def make_tree_from_parent_vec(aux, Ks, N):
Nx = N
NN = np.bincount(aux.Ks, minlength=Nx)[1:]
Fathers = np.unique(Ks[np.logical_and(Ks != np.arange(1, Ks.size + 1), Ks > 0)])
Related = np.array([np.where(Fathers[i] == Ks)[0] for i in range(Fathers.size)])
if Fathers.size != 0:
RelStarts, RelEnds, RelCN, RelVec = cell.cell_2_vec(Related)
else:
RelStarts, RelEnds, RelCN, RelVec = [], [], [], []
CallForFathers = np.add(np.where(is_member(aux.Ks, Fathers)), 1)
RFathers = np.zeros((np.max(Fathers),))
for i in range(Fathers.size):
RFathers[Fathers[i] - 1] = i + 1
SegStartI = np.array([2] + np.add(CallForFathers, 1).tolist()[0])
SegEndI = np.array(np.subtract(CallForFathers, 1).tolist()[0] + [Nx - 1])
Level = np.multiply(SegStartI, 0)
Level[np.where(is_member(np.subtract(SegStartI, 1), Fathers))] = 1
for i in range(1, 51):
FF = np.where(Level == i)
ToWhoTheyCall = aux.Ks[SegEndI[FF]]
Level[np.where(is_member(np.subtract(SegStartI, 1), ToWhoTheyCall))] = i + 1
FLevel = np.add(np.multiply(Fathers, 0), 1)
for i in range(1, np.max(Level) + 1):
FF = np.where(Level == i)
ToWhoTheyCall = aux.Ks[SegEndI[FF]]
ToWhichFatherDoTheyCall = RFathers[np.subtract(ToWhoTheyCall[np.subtract(np.where(ToWhoTheyCall <= np.max(Fathers)), 1)], 1)][0]
FLevel[np.subtract(ToWhichFatherDoTheyCall.astype(int), 1)] = i + 1
SegEndI[-1] = N
Depth = max(Level)
ret = {}
ret['FLevel'] = FLevel
ret['Level'] = Level
ret['Depth'] = Depth
ret['SegStartI'] = SegStartI
ret['SegEndI'] = SegEndI
ret['Fathers'] = Fathers
ret['RelVec'] = RelVec
ret['RelStarts'] = RelStarts
ret['RelEnds'] = RelEnds
ret['CallForFathers'] = CallForFathers[0]
return ret
def make_tree_from_parent_vec(N, aux):
def is_member(A, B):
return [np.sum(a == B) for a in A]
Nx = N
NN = np.bincount(aux.Ks, minlength=Nx)
Fathers = []
for i in Ks:
if i not in range(1, Ks.size + 1) and i > 0:
Fathers.append(i)
Fathers = np.array(Fathers)
Fathers = np.unique(Fathers)
Related = []
for i in Fathers:
Related.append(np.where(Ks == i))
if Fathers.size != 0:
RelStarts, RelEnds, RelCN, RelVec = cell.cell_2_vec(Related)
else:
RelStarts, RelEnds, RelCN, RelVec = [], [], [], []
CallForFathers = np.where(is_member(aux.Ks, Fathers))
RFathers = [[] for i in range(np.max(Fathers))]
for i in Fathers:
RFathers[i] = [i for i in range(Fathers.size)]
RFathers = np.array(RFathers)
SegStartI = np.hstack((np.array(2), CallForFathers + 1))
SegEndI = np.hstack((CallForFathers - 1, np.array(Nx - 1)))
Level = SegStartI * 0
Level[is_member(SegStart - 1, Fathers)] = 1
for i in range(50):
FF = np.where(Level == i)
ToWhoTheyCall = Aux.Ks[SegEndI[FF] + 1]
