(self):
"""return the roots of the characteristic equation.
possible cases for roots:
Case 1:
mu1 = a1 + j*b1
mu2 = a2 + j*b2
mu3 = cc(mu1)
mu4 = cc(mu2)
Case 2:
mu1 = mu2 = a + j*b
mu3 = mu4 = cc(mu1)
For the orthotropic case (A16 = A26 = 0): purely imaginary roots, ai = 0
"""
# LTH-FL 33100-05, or better: lehknitskij
# NOTE: this is for fluxes, not stresses
# get coefficients out of compliance matrix
a = np.linalg.inv(self._lam.A())
c4 = a[0, 0] # a11
c3 = -2 * a[0, 2] # -2*a16
c2 = 2 * a[0, 1] + a[2, 2] # 2*a12 + a66
c1 = -2 * a[1, 2] # -2*a26
c0 = a[1, 1] # a22
# build polynomial and calculate roots
# TODO: that algorithm is not too accurate -> find anything better??
poly = np.polynomial.Polynomial([c0, c1, c2, c3, c4])
roots = poly.roots()
# verify that they are correct: value <= TOL
TOL = 1e-9
assert all([abs(poly(r)) < TOL for r in roots])
# the 4 roots are either complex, then there is two pairs of congujate complex numbers.
# Or they are purely imaginary.
# In either case we want the roots with positive imaginary parts only.
#print('number of roots found:', len(roots))
#print(roots)
proots = roots[roots.imag > 0]
# make sure its 2 roots. Should be ...
assert len(proots) == 2
# set real part to zero if < 1e-9
proots.real[np.abs(proots.real) < 1e-9] = 0
return np.sort_complex(proots)
def test_dense_ising_T_pi4():
T = 3
J = np.pi / 4
g = np.pi / 4
h = 1.0
# init=np.random.normal(size=(2,2))+1.0j*np.random.normal(size=(2,2))
# init=init.T.conj()+init
# init=init@init
# init/=np.trace(init)#init is a proper density matrix
init = np.eye(2) / 2
diT = dense.ising.ising_T(T, J, g, h, init)
assert diT.dtype == np.complex_
assert diT.shape == (64, 64)
ditev, ditevv = la.eig(diT)
ditevc = np.zeros((4**T))
ditevc[-1] = 1.0
pdev = ditevv[:, np.argmax(np.abs(ditev))]
print(pdev / pdev[0])
assert np.sort_complex(ditev) == pytest.approx(ditevc, rel=5e-4, abs=5e-4)
assert pdev / pdev[0] == pytest.approx(dense.ising.perfect_dephaser_im(T))
def Complete_Format(self, Packet_Number):
if self.Check_effection(Packet_Number):
return 1
Bfee_count = self.Get_Bfee_count(Packet_Number)
Perm = self.Get_Perm(Packet_Number)
Nrx = self.Get_Nrx(Packet_Number)
Ntx = self.Get_Ntx(Packet_Number)
Noise = self.Get_Noise(Packet_Number)
RSSI = self.Get_RSSI(Packet_Number)
CSI = np.sort_complex(np.zeros((30, 3, 2)))
#self.CSI = np.complex(self.CSI)
CSI_Packet = self.Get_CSI(Packet_Number)
if (len(CSI_Packet) < 20):
return CSI_Packet
count = 0
for Subcarrier in range(30):
for Nrx_n in range(len(Nrx)):
for Ntx_n in range(len(Ntx)):
CSI[Subcarrier, Nrx_n, Ntx_n] = CSI_Packet[count]
count = count + 1
return [Bfee_count, Perm, Nrx, Ntx, Noise, RSSI, CSI]
