def submodel_vp_colored(m,sm,rotated=None):
""" m is the full model with the VP's already calculated
sm is the submodel that we will save. it is modified and
likely otherwise unusable - be careful """
if( rotated == None ):
vpcoloredmodel = Model('vp colored atoms',m.lx,m.ly,m.lz,sm.atoms)
else:
vpcoloredmodel = Model('vp colored atoms',m.lx,m.ly,m.lz,rotated.atoms)
for i,atom in enumerate(vpcoloredmodel.atoms):
atomi = m.atoms[m.atoms.index(sm.atoms[i])]
if(atomi.vp.type == 'Full-icosahedra'):
sm.atoms[i].vp = atomi.vp.copy()
if( rotated != None ): rotated.atoms[i].vp = atomi.vp.copy()
atom.z = 1 # Dark blue
elif(atomi.vp.type == 'Icosahedra-like'):
sm.atoms[i].vp = atomi.vp.copy()
if( rotated != None ): rotated.atoms[i].vp = atomi.vp.copy()
atom.z = 2 # Light blue
elif(atomi.vp.type == 'Mixed'):
sm.atoms[i].vp = atomi.vp.copy()
if( rotated != None ): rotated.atoms[i].vp = atomi.vp.copy()
atom.z = 3 # Orange
elif(atomi.vp.type == 'Crystal-like'):
sm.atoms[i].vp = atomi.vp.copy()
if( rotated != None ): rotated.atoms[i].vp = atomi.vp.copy()
atom.z = 4 # Red
elif(atomi.vp.type == 'Undef'):
sm.atoms[i].vp = atomi.vp.copy()
if( rotated != None ): rotated.atoms[i].vp = atomi.vp.copy()
atom.z = 5 # Grey
vpcoloredmodel.write_real_xyz('vpcolored.xyz')
class Autoencoder:
def __init__(self):
self.config = Config()
self.model = Model(self.config)
self.model_dir = 'Psych209_AE'
self.writer = tf.summary.FileWriter(self.model_dir)
self.saver = tf.train.Saver(max_to_keep=200)
self.global_step = 0
self.sess = tf.Session()
self.num_epochs = 100
self.save_interval = 1000
self.print_interval = 100
def run(self):
train, test = utils.load_data('Data/movie_lines.txt')
embeddings = utils.load_embeddings()
self.sess.run(tf.global_variables_initializer())
self.writer.add_graph(sess.graph)
merged_summaries = tf.summary.merge_all()
print 'Starting training...'
for epoch in range(self.num_epochs):
print '-----Epoch', epoch, '-----'
batches = utils.get_batches(train, self.config.batch_size)
start_time = datetime.datetime.now()
for batch in tqdm(batches):
ops, feed = self.model.train_step(batch, training=True)
_, loss, summary = sess.run(ops + (merged_summaries, ), feed)
self.writer.add_summary(summary, self.global_step)
self.global_step += 1
# training status
if self.global_step % self.print_interval == 0:
perplexity = math.exp(
float(loss)) if loss < 300 else float('inf')
tqdm.write(
"----- Step %d -- Loss %.2f -- Perplexity %.2f" %
(self.global_step, loss, perplexity))
# run test periodically
ops, feed = self.model.train_step(batch, training=False)
_, loss, summary = sess.run(ops + (merged_summaries, ),
feed)
# save checkpoint
if self.global_step % self.save_interval == 0:
self.save_session(sess)
end_time = datetime.datetime.now()
print 'Epoch finish in ', end_time - start_time, 'ms'
def save_session(self, sess):
print 'Saving session at checkpoint', self.global_step
name = self.model_dir + str(self.global_step)
self.saver.save(sess, name)
print 'Save complete with name', name
def __init__(self):
self.config = Config()
self.model = Model(self.config)
self.model_dir = 'Psych209_AE'
self.writer = tf.summary.FileWriter(self.model_dir)
self.saver = tf.train.Saver(max_to_keep=200)
self.global_step = 0
self.sess = tf.Session()
self.num_epochs = 100
self.save_interval = 1000
self.print_interval = 100
def main():
""" Columns that I should have:
% atoms that are xtal-like (i.e. what % of all the xtal-like atoms are in this sub-model?)