Level[np.where(is_member(SegStartI - 1, ToWhoTheyCall))] = i + 1
FLevel = Fathers * 0 + 1
for i in range(np.max(Level)):
FF = np.where(Level == i + 1)
ToWhoTheyCall = aux.Ks[SegEndI[FF] + 1]
ToWhichFatherDoTheyCall = RFathers[ToWhoTheyCall[np.where(
ToWhoTheyCall <= np.max(Fathers))]]
FLevel[ToWhichFatherDoTheyCall] = i + 1
SegEndI[-1] = N
Depth = np.max(Level)
# TODO, return all relevant values
def make_tree_from_parent_vec(N, aux): def is_member(A, B): return [np.sum(a == B) for a in A] Nx = N NN = np.bincount(aux.Ks, minlength=Nx) Fathers = [] for i in Ks: if i not in range(1, Ks.size + 1) and i > 0: Fathers.append(i) Fathers = np.array(Fathers) Fathers = np.unique(Fathers) Related = [] for i in Fathers: Related.append(np.where(Ks == i)) if Fathers.size != 0: RelStarts, RelEnds, RelCN, RelVec = cell.cell_2_vec(Related) else: RelStarts, RelEnds, RelCN, RelVec = [], [], [], [] CallForFathers = np.where(is_member(aux.Ks, Fathers)) RFathers = [ [] for i in range(np.max(Fathers)) ] for i in Fathers: RFathers[i] = [i for i in range(Fathers.size)] RFathers = np.array(RFathers) SegStartI = np.hstack((np.array(2), CallForFathers + 1)) SegEndI = np.hstack((CallForFathers - 1, np.array(Nx - 1))) Level = SegStartI * 0 Level[is_member(SegStart - 1, Fathers)] = 1 for i in range(50): FF = np.where(Level == i) ToWhoTheyCall = Aux.Ks[SegEndI[FF] + 1] Level[np.where(is_member(SegStartI - 1, ToWhoTheyCall))] = i + 1 FLevel = Fathers * 0 + 1 for i in range(np.max(Level)): FF = np.where(Level == i + 1) ToWhoTheyCall = aux.Ks[SegEndI[FF] + 1] ToWhichFatherDoTheyCall = RFathers[ToWhoTheyCall[np.where(ToWhoTheyCall <= np.max(Fathers))]] FLevel[ToWhichFatherDoTheyCall] = i + 1 SegEndI[-1] = N Depth = np.max(Level) # TODO, return all relevant values
def create_auxilliary_data_3(A, N, NSeg, Parent, cmVec,parent_seg,bool_model,seg_start,n_segs,seg_to_comp):
bool_model = np.array(bool_model)
FTYPESTR = 'float'
print 'createAuxData3'
FatherBase = [0 for i in range(N - 1)]
for i in range(N - 1, 0, -1):
if A[i - 1, i] != 0:
k = i + 1
else:
k = np.where(A[i:,i - 1] != 0)[0] + i + 1
k = k[0]
FatherBase[i - 1] = k
FatherBase = np.array(FatherBase)
print FatherBase
d = np.diag(A).T
e, f = [0 for i in range(N)], [0 for i in range(N)]
for i in range(1, N):
f[i - 1] = A[i - 1, FatherBase[i - 1] - 1]
e[i] = A[FatherBase[i - 1] - 1, i - 1]
f[-1] = 0
f = np.array(f)
e = np.array(e)
Ksx = np.array(parent_seg)
Ks = [0]
for i in range(2, Ksx.size + 1):
Ks.append(N + 1 - Ksx[N + 2 - i - 1])
Ks = np.array(Ks)
aux = Aux()
aux.Ks = Ks.astype(np.int)
FatherBase = Ks[1:]