def parfiltid(input, out, p, NFIR = 1): # We don't want to have any poles in the origin; For that we have the parallel FIR part. # Remove nonzeros p = p[p.nonzero()] # making the filter stable by flipping the poles into the unit circle for k in range(p.size): if abs(p[k]) > 1: p[k] = 1.0/np.conj(p[k]) # Order it to complex pole pairs + real ones afterwards p = np.sort_complex(p) # in order to have second-order sections only (i.e., no first order) pnum = len(p) # number of poles ppnum = 2 * np.floor(pnum/2) # the even part of pnum ODD = 0 #if pnum is odd if pnum > ppnum: ODD = 1 OUTL = len(out) INL = len(input) # making input the same length as the output if INL > OUTL: input = input[:OUTL] if INL < OUTL: input = np.hstack([input, np.zeros(OUTL - INL, dtype=np.float64)]) L = OUTL # Allocate memory M = np.zeros((input.size, p.size + NFIR), dtype=np.float64) # constructing the modeling signal matrix for k in range(0, int(ppnum), 2): #second-order sections #impluse response of the two-pole filter resp = sig.lfilter(np.array([1]), np.poly(p[k:k+2]), input) M[:,k] = resp #the response delayed by one sample M[:,k+1] = np.hstack((0., resp[:L-1])) # if the number of poles is odd, we have a first-order section if ODD: resp = sig.lfilter(np.array([1]), np.poly(p[-1]),input) M[:,pnum-1] = resp # parallel FIR part for k in range(0, NFIR): M[:,pnum+k] = np.hstack([np.zeros(k, dtype=np.float64), input[:L-k+1]]) y = out # Looking for min(||y-M*par||) as a function of par: # least squares solution by equation solving mconj = M.conj().T A = np.dot(mconj, M) b = np.dot(mconj, y) par = np.linalg.solve(A, b) #print (np.dot(A, par) == b).all() # Allocate memory size = int(np.ceil(ppnum/2)) Am = np.zeros((3, size), dtype=np.float64) Bm = np.zeros((2, size), dtype=np.float64) # constructing the Bm and Am matrices for k in range(0, size): Am[:,k] = np.poly(p[2*k:2*k+2]) Bm[:,k] = np.hstack(par[2*k:2*k+2]) # we extend the first-order section to a second-order one by adding zero coefficients if ODD: Am = np.append(Am, np.vstack(np.hstack([np.poly(p[pnum]),0.])), 1) Bm = np.append(Bm, np.vstack([par[pnum], 0.]), 1) FIR = [] # constructing the FIR part if NFIR > 0: FIR = np.hstack(par[pnum:pnum+NFIR]) return Bm, Am, FIR
import numpy as np np.sort_complex([5, 3, 6, 2, 1]) np.sort_complex([1 + 2j, 2 - 1j, 3 - 2j, 3 - 3j, 3 + 5j])
def calculate_singularities(mc, input_source=None, output_port=None, MNA=None,
x0=None, shift=0, outfile=None, verbose=0):
"""Calculate poles and zeros.
By default, only poles are calculated, as they need no information
other than the circuit description.
To activate zeros calculation, it is necessary:
* to specify an input source (``input_source``),
* to specify an output port (``output_port``).
**Parameters:**
mc : circuit instance
The circuit to be analyzed.
input_source : string or element, optional
If zeros are to be calculated, set this to the input surce.
output_port : external node (ref. to gnd) or tuple of external nodes, opt
If zeros are to be calculated, set this to the output nodes.
MNA : ndarray, optional
The Modified Nodal Analysis matrix, if available.
In case the circuit is non-linear, MNA should include the contributes
of the non-linear elements (ie the Jacobian :math:`J`).
x0 : ndarray or op_solution, optional
The linearization point. Only needed for non-linear circuits.
shift : float, optional
Shift frequency at which the algorithm should be run.
outfile : str or None, optional
The data filename.
verbose : int, optional
Verbosity level, from 0 (silent, default) to 6 (debug).