% atoms that are ico-like
% atoms that are mixed
% atoms that are undefined
Can you see planes?
Is the spot visible in the sub-model's FT?
What is the ratio of the max intensity of the spot in the original FT / that in the sub-model's FT? (Do this on a per-atom basis to account for the different number of atoms.)"""
# paramfile, jobid, original ft, main ft direc, VP categories paramfile
# It is necessary to have a "spot_ft" directory and a "submodels" directory
# in the main ft directory for this to work.
paramfile = sys.argv[1] # Paramfile
with open(paramfile) as f:
params = f.readlines()
params = [line.strip() for line in params]
modelfile = params[0]
num_spots = int(params[1])
jobid = sys.argv[2] # jobid
orig_ft = sys.argv[3] # original ft
main_direc = sys.argv[4] # spot ft direc
spot_ft_direc = main_direc+'/spot_fts/'
spot_fts = sorted(os.listdir(spot_ft_direc))
spot_fts = [spot_ft_direc+line for line in spot_fts]
submodel_direc = main_direc+'/submodels/'
vp_paramfile = sys.argv[5] # VP paramfile for categorizing
# Do the VP analysis first, it's faster.
m = Model(modelfile)
vp_dict = load_param_file(vp_paramfile)
vp_dict['Undef'] = []
cutoff = {}
cutoff[(40,40)] = 3.6
cutoff[(13,29)] = 3.6
cutoff[(29,13)] = 3.6
cutoff[(40,13)] = 3.6
cutoff[(13,40)] = 3.6
cutoff[(29,40)] = 3.6
cutoff[(40,29)] = 3.6
cutoff[(13,13)] = 3.6
cutoff[(29,29)] = 3.6
voronoi_3d(m,cutoff)
set_atom_vp_types(m,vp_dict)
vp_models = []
for i,vptype in enumerate(vp_dict):
vp_models.append( Model('{0} atoms'.format(vptype),m.lx,m.ly,m.lz, [atom for atom in m.atoms if atom.vp.type == vptype]) )
vp_models[i].write_real_xyz('{0}.real.xyz'.format(vptype))
vpcoloredmodel = Model('vp colored atoms',m.lx,m.ly,m.lz, m.atoms)
for atom in vpcoloredmodel.atoms:
if(atom.vp.type == 'Full-icosahedra'):
atom.z = 1
elif(atom.vp.type == 'Icosahedra-like'):
atom.z = 2
elif(atom.vp.type == 'Mixed'):
atom.z = 3
elif(atom.vp.type == 'Crystal-like'):
atom.z = 4
elif(atom.vp.type == 'Undef'):
atom.z = 5
vpcoloredmodel.write_our_xyz('vpcolored.xyz')
#sm = Model(sys.argv[6])
#rm = Model(sys.argv[7])
#submodel_vp_colored(m,sm,rm)
#vor_stats(sm)
#vor_stats(rm)
atom_dict_m = generate_atom_dict(m)
smtable = {}
submodelfiles = os.listdir(submodel_direc)
submodelfiles = [smf for smf in submodelfiles if('real' not in smf)]
submodelfiles = [smf for smf in submodelfiles if('cif' not in smf)]
submodelfiles = [submodel_direc+smf for smf in submodelfiles]
submodelfiles.sort()
for i,submodelfile in enumerate(submodelfiles):
sm = Model(submodelfile)
for atom in sm.atoms:
if( atom in m.atoms ):
atom.vp = m.atoms[m.atoms.index(atom)].vp.copy()
smtable[submodelfile] = {}
for vptype in vp_dict:
atom_dict_sm = generate_atom_dict(sm)
#smtable[submodelfile][vptype] = len(atom_dict_sm[vptype]) / float(len(atom_dict_m[vptype]))
smtable[submodelfile][vptype] = len(atom_dict_sm[vptype]) / float(sm.natoms)
#for smf in submodelfiles:
# typedict = smtable[smf]
# print(smf)
# for vptype,ratio in typedict.iteritems():
# print(" {1}% {0}".format(vptype,round(100*ratio)))
# print(" Ratio of ico-like / xtal-like: {0}".format(typedict['Icosahedra-like']/typedict['Crystal-like']))
# print('')
print_table(smtable)
#return
# Calculate the ratio of spot intensitites
spot_ints = []
orig_intensities = read_intensity_file(orig_ft)
for i in range(0,num_spots):
for intensityfile in spot_fts:
id = get_spot_id(intensityfile)
if(i == id):
print("Intensity file: {0} => i={1}".format(intensityfile,id))
# Find and generate corresponding sub-model
# Need this to rescale by number of atoms
# smf will stay at the correct string for appending to table
for smf in submodelfiles:
if(get_spot_id(smf) == id):
sm = Model(smf)
break
j = id*7 +2
x0,y0,z0 = tuple([float(x) for x in params[j+1].split()])
sx,sy,sz = tuple([float(x) for x in params[j+2].split()])
cxy,cxz,cyz = tuple([float(x) for x in params[j+3].split()])
xi,xf,xc = tuple([float(x)*2 for x in params[j+4].split()][:3])
yi,yf,yc = tuple([float(x)*2 for x in params[j+5].split()][:3])
zi,zf,zc = tuple([float(x)*2 for x in params[j+6].split()][:3])
ft_intensities = read_intensity_file(intensityfile)
maxoi = 0.0
maxi = 0.0
for x in range(xi-1,xf):
for y in range(yi-1,yf):
for z in range(zi-1,zf):
#print(x,y,z,orig_intensities[x][y][z])
if( ft_intensities[x][y][z] > maxi ):
maxi = ft_intensities[x][y][z]
print(orig_intensities[x][y][z],x,y,z)
if( orig_intensities[x][y][z] > maxoi ):
maxoi = orig_intensities[x][y][z]
print(" Unscaled intensity ratio: {0}/{1}={2}".format(maxi,maxoi,maxi/maxoi))
smtable[smf]['UIR'] = maxi/maxoi
smtable[smf]['MUIR'] = sm.natoms/float(m.natoms)
maxoi /= m.natoms # rescale by number of atoms (ft is lineary scaling)
maxi /= sm.natoms # rescale by number of atoms (ft is lineary scaling)
smtable[smf]['IR'] = maxi/maxoi
print(" Intensity ratio: {0}/{1}={2}".format(maxi,maxoi,smtable[smf]['IR']))
spot_ints.append((id,maxi,maxoi,maxi/maxoi))
print(spot_ints)
print_table(smtable)
print_table(smtable)
return
# sys.argv[1] should be the modelfile in the xyz format
# sys.argv[2] should be the cutoff desired
modelfile = sys.argv[1]
cutoff = float(sys.argv[2])
ag = AtomGraph(modelfile,cutoff)
model = Model(modelfile)
model.generate_neighbors(cutoff)
submodelfile = sys.argv[3]
mixedmodel = Model('Mixed atoms',model.lx,model.ly,model.lz, [atom for atom in ag.model.atoms if atom.vp.type == 'Mixed'])
icolikemodel = Model('Ico-like atoms',model.lx,model.ly,model.lz, [atom for atom in ag.model.atoms if(atom.vp.type == 'Icosahedra-like' or atom.vp.type == 'Full-icosahedra')])
fullicomodel = Model('Full-icosahedra atoms',model.lx,model.ly,model.lz, [atom for atom in ag.model.atoms if atom.vp.type == 'Full-icosahedra'])
xtalmodel = Model('Xtal-like atoms',model.lx,model.ly,model.lz, [atom for atom in ag.model.atoms if atom.vp.type == 'Crystal-like'])
undefmodel = Model('Undef atoms',model.lx,model.ly,model.lz, [atom for atom in ag.model.atoms if atom.vp.type == 'Undef'])
#mixedmodel.write_cif('mixed.cif')
#mixedmodel.write_our_xyz('mixed.xyz')
#icolikemodel.write_cif('icolike.cif')
#icolikemodel.write_our_xyz('icolike.xyz')
#fullicomodel.write_cif('fullico.cif')
#fullicomodel.write_our_xyz('fullico.xyz')
#xtalmodel.write_cif('xtal.cif')
#xtalmodel.write_our_xyz('xtal.xyz')
#undefmodel.write_cif('undef.cif')
#undefmodel.write_our_xyz('undef.xyz')
icomixedmodel = Model('ico+mix atoms',model.lx,model.ly,model.lz, mixedmodel.atoms + icolikemodel.atoms)
#mixedmodel.write_cif('icomixed.cif')
#mixedmodel.write_our_xyz('icomixed.xyz')
vpcoloredmodel = Model('vp colored atoms',model.lx,model.ly,model.lz, ag.model.atoms)
for atom in vpcoloredmodel.atoms:
if(atom.vp.type == 'Full-icosahedra'):
atom.z = 1
elif(atom.vp.type == 'Icosahedra-like'):
atom.z = 2
elif(atom.vp.type == 'Mixed'):
atom.z = 3
elif(atom.vp.type == 'Crystal-like'):
atom.z = 4
elif(atom.vp.type == 'Undef'):
atom.z = 5
#vpcoloredmodel.write_cif('vpcolored.cif')
#vpcoloredmodel.write_our_xyz('vpcolored.xyz')
subvpcoloredmodel = Model(submodelfile)
for atom in subvpcoloredmodel.atoms:
atom.z = vpcoloredmodel.atoms[ag.model.atoms.index(atom)].z
subvpcoloredmodel.write_cif('subvpcolored.cif')
subvpcoloredmodel.write_our_xyz('subvpcolored.xyz')
return
golden = False
#cluster_prefix = 'ico.t3.'