Father = np.append(FatherBase, [FatherBase.size + 2, FatherBase.size + 2])
FIdxsX = []
for i in range(1, int(np.ceil(np.log2(N)) + 3 + 1)):
CurF = np.array(list(range(1, Father.size + 1)))
for j in range(1, 2 ** (i - 1) + 1):
CurF = Father[np.subtract(CurF, 1)].astype(np.int)
FIdxsX.append(CurF)
FIdxsX = np.array(FIdxsX)
ind = np.where(np.all(FIdxsX == FIdxsX[-1], 1))[0][0] + 1
if ind != 0:
FIdxsX = FIdxsX[:ind - 1,:]
LognDepth = FIdxsX.shape[0]
FIdxsX = FIdxsX[:,:N]
aux.FIdxsX = FIdxsX
aux.LognDepth = LognDepth
Nx = N
SonNoVec, ParentUsed = np.zeros(Nx), np.zeros(Nx)
for seg in range(1, Nx + 1):
if seg == 1:
parentIndex = 1
else:
parentIndex = Nx + 1 - aux.Ks[Nx + 2 - seg - 1]
ParentUsed[parentIndex - 1] = ParentUsed[parentIndex - 1] + 1
SonNoVec[seg - 1] = ParentUsed[parentIndex - 1]
SonNoVec[0] = 0
aux.SonNoVec = SonNoVec
if np.max(SonNoVec) > 2:
raise ValueError('error np.max(SonNoVec) > 2')
tree_dict = make_tree_from_parent_vec(aux, Ks, N)
Depth = tree_dict['Depth']
Level = tree_dict['Level']
FLevel = tree_dict['FLevel']
SegStartI = tree_dict['SegStartI']
SegEndI = tree_dict['SegEndI']
Fathers = tree_dict['Fathers']
aux.Depth = Depth
aux.Level = Level
aux.FLevel = FLevel
aux.SegStartI = SegStartI
aux.SegEndI = SegEndI
aux.Fathers = Fathers
RelVec = tree_dict['RelVec']
RelStarts = tree_dict['RelStarts']
RelEnds = tree_dict['RelEnds']
aux.RelVec = RelVec
aux.RelStarts = RelStarts
aux.RelEnds = RelEnds
LastLevelsI = np.where(Level == np.max(Level))[0][0] + 1
EndLastLevelsI = SegEndI[LastLevelsI - 1]
KsB = Ks
KsB = np.append(KsB, [EndLastLevelsI])
aux.KsB = KsB
FN = '../../Data/BasicConst' + str(N) + 'Seg.mat'
FNP = '../../Data/BasicConst' + str(N) + 'SegP.mat'
FNM = '../../Data/ParamsMat' + str(N) + '.mat'
aux.d = d
aux.e = e
aux.f = f
aux.Cms = cmVec
FN_dict = {}
FN_dict['N'] = np.array([np.uint16(N)])
FN_dict['e'] = np.double(e)
FN_dict['f'] = np.double(f)
FN_dict['Ks'] = np.uint16(Ks)
FN_dict['auxCms'] = np.double(aux.Cms);
FN_dict['nrnHasHH'] = np.uint16(bool_model)
sio.savemat(FN, FN_dict)
CompByLevel32 = np.zeros((0, 32))
CompByFLevel32 = np.zeros((0, 32))
nFComps, nComps = np.array([]), np.array([])
LRelated, FLRelated = [], []
nRoundForThisLevel = np.array([])
for CurLevel in range(Depth + 1):
CurComps = np.add(np.where(Level == CurLevel)[0], 1)
nComps = np.append(nComps, [CurComps.size])
Longer = np.multiply(np.ones(int(np.ceil(CurComps.size / 32.0) * 32)), CurComps[-1])
Longer[:CurComps.size] = CurComps
StuffToAdd = Longer.reshape((Longer.size / 32, 32))
StartPoint = CompByLevel32.shape[0] + 1
CompByLevel32 = np.vstack((CompByLevel32, StuffToAdd))
EndPoint = CompByLevel32.shape[0]
