def calcUnconstrainedLogLikelihood1(self):
"""Calculate likelihood under the multinomial model.
This calculates the unconstrained (multinomial) log like
without regard to character partitions. The result is placed
in the data variable unconstrainedLogLikelihood. If there is
more than one partition, it makes a new temporary alignment
and puts all the sequences in one part in that alignment. So
it ultimately only works on one data partition. If there is
more than one alignment, there is possibly more than one
datatype, and so this method will refuse to do it. Note that
the unconstrained log like of the combined data is not the sum
of the unconstrained log likes of the separate partitions.
See also calcUnconstrainedLogLikelihood2
"""
if len(self.alignments) > 1:
gm = ["Data.calcUnconstrainedLogLikelihood()"]
gm.append(
"This method is not implemented for more than one alignment.")
raise Glitch, gm
if self.nParts == 1: # no problem
self.unconstrainedLogLikelihood = pf.getUnconstrainedLogLike(
self.parts[0].cPart)
else:
a = self.alignments[0]
import copy
newAlig = Alignment()
newAlig.dataType = a.dataType
newAlig.symbols = a.symbols
newAlig.dim = a.dim
newAlig.equates = a.equates
newAlig.taxNames = a.taxNames
for s in a.sequences:
newAlig.sequences.append(copy.deepcopy(s))
newAlig.checkLengthsAndTypes()
newAlig._initParts()
#newAlig.dump()
self.unconstrainedLogLikelihood = pf.getUnconstrainedLogLike(
newAlig.parts[0].cPart)
del (newAlig)
def calcUnconstrainedLogLikelihood1(self):
"""Calculate likelihood under the multinomial model.
This calculates the unconstrained (multinomial) log like
without regard to character partitions. The result is placed
in the data variable unconstrainedLogLikelihood. If there is
more than one partition, it makes a new temporary alignment
and puts all the sequences in one part in that alignment. So
it ultimately only works on one data partition. If there is
more than one alignment, there is possibly more than one
datatype, and so this method will refuse to do it. Note that
the unconstrained log like of the combined data is not the sum
of the unconstrained log likes of the separate partitions.
See also calcUnconstrainedLogLikelihood2
"""
if len(self.alignments) > 1:
gm = ["Data.calcUnconstrainedLogLikelihood()"]
gm.append("This method is not implemented for more than one alignment.")
raise Glitch, gm
if self.nParts == 1: # no problem
self.unconstrainedLogLikelihood = pf.getUnconstrainedLogLike(self.parts[0].cPart)
else:
a = self.alignments[0]
import copy
newAlig = Alignment()
newAlig.dataType = a.dataType
newAlig.symbols = a.symbols
newAlig.dim = a.dim
newAlig.equates = a.equates
newAlig.taxNames = a.taxNames
for s in a.sequences:
newAlig.sequences.append(copy.deepcopy(s))
newAlig.checkLengthsAndTypes()
newAlig._initParts()
#newAlig.dump()
self.unconstrainedLogLikelihood = pf.getUnconstrainedLogLike(newAlig.parts[0].cPart)
del(newAlig)