**Returns:**
pz_sol : pz_solution instance
The PZ solution
"""
calc_zeros = (input_source is not None) and (output_port is not None)
if calc_zeros:
if type(input_source) not in py3compat.string_types:
input_source = input_source.part_id
if type(output_port) in py3compat.string_types:
output_port = plotting._split_netlist_label(output_port)[0]
output_port = [o[1:] for o in output_port]
output_port = [o.lower() for o in output_port]
if len(output_port) == 1:
# we refer to the ground implicitely
output_port += ['0']
if np.isscalar(output_port):
output_port = (output_port, mc.gnd)
if (type(output_port) == tuple or type(output_port) == list) \
and type(output_port[0]) in py3compat.string_types:
output_port = [mc.ext_node_to_int(o) for o in output_port]
we_got_source = False
for e in mc:
if e.part_id == input_source:
we_got_source = True
break
if not we_got_source:
raise ValueError('Source %s not found in circuit.' % input_source)
RIIN = []
ROUT = []
if MNA is None:
MNA, N = dc_analysis.generate_mna_and_N(mc)
if mc.is_nonlinear():
# setup x0
if x0 is None:
printing.print_warning("PZ: No linearization point provided. Using x0 = 0.")
x0 = np.zeros((MNA.shape[0] - 1, 1))
else:
if isinstance(x0, results.op_solution):
x0 = x0.asarray()
# else
# hopefully x0 is an ndarray!
printing.print_info_line(("Using the supplied op as " +
"linearization point.", 5), verbose)
J, _ = dc_analysis.build_J_and_Tx(x0, MNA.shape[0]-1, mc, time=0., sparse=False)
MNA[1:, 1:] += J
D = transient.generate_D(mc, MNA[1:, 1:].shape)
MNAinv = np.linalg.inv(MNA[1:, 1:] + shift*D[1:, 1:])
nodes_m1 = mc.get_nodes_number() - 1
vde1 = -1
MC = np.zeros((MNA.shape[0] - 1, 1))
TCM = None
dei_source = 0
for e1 in mc:
if circuit.is_elem_voltage_defined(e1):
vde1 += 1
if isinstance(e1, components.Capacitor):
MC[e1.n1 - 1, 0] += 1. if e1.n1 > 0 else 0.
MC[e1.n2 - 1, 0] -= 1. if e1.n2 > 0 else 0.
elif isinstance(e1, components.Inductor):
MC[nodes_m1 + vde1] += -1.
elif calc_zeros and e1.part_id == input_source:
if isinstance(e1, components.sources.VSource):
MC[nodes_m1 + vde1] += -1.
elif isinstance(e1, components.sources.ISource):
MC[e1.n1 - 1, 0] += 1. if e1.n1 > 0 else 0.
MC[e1.n2 - 1, 0] -= 1. if e1.n2 > 0 else 0.
else:
raise Exception("Unknown input source type %s" % input_source)
else:
continue
TV = -1. * np.dot(MNAinv, MC)
dei_victim = 0
vde2 = -1
for e2 in mc:
if circuit.is_elem_voltage_defined(e2):
vde2 += 1
if isinstance(e2, components.Capacitor):
v = 0
if e2.n1:
v += TV[e2.n1 - 1, 0]
if e2.n2:
v -= TV[e2.n2 - 1, 0]
elif isinstance(e2, components.Inductor):
v = TV[nodes_m1 + vde2, 0]
else:
continue
if calc_zeros and e1.part_id == input_source:
RIIN += [v]
else:
if not dei_source:
TCM = _enlarge_matrix(TCM)
TCM[dei_victim, dei_source] = v*e1.value
dei_victim += 1
if calc_zeros and e1.part_id == input_source:
ROUTIN = 0
o1, o2 = output_port
if o1:
ROUTIN += TV[o1 - 1, 0]
if o2:
ROUTIN -= TV[o2 - 1, 0]
else:
dei_source += 1
# reset, get ready to restart
MC[:, :] = 0.
if TCM is not None:
if np.linalg.det(TCM):
poles = 1./(2.*np.pi)*(1./np.linalg.eigvals(TCM) + shift)
else:
return calculate_singularities(mc, input_source, output_port, MNA=MNA,
outfile=outfile,
shift=shift+np.abs(1+np.random.uniform())*1e3)
else:
poles = []
if calc_zeros and TCM is not None:
# re-loop, get the ROUT elements
vde1 = -1
MC = np.zeros((MNA.shape[0] - 1, 1))
ROUT = []
for e1 in mc:
if circuit.is_elem_voltage_defined(e1):
vde1 += 1
if isinstance(e1, components.Capacitor):
MC[e1.n1 - 1, 0] += 1. if e1.n1 > 0 else 0.
