def DetModel(string):
from dcprogs.likelihood import DeterminantEq
string = string.lower().rstrip().lstrip()
if 'transpose' in string:
return DetModel(string.replace('transpose', '')).transpose()
if string == "classic": return DeterminantEq(QMat(string), 1e-4)
if string == "ch82": return DeterminantEq(QMat(string), 0.2)
if string == "cb": return DeterminantEq(QMat(string), 0.2)
if string == "cks": return DeterminantEq(QMat(string), 0.2)
else: raise Exception("Unknown eG model {0}".format(string))
def step(context):
from dcprogs.likelihood import QMatrix, DeterminantEq
qmatrix = QMatrix(
[[
-1.89907444e+02, 0.00000000e+00, 2.17917781e+01, 1.68115666e+02,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00
],
[
0.00000000e+00, -5.59197168e+02, 0.00000000e+00, 0.00000000e+00,
2.28813670e+01, 5.36315801e+02, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00
],
[
1.70657017e+04, 0.00000000e+00, -1.70657017e+04, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00
],
[
6.93510518e+04, 0.00000000e+00, 0.00000000e+00, -7.70252433e+04,
0.00000000e+00, 0.00000000e+00, 7.67419154e+03, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00
],
[
0.00000000e+00, 1.62124166e+04, 0.00000000e+00, 0.00000000e+00,
-1.62124166e+04, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00
],
[
0.00000000e+00, 7.15538758e+04, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, -1.06690747e+06, 9.95292553e+05, 0.00000000e+00,
6.10438859e+01, 0.00000000e+00, 0.00000000e+00
],
[
0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 1.45242879e+04,
0.00000000e+00, 3.69900307e+05, -3.87202655e+05, 2.77806009e+03,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00
],
[
0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 2.01908451e+03, -2.28998927e+03,
2.70904758e+02, 0.00000000e+00, 0.00000000e+00
],
[
0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 2.87990000e+04, 0.00000000e+00, 1.09766807e+02,
-2.90893700e+04, 1.80603172e+02, 0.00000000e+00
],
[
0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
2.19533614e+02, -3.09835200e+02, 9.03015859e+01
],
[
0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 3.29300422e+02, -3.29300422e+02
]], 3)
context.determinant = DeterminantEq(qmatrix, 4e-5).transpose()
def step(context, tau):
from sys import exc_info
from dcprogs.likelihood import DeterminantEq
try:
context.determinant = DeterminantEq(context.qmatrix, tau)
except:
context.initialization_exception = exc_info()
def step(context):
from sys import exc_info
from dcprogs.likelihood import DeterminantEq
print(context.matrix)
try:
context.determinant = DeterminantEq(context.matrix, context.nopen,
context.tau)
except:
context.initialization_exception = exc_info()
def step(context, n):
from dcprogs.likelihood import DeterminantEq
from dcprogs.likelihood.random import qmatrix as random_qmatrix
context.matrices = []
context.equations = []
while len(context.equations) < n:
try:
matrix = random_qmatrix()
equation = DeterminantEq(matrix, 1e-4)
except:
continue
else:
context.matrices.append(matrix)
context.equations.append(equation)
from numpy import all, abs, arange
from dcprogs.likelihood import QMatrix, DeterminantEq, MissedEventsG
# Define parameters.
qmatrix = QMatrix([ [-3050, 50, 3000, 0, 0],
[2./3., -1502./3., 0, 500, 0],
[ 15, 0, -2065, 50, 2000],
[ 0, 15000, 4000, -19000, 0],
[ 0, 0, 10, 0, -10] ], 2)
tau = 1e-4
# Create eG from prior knowledge of roots
determinant_eq = DeterminantEq(qmatrix, tau)
af_roots = [( -3045.285776037674, 1), (-162.92946543451328, 1)]
fa_roots = [(-17090.192769236815, 1), (-2058.0812921673496, 1), (-0.24356535498785126, 1)]
eG_from_roots = MissedEventsG(determinant_eq, af_roots, determinant_eq.transpose(), fa_roots)
# Create eG automaticallye
eG_automatic = MissedEventsG(qmatrix, tau)
# Checks the three initialization are equivalent at tau
assert all(abs(eG_from_roots.af(tau) - eG_automatic.af(tau)) < 1e-8)
assert all(abs(eG_from_roots.fa(tau) - eG_automatic.fa(tau)) < 1e-8)
# Checks the three initialization are equivalent at different times
# The functions can be applied to arrays.
times = arange(tau, 10*tau, 0.1*tau)
assert eG_from_roots.af(times).shape == (len(times), 2, 3)
assert eG_from_roots.fa(times).shape == (len(times), 3, 2)
assert all(abs(eG_from_roots.af(times) - eG_automatic.af(times)) < 1e-8)
assert all(abs(eG_from_roots.fa(times) - eG_automatic.fa(times)) < 1e-8)
