def toCIspinorbReverse(norb, ne, civec2):
print '\n[toCIspinorbReverse]: spin-orbit rep C(Ia,Ib)<-C(I)'
nsA = misc.binomial(norb, ne[0])
nsB = misc.binomial(norb, ne[1])
nelec = ne[0] + ne[1]
nsorb = 2 * norb
print 'nsA,nsB=', nsA, nsB
civec = numpy.zeros((nsA, nsB))
for isA in range(nsA):
bsA = addr2str_o1(norb, ne[0], isA)
strA = bit2string(bsA, norb)
strA = [2 * i for i in strA]
for isB in range(nsB):
bsB = addr2str_o1(norb, ne[1], isB)
strB = bit2string(bsB, norb)
strB = [2 * i + 1 for i in strB]
# conversion
string = strA + strB
sgn = string_sgn(string)
string.sort()
addr = str2addr_o1(nsorb, nelec, string2bit(string))
civec[isA, isB] = civec2[addr] * sgn
#print civec2[numpy.argwhere(civec2>1.e-10)]
norm = numpy.linalg.norm(civec)
print "NORM1=", norm
#if(abs(norm-1.0)>1.e-6): exit(1)
return civec
def toCItensor2(nsorb, nelec, civec):
print '\n[toCItensor2]'
ndim = misc.binomial(nsorb, nelec)
print 'nsorb=', nsorb
print 'nelec=', nelec
print 'C_K^N=', ndim, civec.size
if ndim != civec.size:
print 'ndim='
print 'civec.size='
exit(1)
# phys=K-N+1
nphys = nsorb - nelec + 1
citensor = numpy.zeros(nphys**nelec)
print 'd_K^N=', citensor.shape
stride = [nphys**(nelec - i - 1) for i in range(nelec)]
offset = numpy.array(range(nelec))
for i in range(civec.size):
bstring = bin(addr2str_o1(nsorb, nelec, i))
orblst0 = numpy.array(bit2string(int(bstring, 2), nsorb))
orblst1 = orblst0 - offset
addr = numpy.dot(orblst1, stride)
#print orblst0,orblst1,addr
citensor[addr] = civec[i]
print 'NORM of Tensor=', numpy.linalg.norm(citensor)
return citensor
def toCItensor(nsorb, nelec, civec, iop=0):
print '\n[toCItensor]: iop=', iop
ndim = misc.binomial(nsorb, nelec)
print 'nsorb=', nsorb
print 'nelec=', nelec
print 'C_K^N=', ndim, civec.size
if ndim != civec.size:
print 'ndim='
print 'civec.size='
exit(1)
citensor = numpy.zeros(nsorb**nelec)
print 'd_K^N=', citensor.shape
stride = [nsorb**(nelec - i - 1) for i in range(nelec)]
if iop == 0:
for i in range(civec.size):
bstring = bin(addr2str_o1(nsorb, nelec, i))
orblst = tuple(bit2string(int(bstring, 2), nsorb))
addr = numpy.dot(orblst, stride)
citensor[addr] = civec[i]
else:
for i in range(civec.size):
bstring = bin(addr2str_o1(nsorb, nelec, i))
orblst = tuple(bit2string(int(bstring, 2), nsorb))
for j in itools.permutations(orblst, nelec):
addr = numpy.dot(j, stride)
citensor[addr] = string_sgn(j) * civec[i]
fac = 1.0 / math.sqrt(math.factorial(nelec))
citensor = citensor * fac
print 'NORM of Tensor=', numpy.linalg.norm(citensor)
return citensor
def toCIspinorb(norb, ne, civec):
print '\n[toCIspinorb]: spin-orbit rep C(Ia,Ib)->C(I)'
nsA, nsB = civec.shape
nelec = ne[0] + ne[1]
nsorb = 2 * norb
ndim = misc.binomial(nsorb, nelec)
print 'nsA,nsB=', nsA, nsB
print 'nstring=', ndim
print 'nfactor=', ndim * 1.0 / (nsA * nsB)
civec2 = numpy.zeros(ndim)
for isA in range(nsA):
bsA = addr2str_o1(norb, ne[0], isA)
strA = bit2string(bsA, norb)
strA = [2 * i for i in strA]
for isB in range(nsB):
bsB = addr2str_o1(norb, ne[1], isB)
strB = bit2string(bsB, norb)
strB = [2 * i + 1 for i in strB]
# conversion
string = strA + strB
sgn = string_sgn(string)
string.sort()
#print string2bit(string)
#print string
#print civec[isA,isB]
#print addr
#addr=fci.cistring.str2addr_o1(nsorb,nelec,string2bit(string))
addr = str2addr_o1(nsorb, nelec, string2bit(string))
#print nsorb,string,sgn,addr,bin(string2bit(string))
civec2[addr] = civec[isA, isB] * sgn
#print civec2[numpy.argwhere(civec2>1.e-10)]
norm = numpy.linalg.norm(civec2)
print "NORM1=", norm
if (abs(norm - 1.0) > 1.e-6): exit(1)
return civec2
def toONtensorReverse(nsorb, nelec, ontensor):
print '\n[toONtensorReverse]:'
ndim = misc.binomial(nsorb, nelec)
print 'nsorb=', nsorb
print 'nelec=', nelec