MC[e1.n2 - 1, 0] -= 1. if e1.n2 > 0 else 0.
elif isinstance(e1, components.Inductor):
MC[nodes_m1 + vde1, 0] += -1.
else:
continue
TV = -1.*np.dot(MNAinv, MC)
v = 0
o1, o2 = output_port
if o1:
v += TV[o1 - 1, 0]
if o2:
v -= TV[o2 - 1, 0]
ROUT += [v*e1.value]
# reset, get ready to restart
MC[:, :] = 0.
# Reshape the matrices and evaluate the zero correction.
RIIN = np.array(RIIN).reshape((-1, 1))
RIIN = np.tile(RIIN, (1, RIIN.shape[0]))
ROUT = np.diag(np.atleast_1d(np.array(ROUT)))
if ROUT.any():
try:
if not np.all(ROUTIN) or not np.isfinite(ROUTIN):
# immediate catch
raise ValueError("ROUT-IN is either Inf, NaN or null")
ZTCM = TCM - np.dot(RIIN, ROUT)/ROUTIN
if not (np.isfinite(ZTCM).all()):
raise ValueError("Array contains infs, NaNs or both")
# immediate catch
eigvals = np.linalg.eigvals(ZTCM)
if not np.all(eigvals) or not np.isfinite(eigvals).all():
# immediate catch
raise ValueError("ZTCM eigenvalues contain either Inf, NaN or null values")
zeros = 1./(2.*np.pi)*(1./np.linalg.eigvals(ZTCM) + shift)
except ValueError:
return calculate_singularities(mc, input_source, output_port,
MNA=MNA, outfile=outfile,
shift=shift+np.abs(np.random.uniform()+1)*1e3)
elif shift < 1e12:
return calculate_singularities(mc, input_source, output_port,
MNA=MNA, outfile=outfile,
shift=shift*np.abs(np.random.uniform()+1)*10)
else:
zeros = []
else:
zeros = []
poles = np.array([a for a in poles if np.abs(a) < options.pz_max], dtype=np.complex_)
zeros = np.array([a for a in zeros if np.abs(a) < options.pz_max], dtype=np.complex_)
poles = np.sort_complex(poles)
zeros = np.sort_complex(zeros)
res = results.pz_solution(mc, poles, zeros, outfile)
return res
def datestr2num(s):
return datetime.datetime.strptime(s,"%d-%m-%Y").toordinal()
dates,closes = np.loadtxt('AAPL.csv',delimiter=',',usecols=(1,6),\
converters={1:datestr2num},unpack=True)
indices =np.lexsort((dates,closes))
print "Indices",indices
print ["%s %s" % (datetime.date.fromordinal(int(dates[i])),closes[i]) for i in indices]
print u"复数"
#设置随机种子
np.random.seed(42)
complex_numbers = np.random.random(5) + 1j*np.random.random(5)
print "Complex numbers\n",complex_numbers
print "Sorted\n",np.sort_complex(complex_numbers)
print u"搜索"
#argmax返回数组中最大值对应的下标
a= np.array([2,4,8])
print "argmax",np.argmax(a)
#nanargmax提供相同功能，但忽略NaN值
#argmin和nanargmin功能类似，只是换成最小值
#argwhere根据条件搜索非零元素，并分组返回对应下标
#searchsorted可以为指定的插入值寻找维持数组排序的索引位置
#这个位置可以保持数组有序性
a =np.arange(5)
indices = np.searchsorted(a,[-2,7])
print "Indices",indices
print "The full array", np.insert(a,indices,[-2,7])
print u"抽取元素"
def pair_complex_numbers(a, tol=1e-9, realness_tol=1e-9,
positives_first=False, reals_first=True):
"""
Given an array-like somearray, it first tests and clears out small
imaginary parts via `numpy.real_if_close`. Then pairs complex numbers
together as consecutive entries. A real array is returned as is.
