def getVector(tags, llrThreshold, converge, pairs, p):
"""
Output: distribution over users after the LLR algorithm is converged.
"""
vectors = []
for tag in tags:
c = set([tag])
vectors.append(clustering.runClustering(pairs, p, c, llrThreshold, converge))
return vectors
def readAndSave(simulationfiles, datafile, mindist=1.5, maxdist=5.0, eps=1.6,
ms=4, scale_eps=False):
masses = []
H0s = []
inClusterH0s = []
outClusterH0s = []
zeropoints = []
inClusterZeros = []
outClusterZeros = []
allDispersions = []
clusterDispersions = []
unclusteredDispersions = []
radialVelocities = []
tangentialVelocities = []
LGdistances = []
timingArgumentMasses = []
f = open(simulationfiles, 'r')
for simdir in f.readlines():
dirname = simdir.strip()
staticVel = filereader.readAllFiles(dirname, "Subhalo/Velocity", 3)
mass = filereader.readAllFiles(dirname, "Subhalo/Mass", 1)
fullCOP = filereader.readAllFiles(dirname, "Subhalo/CentreOfPotential", 3)
FoFcontamination = filereader.readAllFiles(dirname, "FOF/ContaminationCount",
1)
groupNumbers = filereader.readAllFiles(dirname, "Subhalo/GroupNumber", 1)
fullCOP = fullCOP/h0 # to Mpc
# creation of mask with True if there is no contamination in Subhalo
contaminationMask = np.asarray([FoFcontamination[int(group)-1]<1 for group in
groupNumbers])
# save contaminated haloes for finding closest one
contaminatedPositions = fullCOP[contaminationMask==False, :]
# elimination of contaminated haloes
staticVel = staticVel[contaminationMask,:]
mass = mass[contaminationMask]
cop = fullCOP[contaminationMask, :]
groupNumbers = groupNumbers[contaminationMask]
# to physical units
mass = mass/h0*1e10 # to M_☉
LGs = LGfinder.findLocalGroup(staticVel, mass, cop, quiet=True,
outputAll=True)
unmaskedLGs = LGfinder.maskedToUnmasked(LGs, cop, fullCOP)
bestLGindex = LGfinder.chooseClosestToCentre(unmaskedLGs, contaminationMask, fullCOP)
LG = LGs[bestLGindex]
if mass[LG[0]] > mass[LG[1]]:
centreIndex = LG[1]
else:
centreIndex = LG[0]
massCentre = physUtils.massCentre(cop[LG[0]], cop[LG[1]], mass[LG[0]],
mass[LG[1]])
closestContDist = physUtils.findClosestDistance(massCentre,
contaminatedPositions)
if closestContDist < maxdist:
print("Warning: closest contaminated halo closer than the edge of\
Hubble flow fitting range in simulation " + dirname +
".\nExcluding simulation from analysis.")
continue
centre = cop[centreIndex]
centreVel = staticVel[centreIndex]
closestContDist = physUtils.findClosestDistance(centre,
contaminatedPositions)
distances = np.array([physUtils.distance(centre, c) for c in
cop])
vel = physUtils.addExpansion(staticVel, cop, centre)
LGrelVel = vel[LG[0]] - vel[LG[1]]
LGrelVelComponents = physUtils.velocityComponents(LGrelVel, cop[LG[1]]-cop[LG[0]])
LGdistance = physUtils.distance(cop[LG[0]], cop[LG[1]])
closestContMask = distances < closestContDist
distRangeMask = np.array([d < maxdist and d > mindist for d in
distances])
# contamination and distance range cut
mask = np.logical_and(closestContMask, distRangeMask)
cop = cop[mask]
vel = vel[mask]
distances = distances[mask]
# radial velocities
radvel = np.array([physUtils.velocityComponents(vel[j] - centreVel,
cop[j] - centre)[0]
for j in range(len(vel))])
##### extracting clustering data #####
clusteringDB = clustering.runClustering(cop, centre, ms, eps,
meansep=scale_eps)
labels = clusteringDB.labels_
uniqueLabels = set(labels)
clusterMemberMask = labels != -1 # True for haloes in cluster
## all haloes ##
(H0, zero) = simpleFit(distances, radvel)
radvelResiduals = np.empty(radvel.shape)
for i in range(len(radvel)):
radvelResiduals[i] = radvel[i] - (distances[i] - zero) * H0
zeropoints.append(zero)
H0s.append(H0)
allDispersions.append(np.std(radvelResiduals, ddof=1))