print 'C_K^N=', ndim
civec = numpy.zeros(ndim)
for i in range(civec.size):
bstring = bin(addr2str_o1(nsorb, nelec, i))
addr = int(bstring, 2)
civec[i] = ontensor[addr]
print 'NORM of Tensor=', numpy.linalg.norm(civec)
return civec
def toCItensorReverse(nsorb, nelec, citensor, iop=0):
#print '\n[toCItensorReverse]: iop=',iop
ndim = misc.binomial(nsorb, nelec)
#print 'nsorb=',nsorb
#print 'nelec=',nelec
#print 'C_K^N=',ndim
civec = numpy.zeros(ndim)
stride = [nsorb**(nelec - i - 1) for i in range(nelec)]
for i in range(civec.size):
bstring = bin(addr2str_o1(nsorb, nelec, i))
orblst = tuple(bit2string(int(bstring, 2), nsorb))
addr = numpy.dot(orblst, stride)
civec[i] = citensor[addr]
if iop == 1:
fac = math.sqrt(math.factorial(nelec))
civec = civec * fac
#print 'NORM of Tensor=',numpy.linalg.norm(civec)
return civec
def toONtensor(nsorb, nelec, civec):
print '\n[toONtensor]:'
ndim = misc.binomial(nsorb, nelec)
print 'nsorb=', nsorb
print 'nelec=', nelec
print 'C_K^N=', ndim, civec.size
if ndim != civec.size: exit(1)
ontensor = numpy.zeros(2**nsorb)
print 'd_2^K=', ontensor.shape
for i in range(civec.size):
bstring = bin(addr2str_o1(nsorb, nelec, i))
addr = int(bstring, 2)
print 'idx=', i, bstring, addr
# idx= 0 0b1 1
# idx= 1 0b10 2
# idx= 2 0b100 4
ontensor[addr] = civec[i]
print 'NORM of Tensor=', numpy.linalg.norm(ontensor)
return ontensor
def remove_redundant_reads(self):
''' remove additional redundant reads that are not warranted
by a binomial test.'''
def rm_max(list, max):
# remove reads that are more than the maximum specified.
list_new = []
pre = 0
count = 0
for x in list:
if x == pre:
count += 1
else:
pre = x
count = 1
if count > max:
continue
list_new.append(x)
return list_new
for filename in self.filename_list:
max = misc.binomial(
self.read_total_per_file[filename], 1.0/self.genome_size)
# calculate maximum duplicates for each genomic location.
info ("The maximum at one position for %s is %d", filename, max)
total_before = self.read_total_per_file[filename]
total_after = 0
for chr in self.chr_list:
for strand in ["f", "r"]:
reads = self.data_dict_by_strands[chr][filename][strand]
reads.sort()
reads = rm_max(reads, max)
total_after += len(reads)
self.data_dict_by_strands[chr][filename][strand] = reads
self.read_total_per_file[filename] = total_after
debug("Total # of reads before removing redundancies is %d",
total_before)
debug("Total # of reads after removing redundancies is %d",
total_after)
info("The percentage of redundant reads for %s is %f ",
filename, 1-float(total_after)/total_before)
def remove_redundant_reads(self):
''' remove additional redundant reads that are not warranted
by a binomial test.'''
def rm_max(list, max):
# remove reads that are more than the maximum specified.
list_new = []
pre = 0
count = 0
for x in list:
if x == pre:
count += 1
else:
pre = x
count = 1
if count > max:
continue
list_new.append(x)
return list_new
for filename in self.filename_list:
max = misc.binomial(self.read_total_per_file[filename],
1.0 / self.genome_size)
# calculate maximum duplicates for each genomic location.
info("The maximum at one position for %s is %d", filename, max)
total_before = self.read_total_per_file[filename]
total_after = 0
for chr in self.chr_list:
for strand in ["f", "r"]:
reads = self.data_dict_by_strands[chr][filename][strand]
reads.sort()
reads = rm_max(reads, max)
total_after += len(reads)
self.data_dict_by_strands[chr][filename][strand] = reads
self.read_total_per_file[filename] = total_after
debug("Total # of reads before removing redundancies is %d",
total_before)
debug("Total # of reads after removing redundancies is %d",
total_after)
info("The percentage of redundant reads for %s is %f ", filename,
1 - float(total_after) / total_before)
