def get_scaled_fixation_probabilities(gammas):
"""
The scaling factor is the same for each allele.
It is something like 2N.
@param gammas: scaled selections near zero, positive is better
@return: a matrix of fixation probabilities
"""
k = len(gammas)
F = np.zeros((k, k))
for i, gi in enumerate(gammas):
for j, gj in enumerate(gammas):
if i == j:
continue
F[i, j] = bernoulli.bgf(gj - gi)
return F
def get_scaled_fixation_probabilities(gammas):
"""
The scaling factor is the same for each allele.
It is something like 2N.
@param gammas: scaled selections near zero, positive is better
@return: a matrix of fixation probabilities
"""
k = len(gammas)
F = np.zeros((k, k))
for i, gi in enumerate(gammas):
for j, gj in enumerate(gammas):
if i == j:
continue
F[i, j] = bernoulli.bgf(gj - gi)
return F
def to_gtr_hb_known_energies(M, u, v):
"""
The hb in the function name stands for Halpern-Bruno.
@param M: a time-reversible rate matrix
@param u: energies of states of M
@param v: target energies
@return: a time-reversible rate matrix
"""
n = len(v)
R = M.copy()
# adjust the entries of the rate matrix
for a in range(n):
for b in range(n):
if a != b:
du = u[b] - u[a]
dv = v[b] - v[a]
log_tau = du - dv
coeff = bernoulli.bgf(-log_tau)
R[a, b] *= coeff
# reset the diagonal entries of the rate matrix
R -= np.diag(np.sum(R, axis=1))
return R
def to_gtr_hb_known_energies(M, u, v):
"""
The hb in the function name stands for Halpern-Bruno.
@param M: a time-reversible rate matrix
@param u: energies of states of M
@param v: target energies
@return: a time-reversible rate matrix
"""
n = len(v)
R = M.copy()
# adjust the entries of the rate matrix
for a in range(n):
for b in range(n):
if a != b:
du = u[b] - u[a]
dv = v[b] - v[a]
log_tau = du - dv
coeff = bernoulli.bgf(-log_tau)
R[a, b] *= coeff
# reset the diagonal entries of the rate matrix
R -= np.diag(np.sum(R, axis=1))
return R