Parameters
----------
a : array_like
Array like object needs to be paired
tol: float
The sensitivity threshold for the real and complex parts to be
assumed as equal.
realness_tol: float
The sensitivity threshold for the complex parts to be assumed
as zero.
positives_first: bool
The boolean that defines whether the positive complex part
should come first for the conjugate pairs
reals_first: bool
The boolean that defines whether the real numbers are at the
beginning or the end of the resulting array.
Returns
-------
paired_array : ndarray
The resulting paired array
"""
try:
array_r_j = np.array(a, dtype='complex').flatten()
except:
raise ValueError('Something in the argument array prevents me to '
'convert the entries to complex numbers.')
# is there anything to pair?
if array_r_j.size == 0:
return np.array([], dtype='complex')
# is the array 1D or more?
if array_r_j.ndim > 1 and np.min(array_r_j.shape) > 1:
raise ValueError('Currently, I can\'t deal with matrices, so I '
'need 1D arrays.')
# A shortcut for splitting a complex array into real and imag parts
def return_imre(arr):
return np.real(arr), np.imag(arr)
# a close to realness function that operates element-wise
real_if_close_array = np.vectorize(
lambda x: np.real_if_close(x, realness_tol), otypes=[np.complex_],
doc='Elementwise numpy.real_if_close')
array_r_j = real_if_close_array(array_r_j)
array_r, array_j = return_imre(array_r_j)
# are there any complex numbers to begin with or all reals?
# if not kick the argument back as a real array
imagness = np.abs(array_j) >= realness_tol
# perform the imaginary entry separation once
array_j_ent = array_r_j[imagness]
num_j_ent = array_j_ent.size
if num_j_ent == 0:
# If no complex entries exist sort and return unstable first
return np.sort(array_r)
elif num_j_ent % 2 != 0:
# Check to make sure there are even number of complex numbers
# Otherwise stop with "odd number --> no pair" error.
raise ValueError('There are odd number of complex numbers to '
'be paired!')
else:
# Still doesn't distinguish whether they are pairable or not.
sorted_array_r_j = np.sort_complex(array_j_ent)
sorted_array_r, sorted_array_j = return_imre(sorted_array_r_j)
# Since the entries are now sorted and appear as pairs,
# when summed with the next element the result should
# be very small
if any(np.abs(sorted_array_r[:-1:2] - sorted_array_r[1::2]) > tol):
# if any difference is bigger than the tolerance
raise ValueError('Pairing failed for the real parts.')
# Now we have sorted the real parts and they appear in pairs.
# Next, we have to get rid of the repeated imaginary, if any,
# parts in the --... , ++... pattern due to sorting. Note
# that the non-repeated imaginary parts now have the pattern
# -,+,-,+,-,... and so on. So we can check whether sign is
# not alternating for the existence of the repeatedness.
def repeat_sign_test(myarr, mylen):
# Since we separated the zero imaginary parts now any sign
# info is either -1 or 1. Hence we can test whether -1,1
# pattern is present. Otherwise we count how many -1s occured
# double it for the repeated region. Then repeat until we
# we exhaust the array with a generator.
x = 0
myarr_sign = np.sign(myarr).astype(int)
while x < mylen:
if np.array_equal(myarr_sign[x:x+2], [-1, 1]):
x += 2
elif np.array_equal(myarr_sign[x:x+2], [1, -1]):
myarr[x:x+2] *= -1
x += 2
else:
still_neg = True
xl = x+2
while still_neg:
if myarr_sign[xl] == 1:
still_neg = False
else:
xl += 1
yield x, xl - x
x += 2*(xl - x)
for ind, l in repeat_sign_test(sorted_array_j, num_j_ent):
indices = np.dstack(
(range(ind, ind+l), range(ind+2*l-1, ind+l-1, -1))
)[0].reshape(1, -1)
sorted_array_j[ind:ind+2*l] = sorted_array_j[indices]
if any(np.abs(sorted_array_j[:-1:2] + sorted_array_j[1::2]) > tol):
# if any difference is bigger than the tolerance
raise ValueError('Pairing failed for the complex parts.')