LRelated.append(list(range(StartPoint, EndPoint + 1)))
nRoundForThisLevel = np.append(nRoundForThisLevel, [CompByLevel32.shape[0]])
if CurLevel < Depth:
CurComps = np.add(np.where(FLevel == CurLevel + 1)[0], 1)
nFComps = np.append(nFComps, [CurComps.size])
Longer = np.multiply(np.ones(int(np.ceil(CurComps.size / 32.0) * 32)), CurComps[-1])
Longer[:CurComps.size] = CurComps
StuffToAdd = Longer.reshape((Longer.size / 32, 32))
StartPoint = CompByFLevel32.shape[0] + 1
CompByFLevel32 = np.vstack((CompByFLevel32, StuffToAdd))
EndPoint = CompByFLevel32.shape[0]
FLRelated.append(list(range(StartPoint, EndPoint + 1)))
LRelated = np.array(LRelated)
FLRelated = np.array(FLRelated).astype(object)
LRelStarts, LRelEnds, LRelCN, LRelVec = cell.cell_2_vec(LRelated)
LRelStarts = np.add(LRelStarts, 1)
LRelEnds = np.add(LRelEnds, 1)
if Depth == 0:
FLRelStarts, FLRelEnds, FLRelCN, FLRelVec = [], [], [], []
else:
FLRelStarts, FLRelEnds, FLRelCN, FLRelVec = cell.cell_2_vec(FLRelated)
FLRelStarts = np.add(FLRelStarts, 1)
FLRelEnds = np.add(FLRelEnds, 1)
NComps = NSeg.size
CompsMid = np.zeros((NComps, 1))
for i in range(1, NComps + 1):
CompsMid[i - 1] = seg_start[i - 1] + np.floor(n_segs[i - 1] / 2)
aux.LRelStarts = LRelStarts
aux.LRelEnds = LRelEnds
aux.FLRelStarts = FLRelStarts
aux.FLRelEnds = FLRelEnds
CompDepth32 = CompByLevel32.shape[0]
CompFDepth32 = CompByFLevel32.shape[0]
CompsMid[0] += 1
Nd32 = np.ceil(N / 32.0)
N32 = Nd32 * 32
nFathers = Fathers.size
CallForFathers = tree_dict['CallForFathers']
nCallForFather = CallForFathers.size
aux.nCallForFather = nCallForFather
FNP_dict = {}
FNP_dict['N'] = np.array(np.uint16([N]))
FNP_dict['NComps'] = np.array(np.uint16([NComps]))
FNP_dict['e'] = np.double(e)
FNP_dict['f'] = np.double(f)
FNP_dict['Ks'] = np.uint16(Ks)
FNP_dict['SegToComp'] = np.uint16(np.subtract(seg_to_comp, 1))
FNP_dict['cmVec'] = np.double(cmVec)
FNP_dict['nrnHasHHSize'] = np.array(np.uint16([bool_model.shape[0]]))
FNP_dict['nrnHasHHT'] = bool_model.T
FNP_dict['SonNoVec'] = np.uint16(SonNoVec)
FNP_dict['Depth'] = np.array(np.uint16([Depth]))
FNP_dict['LognDepth'] = np.array(np.uint16([LognDepth]))
FNP_dict['nFathers'] = np.array(np.uint16([nFathers]))
FNP_dict['nCallForFather'] = np.array(np.uint16([nCallForFather]))
FNP_dict['RelStarts'] = np.uint16(RelStarts)
FNP_dict['RelEnds'] = np.uint16(RelEnds)
FNP_dict['RelVec'] = np.uint16(RelVec)
FNP_dict['SegStartI'] = np.uint16(SegStartI)
FNP_dict['SegEndI'] = np.uint16(SegEndI)
FNP_dict['auxFathers'] = np.uint16(aux.Fathers)
FNP_dict['FIdxsXT'] = np.uint16(FIdxsX.T)
FNP_dict['CompDepth32'] = np.uint16(np.array([CompDepth32]))