## outside clusters ##
(H0, zero, dispersion) = outClusterFit(clusteringDB, radvel, distances,
minHaloes=10)
outClusterH0s.append(H0)
outClusterZeros.append(zero)
unclusteredDispersions.append(dispersion)
## inside clusters ##
(H0, zero, dispersion) = inClusterFit(clusteringDB, radvel, distances,
minSize=10)
inClusterH0s.append(H0)
inClusterZeros.append(zero)
clusterDispersions.append(dispersion)
# LG mass from MW and andromeda
M_big2 = mass[LG[0]] + mass[LG[1]]
masses.append(M_big2)
radialVelocities.append(LGrelVelComponents[0])
tangentialVelocities.append(LGrelVelComponents[1])
LGdistances.append(LGdistance)
timingArgumentMass = timingargument.timingArgumentMass(-1 *
LGrelVelComponents[0],
LGdistance*1000.0,
13.815, G)
timingArgumentMasses.append(timingArgumentMass)
##### finalizing data #####
masses = np.array(masses)
H0s = np.array(H0s)
inClusterH0s = np.array(inClusterH0s)
outClusterH0s = np.array(outClusterH0s)
zeropoints = np.array(zeropoints)
inClusterZeros = np.array(inClusterZeros)
outClusterZeros = np.array(outClusterZeros)
allDispersions = np.array(allDispersions)
clusterDispersions = np.array(clusterDispersions)
unclusteredDispersions = np.array(unclusteredDispersions)
radialVelocities = np.array(radialVelocities)
tangentialVelocities = np.array(tangentialVelocities)
LGdistances = np.array(LGdistances)
timingArgumentMasses = np.array(timingArgumentMasses)
data = np.array([masses, timingArgumentMasses, H0s, inClusterH0s,
outClusterH0s, zeropoints, inClusterZeros, outClusterZeros,
allDispersions, unclusteredDispersions, clusterDispersions,
radialVelocities, tangentialVelocities, LGdistances]).T
np.savetxt(datafile, data)
return data
def readAndSave(simulationfiles,
datafile,
mindist=1.5,
maxdist=5.0,
eps=1.6,
ms=4,
scale_eps=False):
masses = []
H0s = []
inClusterH0s = []
outClusterH0s = []
zeropoints = []
inClusterZeros = []
outClusterZeros = []
allDispersions = []
clusterDispersions = []
unclusteredDispersions = []
radialVelocities = []
tangentialVelocities = []
LGdistances = []
timingArgumentMasses = []
f = open(simulationfiles, 'r')
for simdir in f.readlines():
dirname = simdir.strip()
staticVel = filereader.readAllFiles(dirname, "Subhalo/Velocity", 3)
mass = filereader.readAllFiles(dirname, "Subhalo/Mass", 1)
fullCOP = filereader.readAllFiles(dirname, "Subhalo/CentreOfPotential",
3)
FoFcontamination = filereader.readAllFiles(dirname,
"FOF/ContaminationCount", 1)
groupNumbers = filereader.readAllFiles(dirname, "Subhalo/GroupNumber",
1)
fullCOP = fullCOP / h0 # to Mpc
# creation of mask with True if there is no contamination in Subhalo
contaminationMask = np.asarray(
[FoFcontamination[int(group) - 1] < 1 for group in groupNumbers])
# save contaminated haloes for finding closest one
contaminatedPositions = fullCOP[contaminationMask == False, :]
# elimination of contaminated haloes
staticVel = staticVel[contaminationMask, :]
mass = mass[contaminationMask]
cop = fullCOP[contaminationMask, :]
groupNumbers = groupNumbers[contaminationMask]
# to physical units
mass = mass / h0 * 1e10 # to M_☉
LGs = LGfinder.findLocalGroup(staticVel,
mass,
cop,
quiet=True,
outputAll=True)
unmaskedLGs = LGfinder.maskedToUnmasked(LGs, cop, fullCOP)
bestLGindex = LGfinder.chooseClosestToCentre(unmaskedLGs,
contaminationMask,
fullCOP)
LG = LGs[bestLGindex]
if mass[LG[0]] > mass[LG[1]]:
centreIndex = LG[1]
else:
centreIndex = LG[0]
massCentre = physUtils.massCentre(cop[LG[0]], cop[LG[1]], mass[LG[0]],
mass[LG[1]])
closestContDist = physUtils.findClosestDistance(
massCentre, contaminatedPositions)
if closestContDist < maxdist:
print("Warning: closest contaminated halo closer than the edge of\
Hubble flow fitting range in simulation " + dirname +
".\nExcluding simulation from analysis.")