# Finally we have a properly sorted pairs of complex numbers
# We can now combine the real and complex parts depending on
# the choice of positives_first keyword argument
# Force entries to be the same for each of the pairs.
sorted_array_j = np.repeat(sorted_array_j[::2], 2)
paired_cmplx_part = np.repeat(sorted_array_r[::2], 2).astype(complex)
if positives_first:
sorted_array_j[::2] *= -1
else:
sorted_array_j[1::2] *= -1
paired_cmplx_part += sorted_array_j*1j
if reals_first:
return np.r_[np.sort(array_r_j[~imagness]), paired_cmplx_part]
else:
return np.r_[paired_cmplx_part, np.sort(array_r_j[~imagness])]
def pretty_lti(arg):
"""Given the lti object ``arg`` return a *pretty* representation."""
z, p, k = _get_zpk(arg)
z = np.atleast_1d(z)
p = np.atleast_1d(p)
z = np.round(np.real_if_close(z), 4)
p = np.round(np.real_if_close(p), 4)
k = np.round(np.real_if_close(k), 4)
signs = {1:'+', -1:'-'}
if not len(z) and not len(p):
return "%g" % k
ppstr = ["", "", ""]
if np.allclose(k, 0., atol=1e-5):
return "0"
if k != 1:
if np.isreal(k):
ppstr[1] = "%g " % k
else:
# quadrature modulators support
ppstr[1] += "(%g %s %gj) " % (np.real(k),
signs[np.sign(np.imag(k))],
np.abs(np.imag(k)))
for i, s in zip((0, 2), (z, p)):
rz = None
m = 1
try:
sorted_singularities = cplxpair(s)
quadrature = False
except ValueError:
# quadrature modulator
sorted_singularities = np.sort_complex(s)
quadrature = True
for zindex, zi in enumerate(sorted_singularities):
zi = np.round(np.real_if_close(zi), 4)
if np.isreal(zi) or quadrature:
if len(sorted_singularities) > zindex + 1 and \
sorted_singularities[zindex + 1] == zi:
m += 1
continue
if zi == 0.:
ppstr[i] += "z"
elif np.isreal(zi):
ppstr[i] += "(z %s %g)" % (signs[np.sign(-zi)], np.abs(zi))
else:
ppstr[i] += "(z %s %g %s %gj)" % (signs[np.sign(np.real(-zi))],
np.abs(np.real(zi)),
signs[np.sign(np.imag(-zi))],
np.abs(np.imag(zi)))
if m == 1:
ppstr[i] += " "
else:
ppstr[i] += "^%d " % m
m = 1
else:
if len(sorted_singularities) > zindex + 2 and \
sorted_singularities[zindex + 2] == zi:
m += .5
continue
if rz is None:
rz = zi
continue
ppstr[i] += "(z^2 %s %gz %s %g)" % \
(signs[np.sign(np.real_if_close(np.round(-rz - zi, 3)))],
np.abs(np.real_if_close(np.round(-rz - zi, 3))),
signs[np.sign(np.real_if_close(np.round(rz * zi, 4)))],
np.abs(np.real_if_close(np.round(rz * zi, 4))))
if m == 1:
ppstr[i] += " "
else:
ppstr[i] += "^%d " % m
rz = None
m = 1
ppstr[i] = ppstr[i][:-1] if len(ppstr[i]) else "1"
if ppstr[2] == '1':
return ppstr[1] + ppstr[0]
else:
if ppstr[0] == '1' and len(ppstr[1]) and float(ppstr[1]) != 1.:
ppstr[0] = ppstr[1][:-1]
ppstr[1] = ""
space_pad_ln = len(ppstr[1])
fraction_line = "-" * (max(len(ppstr[0]), len(ppstr[2])) + 2)
ppstr[1] += fraction_line
ppstr[0] = " "*space_pad_ln + ppstr[0].center(len(fraction_line))
ppstr[2] = " "*space_pad_ln + ppstr[2].center(len(fraction_line))
return "\n".join(ppstr)