FNP_dict['CompFDepth32'] = np.uint16(np.array([CompFDepth32]))
FNP_dict['CompByLevel32T'] = np.uint16(CompByLevel32.T)
FNP_dict['CompByFLevel32T'] = np.uint16(CompByFLevel32.T)
aux.CompDepth32 = CompDepth32
aux.CompFDepth32 = CompFDepth32
FNP_dict['auxLRelStartsSize'] = np.uint16(np.array([aux.RelStarts.size]))
FNP_dict['auxLRelStarts_1'] = np.uint16(np.subtract(aux.LRelStarts, 1))
FNP_dict['auxLRelEnds_1'] = np.uint16(np.subtract(aux.LRelEnds, 1))
FNP_dict['auxFLRelStartsSize'] = np.uint16(np.array([aux.FLRelStarts.size]))
FNP_dict['auxFLRelStarts_1'] = np.uint16(np.subtract(aux.FLRelStarts, 1))
FNP_dict['auxFLRelEnds_1'] = np.uint16(np.subtract(aux.FLRelEnds, 1))
FNP_dict['auxKsB_1'] = np.uint16(np.subtract(aux.KsB, 1))
sio.savemat(FNP, FNP_dict)
FNP_dict['auxKsB_1'] = np.uint16(np.subtract(aux.KsB, 1))
return FN, FNP, FNM, aux
def create_auxilliary_data_3(A, N, NSeg, Parent, cmVec, parent_seg, bool_model,
seg_start, n_segs, seg_to_comp):
bool_model = np.array(bool_model)
FTYPESTR = 'float'
print 'createAuxData3'
FatherBase = [0 for i in range(N - 1)]
for i in range(N - 1, 0, -1):
if A[i - 1, i] != 0:
k = i + 1
else:
k = np.where(A[i:, i - 1] != 0)[0] + i + 1
k = k[0]
FatherBase[i - 1] = k
FatherBase = np.array(FatherBase)
print FatherBase
d = np.diag(A).T
e, f = [0 for i in range(N)], [0 for i in range(N)]
for i in range(1, N):
f[i - 1] = A[i - 1, FatherBase[i - 1] - 1]
e[i] = A[FatherBase[i - 1] - 1, i - 1]
f[-1] = 0
f = np.array(f)
e = np.array(e)
Ksx = np.array(parent_seg)
Ks = [0]
for i in range(2, Ksx.size + 1):
Ks.append(N + 1 - Ksx[N + 2 - i - 1])
Ks = np.array(Ks)
aux = Aux()
aux.Ks = Ks.astype(np.int)
FatherBase = Ks[1:]
Father = np.append(FatherBase, [FatherBase.size + 2, FatherBase.size + 2])
FIdxsX = []
for i in range(1, int(np.ceil(np.log2(N)) + 3 + 1)):
CurF = np.array(list(range(1, Father.size + 1)))
for j in range(1, 2**(i - 1) + 1):
CurF = Father[np.subtract(CurF, 1)].astype(np.int)
FIdxsX.append(CurF)
FIdxsX = np.array(FIdxsX)
ind = np.where(np.all(FIdxsX == FIdxsX[-1], 1))[0][0] + 1
if ind != 0:
FIdxsX = FIdxsX[:ind - 1, :]
LognDepth = FIdxsX.shape[0]
FIdxsX = FIdxsX[:, :N]
aux.FIdxsX = FIdxsX
aux.LognDepth = LognDepth
Nx = N
SonNoVec, ParentUsed = np.zeros(Nx), np.zeros(Nx)
for seg in range(1, Nx + 1):
if seg == 1:
parentIndex = 1
else:
parentIndex = Nx + 1 - aux.Ks[Nx + 2 - seg - 1]
ParentUsed[parentIndex - 1] = ParentUsed[parentIndex - 1] + 1
SonNoVec[seg - 1] = ParentUsed[parentIndex - 1]
SonNoVec[0] = 0
aux.SonNoVec = SonNoVec
if np.max(SonNoVec) > 2:
raise ValueError('error np.max(SonNoVec) > 2')
