def fixNegsConstrainedOpt(self, to_minimize=norm_diff, badness=1e6):
"""Uses constrained minimization to find approx q matrix.
to_minimize: metric for comparing orig result and new result.
badness: scale factor for penalizing negative off-diagonal values.
"""
if not sum_neg_off_diags(self._data):
return self
q = ravel(without_diag(self._data))
p = expm(self._data)(t=1)
def err_f(q):
new_q = reshape(array(q), (4, 3))
new_q = with_diag(new_q, -sum(new_q, 1))
p_new = expm(new_q)(t=1)
result = to_minimize(ravel(p), ravel(p_new))
if q.min() < 0:
result += -q.min() * badness
return result
a = array(q)
xmin = fmin(func=err_f, x0=a, disp=0)
r = reshape(xmin, (4, 3))
new_q = with_diag(r, -sum(r, 1))
return self.__class__(new_q, self.Alphabet)
def err_f(q):
new_q = reshape(array(q), (4,3))
new_q = with_diag(new_q, -sum(new_q, 1))
p_new = expm(new_q)(t=1)
result = to_minimize(ravel(p), ravel(p_new))
if q.min() < 0:
result += -q.min() * badness
return result
def err_f(q):
new_q = reshape(array(q), (4, 3))
new_q = with_diag(new_q, -sum(new_q, 1))
p_new = expm(new_q)(t=1)
result = to_minimize(ravel(p), ravel(p_new))
if q.min() < 0:
result += -q.min() * badness
return result
def fixNegsEven(self):
"""Returns copy of self w/o negative off-diags, using 'even' heuristic.
If a negative off-diagonal is encountered, sets it to 0.
Distributes the negative score evenly among the other elements.
"""
m = without_diag(self._data)
for i, row in enumerate(m):
is_neg = row < 0
if any(is_neg):
num_negs = sum(is_neg)
sum_negs = sum(is_neg*row)
is_not_neg = logical_not(is_neg)
num_not_neg = sum(is_not_neg)
new_row = (row + (sum_negs/(num_not_neg+1)))*is_not_neg
m[i] = new_row
return self.__class__(with_diag(m, -sum(m,1)), self.Alphabet)
def fixNegsEven(self):
"""Returns copy of self w/o negative off-diags, using 'even' heuristic.
If a negative off-diagonal is encountered, sets it to 0.
Distributes the negative score evenly among the other elements.
"""
m = without_diag(self._data)
for i, row in enumerate(m):
is_neg = row < 0
if any(is_neg):
num_negs = sum(is_neg)
sum_negs = sum(is_neg * row)
is_not_neg = logical_not(is_neg)
num_not_neg = sum(is_not_neg)
new_row = (row + (sum_negs / (num_not_neg + 1))) * is_not_neg
m[i] = new_row
return self.__class__(with_diag(m, -sum(m, 1)), self.Alphabet)
def fixNegsConstrainedOpt(self, to_minimize=norm_diff, badness=1e6):
"""Uses constrained minimization to find approx q matrix.
to_minimize: metric for comparing orig result and new result.
badness: scale factor for penalizing negative off-diagonal values.
"""
if not sum_neg_off_diags(self._data):
return self
q = ravel(without_diag(self._data))
p = expm(self._data)(t=1)
def err_f(q):
new_q = reshape(array(q), (4,3))
new_q = with_diag(new_q, -sum(new_q, 1))
p_new = expm(new_q)(t=1)
result = to_minimize(ravel(p), ravel(p_new))
if q.min() < 0:
result += -q.min() * badness
return result
a = array(q)
xmin = fmin(func=err_f, x0=a, disp=0)
r = reshape(xmin, (4,3))
new_q = with_diag(r, -sum(r, 1))
return self.__class__(new_q, self.Alphabet)
def test_with_diag(self):
"""with_diag should add diagonal to matrix"""
a = array([[2,3],[4,6],[7,8]])
b = with_diag(a, array([1,5,9]))
self.assertEqual(b, array([[1,2,3],[4,5,6],[7,8,9]]))