continue
centre = cop[centreIndex]
centreVel = staticVel[centreIndex]
closestContDist = physUtils.findClosestDistance(
centre, contaminatedPositions)
distances = np.array([physUtils.distance(centre, c) for c in cop])
vel = physUtils.addExpansion(staticVel, cop, centre)
LGrelVel = vel[LG[0]] - vel[LG[1]]
LGrelVelComponents = physUtils.velocityComponents(
LGrelVel, cop[LG[1]] - cop[LG[0]])
LGdistance = physUtils.distance(cop[LG[0]], cop[LG[1]])
closestContMask = distances < closestContDist
distRangeMask = np.array(
[d < maxdist and d > mindist for d in distances])
# contamination and distance range cut
mask = np.logical_and(closestContMask, distRangeMask)
cop = cop[mask]
vel = vel[mask]
distances = distances[mask]
# radial velocities
radvel = np.array([
physUtils.velocityComponents(vel[j] - centreVel,
cop[j] - centre)[0]
for j in range(len(vel))
])
##### extracting clustering data #####
clusteringDB = clustering.runClustering(cop,
centre,
ms,
eps,
meansep=scale_eps)
labels = clusteringDB.labels_
uniqueLabels = set(labels)
clusterMemberMask = labels != -1 # True for haloes in cluster
## all haloes ##
(H0, zero) = simpleFit(distances, radvel)
radvelResiduals = np.empty(radvel.shape)
for i in range(len(radvel)):
radvelResiduals[i] = radvel[i] - (distances[i] - zero) * H0
zeropoints.append(zero)
H0s.append(H0)
allDispersions.append(np.std(radvelResiduals, ddof=1))
## outside clusters ##
(H0, zero, dispersion) = outClusterFit(clusteringDB,
radvel,
distances,
minHaloes=10)
outClusterH0s.append(H0)
outClusterZeros.append(zero)
unclusteredDispersions.append(dispersion)
## inside clusters ##
(H0, zero, dispersion) = inClusterFit(clusteringDB,
radvel,
distances,
minSize=10)
inClusterH0s.append(H0)
inClusterZeros.append(zero)
clusterDispersions.append(dispersion)
# LG mass from MW and andromeda
M_big2 = mass[LG[0]] + mass[LG[1]]
masses.append(M_big2)
radialVelocities.append(LGrelVelComponents[0])
tangentialVelocities.append(LGrelVelComponents[1])
LGdistances.append(LGdistance)
timingArgumentMass = timingargument.timingArgumentMass(
-1 * LGrelVelComponents[0], LGdistance * 1000.0, 13.815, G)
timingArgumentMasses.append(timingArgumentMass)
##### finalizing data #####
masses = np.array(masses)
H0s = np.array(H0s)
inClusterH0s = np.array(inClusterH0s)
outClusterH0s = np.array(outClusterH0s)
zeropoints = np.array(zeropoints)
inClusterZeros = np.array(inClusterZeros)
outClusterZeros = np.array(outClusterZeros)
allDispersions = np.array(allDispersions)
clusterDispersions = np.array(clusterDispersions)
unclusteredDispersions = np.array(unclusteredDispersions)
radialVelocities = np.array(radialVelocities)
tangentialVelocities = np.array(tangentialVelocities)
LGdistances = np.array(LGdistances)
timingArgumentMasses = np.array(timingArgumentMasses)
data = np.array([
masses, timingArgumentMasses, H0s, inClusterH0s, outClusterH0s,
zeropoints, inClusterZeros, outClusterZeros, allDispersions,
unclusteredDispersions, clusterDispersions, radialVelocities,
tangentialVelocities, LGdistances
]).T
np.savetxt(datafile, data)
return data
cop = cop[mask]
vel = vel[mask]
distances = distances[mask]
mass = mass[mask]
# radial velocities
radvel = np.array([
physUtils.velocityComponents(vel[j] - centreVel,
cop[j] - centre)[0]
for j in range(len(vel))
])
##### extracting interesting data starts #####
clusteringDB = clustering.runClustering(cop,
centre,
ms,
eps,
meansep=False)
labels = clusteringDB.labels_
uniqueLabels = set(labels)
clusterMemberMask = labels != -1 # True for in cluster
# print(sum([not membership for membership in clusterMemberMask]))
# (inClusterH0, inClusterZero) = simpleFit(distances[clusterMemberMask],
# radvel[clusterMemberMask])
# (outClusterH0, outClusterZero) = simpleFit(
# distances[clusterMemberMask == False],
# radvel[clusterMemberMask == False])
# all haloes
(H0, zero) = simpleFit(distances, radvel)
def llrClustering(initialSet, llrThreshold, converge, pairs, p):
return clustering.runClustering(pairs, p, initialSet, llrThreshold, converge)
mask = np.array([d < maxdist and d > mindist for d in distances]) cop = cop[mask] vel = vel[mask] distances = distances[mask] mass = mass[mask] # radial velocities radvel = np.array([physUtils.velocityComponents(vel[j] - centreVel, cop[j] - centre)[0] for j in range(len(vel))]) ##### extracting interesting data starts ##### clusteringDB = clustering.runClustering(cop, centre, ms, eps, meansep=False) labels = clusteringDB.labels_ uniqueLabels = set(labels) clusterMemberMask = labels != -1 # True for in cluster # print(sum([not membership for membership in clusterMemberMask])) # (inClusterH0, inClusterZero) = simpleFit(distances[clusterMemberMask], # radvel[clusterMemberMask]) # (outClusterH0, outClusterZero) = simpleFit( # distances[clusterMemberMask == False], # radvel[clusterMemberMask == False]) # all haloes (H0, zero) = simpleFit(distances, radvel) radvelResiduals = np.empty(radvel.shape) for i in range(len(radvel)):