tree_dict = make_tree_from_parent_vec(aux, Ks, N)
Depth = tree_dict['Depth']
Level = tree_dict['Level']
FLevel = tree_dict['FLevel']
SegStartI = tree_dict['SegStartI']
SegEndI = tree_dict['SegEndI']
Fathers = tree_dict['Fathers']
aux.Depth = Depth
aux.Level = Level
aux.FLevel = FLevel
aux.SegStartI = SegStartI
aux.SegEndI = SegEndI
aux.Fathers = Fathers
RelVec = tree_dict['RelVec']
RelStarts = tree_dict['RelStarts']
RelEnds = tree_dict['RelEnds']
aux.RelVec = RelVec
aux.RelStarts = RelStarts
aux.RelEnds = RelEnds
LastLevelsI = np.where(Level == np.max(Level))[0][0] + 1
EndLastLevelsI = SegEndI[LastLevelsI - 1]
KsB = Ks
KsB = np.append(KsB, [EndLastLevelsI])
aux.KsB = KsB
FN = '../../Data/BasicConst' + str(N) + 'Seg.mat'
FNP = '../../Data/BasicConst' + str(N) + 'SegP.mat'
FNM = '../../Data/ParamsMat' + str(N) + '.mat'
aux.d = d
aux.e = e
aux.f = f
aux.Cms = cmVec
FN_dict = {}
FN_dict['N'] = np.array([np.uint16(N)])
FN_dict['e'] = np.double(e)
FN_dict['f'] = np.double(f)
FN_dict['Ks'] = np.uint16(Ks)
FN_dict['auxCms'] = np.double(aux.Cms)
FN_dict['nrnHasHH'] = np.uint16(bool_model)
sio.savemat(FN, FN_dict)
CompByLevel32 = np.zeros((0, 32))
CompByFLevel32 = np.zeros((0, 32))
nFComps, nComps = np.array([]), np.array([])
LRelated, FLRelated = [], []
nRoundForThisLevel = np.array([])
for CurLevel in range(Depth + 1):
CurComps = np.add(np.where(Level == CurLevel)[0], 1)
nComps = np.append(nComps, [CurComps.size])
Longer = np.multiply(np.ones(int(np.ceil(CurComps.size / 32.0) * 32)),
CurComps[-1])
Longer[:CurComps.size] = CurComps
StuffToAdd = Longer.reshape((Longer.size / 32, 32))
StartPoint = CompByLevel32.shape[0] + 1
CompByLevel32 = np.vstack((CompByLevel32, StuffToAdd))
EndPoint = CompByLevel32.shape[0]
LRelated.append(list(range(StartPoint, EndPoint + 1)))
nRoundForThisLevel = np.append(nRoundForThisLevel,
[CompByLevel32.shape[0]])
if CurLevel < Depth:
CurComps = np.add(np.where(FLevel == CurLevel + 1)[0], 1)
nFComps = np.append(nFComps, [CurComps.size])
Longer = np.multiply(
np.ones(int(np.ceil(CurComps.size / 32.0) * 32)), CurComps[-1])
Longer[:CurComps.size] = CurComps
StuffToAdd = Longer.reshape((Longer.size / 32, 32))
StartPoint = CompByFLevel32.shape[0] + 1
CompByFLevel32 = np.vstack((CompByFLevel32, StuffToAdd))
EndPoint = CompByFLevel32.shape[0]
FLRelated.append(list(range(StartPoint, EndPoint + 1)))
LRelated = np.array(LRelated)
FLRelated = np.array(FLRelated).astype(object)
LRelStarts, LRelEnds, LRelCN, LRelVec = cell.cell_2_vec(LRelated)
LRelStarts = np.add(LRelStarts, 1)
LRelEnds = np.add(LRelEnds, 1)
if Depth == 0:
FLRelStarts, FLRelEnds, FLRelCN, FLRelVec = [], [], [], []
else:
FLRelStarts, FLRelEnds, FLRelCN, FLRelVec = cell.cell_2_vec(FLRelated)
FLRelStarts = np.add(FLRelStarts, 1)
FLRelEnds = np.add(FLRelEnds, 1)
NComps = NSeg.size
CompsMid = np.zeros((NComps, 1))
for i in range(1, NComps + 1):
CompsMid[i - 1] = seg_start[i - 1] + np.floor(n_segs[i - 1] / 2)
aux.LRelStarts = LRelStarts
aux.LRelEnds = LRelEnds
aux.FLRelStarts = FLRelStarts
aux.FLRelEnds = FLRelEnds
CompDepth32 = CompByLevel32.shape[0]
CompFDepth32 = CompByFLevel32.shape[0]
CompsMid[0] += 1
Nd32 = np.ceil(N / 32.0)
N32 = Nd32 * 32
nFathers = Fathers.size
CallForFathers = tree_dict['CallForFathers']
nCallForFather = CallForFathers.size
aux.nCallForFather = nCallForFather
FNP_dict = {}
FNP_dict['N'] = np.array(np.uint16([N]))
FNP_dict['NComps'] = np.array(np.uint16([NComps]))
FNP_dict['e'] = np.double(e)
FNP_dict['f'] = np.double(f)
FNP_dict['Ks'] = np.uint16(Ks)
FNP_dict['SegToComp'] = np.uint16(np.subtract(seg_to_comp, 1))
FNP_dict['cmVec'] = np.double(cmVec)
FNP_dict['nrnHasHHSize'] = np.array(np.uint16([bool_model.shape[0]]))
FNP_dict['nrnHasHHT'] = bool_model.T
FNP_dict['SonNoVec'] = np.uint16(SonNoVec)
FNP_dict['Depth'] = np.array(np.uint16([Depth]))
FNP_dict['LognDepth'] = np.array(np.uint16([LognDepth]))
FNP_dict['nFathers'] = np.array(np.uint16([nFathers]))
FNP_dict['nCallForFather'] = np.array(np.uint16([nCallForFather]))
FNP_dict['RelStarts'] = np.uint16(RelStarts)
FNP_dict['RelEnds'] = np.uint16(RelEnds)
FNP_dict['RelVec'] = np.uint16(RelVec)
FNP_dict['SegStartI'] = np.uint16(SegStartI)
FNP_dict['SegEndI'] = np.uint16(SegEndI)
FNP_dict['auxFathers'] = np.uint16(aux.Fathers)
FNP_dict['FIdxsXT'] = np.uint16(FIdxsX.T)
FNP_dict['CompDepth32'] = np.uint16(np.array([CompDepth32]))
FNP_dict['CompFDepth32'] = np.uint16(np.array([CompFDepth32]))
FNP_dict['CompByLevel32T'] = np.uint16(CompByLevel32.T)
FNP_dict['CompByFLevel32T'] = np.uint16(CompByFLevel32.T)
aux.CompDepth32 = CompDepth32
aux.CompFDepth32 = CompFDepth32
FNP_dict['auxLRelStartsSize'] = np.uint16(np.array([aux.RelStarts.size]))
FNP_dict['auxLRelStarts_1'] = np.uint16(np.subtract(aux.LRelStarts, 1))
FNP_dict['auxLRelEnds_1'] = np.uint16(np.subtract(aux.LRelEnds, 1))
FNP_dict['auxFLRelStartsSize'] = np.uint16(np.array([aux.FLRelStarts.size
]))
FNP_dict['auxFLRelStarts_1'] = np.uint16(np.subtract(aux.FLRelStarts, 1))
FNP_dict['auxFLRelEnds_1'] = np.uint16(np.subtract(aux.FLRelEnds, 1))
FNP_dict['auxKsB_1'] = np.uint16(np.subtract(aux.KsB, 1))
sio.savemat(FNP, FNP_dict)
FNP_dict['auxKsB_1'] = np.uint16(np.subtract(aux.KsB, 1))
return FN, FNP, FNM, aux