from numpy import abs, all, array
from dcprogs import internal_dtype
from dcprogs.likelihood import DeterminantEq, QMatrix
# Define parameters.
matrix = [ [-3050, 50, 3000, 0, 0],
[2./3., -1502./3., 0, 500, 0],
[ 15, 0, -2065, 50, 2000],
[ 0, 15000, 4000, -19000, 0],
[ 0, 0, 10, 0, -10] ]
qmatrix = QMatrix(matrix, 2)
tau = 1e-4
det0 = DeterminantEq(qmatrix, 1e-4);
det1 = DeterminantEq(matrix, 2, 1e-4);
print("{0!s}\n\n{1!s}".format(det0, det1))
x = [0, -1, -1e2]
assert all(abs(det0(x) - det1(x)) < 1e-8)
roots = array([-3045.285776037674, -162.92946543451328])
assert all(abs(det0(roots)) < 1e-6 * abs(roots))
transpose = det0.transpose()
roots = array( [-17090.192769236815, -2058.0812921673496, -0.24356535498785126],
dtype=internal_dtype )
assert all(abs(transpose(roots)) < 1e-6 * abs(roots))
print(" * H({0}):\n{1}\n".format(0, det0.H(0)))
print(" * H({0}):\n{1}\n".format(-1e2, det0.H(-1e2)))
def step(context, event, s, tau):
import numpy
from dcprogs.likelihood import DeterminantEq
s = eval(s, globals().copy(), numpy.__dict__.copy())
if event == "open": context.result = DeterminantEq(context.qmatrix, tau)(s)
else: context.result = DeterminantEq(context.qmatrix.transpose(), tau)(s)
def step(context, s):
from dcprogs.likelihood import DeterminantEq
context.result = DeterminantEq(context.matrix, context.nopen,
context.tau)(s)
from numpy import all
from dcprogs.likelihood import eig
from dcprogs.likelihood import find_upper_bound_for_roots, find_lower_bound_for_roots, \
find_root_intervals, brentq, find_roots, QMatrix, DeterminantEq
qmatrix = QMatrix([ [-3050, 50, 3000, 0, 0],
[2./3., -1502./3., 0, 500, 0],
[ 15, 0, -2065, 50, 2000],
[ 0, 15000, 4000, -19000, 0],
[ 0, 0, 10, 0, -10] ], 2)
det = DeterminantEq(qmatrix, 1e-4);
upper_bound = find_upper_bound_for_roots(det);
lower_bound = find_lower_bound_for_roots(det);
get_eigenvalues = lambda s: eig(det.H(s))[0].T
assert all(get_eigenvalues(lower_bound) > lower_bound)
assert all(get_eigenvalues(upper_bound) < upper_bound)
print("Root Determination\n" \
"==================\n\n" \
" * Interval containing roots: {lower}, {upper}\n" \
" * Eigenvalues of H at lower bound: {eiglow}\n" \
" * Eigenvalues of H at upper bound: {eigup}\n" \
.format( lower=lower_bound, upper=upper_bound,
eiglow = get_eigenvalues(lower_bound),
eigup = get_eigenvalues(upper_bound) ))
intervals = find_root_intervals(det, lower_bound, upper_bound)
def step(context, doopen, qmatrix, tau):
from dcprogs.likelihood import DeterminantEq
context.equation = DeterminantEq(qmatrix, tau)
if doopen != "open": context.equation = context.equation.transpose()
print(context.equation)
from numpy import abs, all, array
from dcprogs import internal_dtype
from dcprogs.likelihood import DeterminantEq, QMatrix
# Define parameters.
matrix = [[-3050, 50, 3000, 0, 0], [2. / 3., -1502. / 3., 0, 500, 0],
[15, 0, -2065, 50, 2000], [0, 15000, 4000, -19000, 0],
[0, 0, 10, 0, -10]]
qmatrix = QMatrix(matrix, 2)
tau = 1e-4
det0 = DeterminantEq(qmatrix, 1e-4)
det1 = DeterminantEq(matrix, 2, 1e-4)
print("{0!s}\n\n{1!s}".format(det0, det1))
x = [0, -1, -1e2]
assert all(abs(det0(x) - det1(x)) < 1e-8)
roots = array([-3045.285776037674, -162.92946543451328])
assert all(abs(det0(roots)) < 1e-6 * abs(roots))
transpose = det0.transpose()
roots = array([-17090.192769236815, -2058.0812921673496, -0.24356535498785126],
dtype=internal_dtype)
assert all(abs(transpose(roots)) < 1e-6 * abs(roots))
print(" * H({0}):\n{1}\n".format(0, det0.H(0)))
print(" * H({0}):\n{1}\n".format(-1e2, det0.H(-1e2)))
print(" * d[sI-H(s)]/ds for s={0}:\n{1}\n".format(0, det0.s_derivative(0)))
print(" * d[sI-H(s)]/ds for s={0}:\n{1}\n".format(-1e2,
from numpy import all, abs, arange
from dcprogs.likelihood import QMatrix, DeterminantEq, MissedEventsG
# Define parameters.
qmatrix = QMatrix([[-3050, 50, 3000, 0, 0], [2. / 3., -1502. / 3., 0, 500, 0],
[15, 0, -2065, 50, 2000], [0, 15000, 4000, -19000, 0],
[0, 0, 10, 0, -10]], 2)
tau = 1e-4
# Create eG from prior knowledge of roots
determinant_eq = DeterminantEq(qmatrix, tau)
af_roots = [(-3045.285776037674, 1), (-162.92946543451328, 1)]
fa_roots = [(-17090.192769236815, 1), (-2058.0812921673496, 1),
(-0.24356535498785126, 1)]
eG_from_roots = MissedEventsG(determinant_eq, af_roots,
determinant_eq.transpose(), fa_roots)
# Create eG automaticallye
eG_automatic = MissedEventsG(qmatrix, tau)
# Checks the three initialization are equivalent at tau
assert all(abs(eG_from_roots.af(tau) - eG_automatic.af(tau)) < 1e-8)
assert all(abs(eG_from_roots.fa(tau) - eG_automatic.fa(tau)) < 1e-8)
# Checks the three initialization are equivalent at different times
# The functions can be applied to arrays.
times = arange(tau, 10 * tau, 0.1 * tau)
assert eG_from_roots.af(times).shape == (len(times), 2, 3)
assert eG_from_roots.fa(times).shape == (len(times), 3, 2)
assert all(abs(eG_from_roots.af(times) - eG_automatic.af(times)) < 1e-8)
assert all(abs(eG_from_roots.fa(times) - eG_automatic.fa(times)) < 1e-8)