#cluster_types = 'Icosahedra-like', 'Full-icosahedra' # Need this for final/further analysis
#cluster_prefix = 'fi.t3.'
#cluster_types = ['Full-icosahedra'] # Need this for final/further analysis
cluster_prefix = 'xtal.t3.'
cluster_types = 'Crystal-like' # Need this for final/further analysis
#cluster_prefix = 'mix.t3'
#cluster_types = ['Mixed'] # Need this for final/further analysis
#cluster_prefix = 'undef.t3'
#cluster_types = ['Undef'] # Need this for final/further analysis
# Decide what time of clustering you want to do
#clusters = ag.get_clusters_with_n_numneighs(cutoff,5,cluster_types) #Vertex
#clusters = ag.get_vertex_sharing_clusters(cutoff,cluster_types) #Vertex
#clusters = ag.get_edge_sharing_clusters(cutoff,cluster_types) #Edge
#clusters = ag.get_face_sharing_clusters(cutoff,cluster_types) #Face
clusters = ag.get_interpenetrating_atoms(cutoff,cluster_types) #Interpenetrating
#clusters = ag.get_interpenetrating_clusters_with_neighs(cutoff,cluster_types) #Interpenetrating+neighs
#clusters = ag.get_connected_clusters_with_neighs(cutoff, cluster_types) #Connected (vertex) + neighs
#v,e,f,i = ag.vefi_sharing(cluster_types)
#print("V: {0} E: {1} F: {2} I: {3}".format(int(v),int(e),int(f),int(i)))
orig_clusters = clusters[:]
# Print orig clusters
j = 0
for i,cluster in enumerate(clusters):
print("Orig cluster {0} contains {1} atoms.".format(i,len(cluster)))
if(golden):
for atom in cluster:
if(atom.vp.type in cluster_types):
atom.z = 0
# Save cluster files
cluster_model = Model("Orig cluster {0} contains {1} atoms.".format(i,len(cluster)),model.lx, model.ly, model.lz, cluster)
cluster_model.write_cif('{1}cluster{0}.cif'.format(i,cluster_prefix))
cluster_model.write_our_xyz('{1}cluster{0}.xyz'.format(i,cluster_prefix))
allclusters = []
for cluster in clusters:
for atom in cluster:
if(atom not in allclusters):
allclusters.append(atom)
#if(atom.vp.type in cluster_types): print(' {0}\t{1}'.format(atom,atom.vp.type))
allclusters = Model("All clusters.",model.lx, model.ly, model.lz, allclusters)
allclusters.write_cif('{0}allclusters.cif'.format(cluster_prefix))
allclusters.write_our_xyz('{0}allclusters.xyz'.format(cluster_prefix))
print("{0}allclusters.cif and {0}allclusters.xyz contain {1} atoms.".format(cluster_prefix, allclusters.natoms))
if(not golden):
x_cluster = []
for i,atom in enumerate(model.atoms):
if atom not in allclusters.atoms:
x_cluster.append(atom)
x_cluster = Model("Opposite cluster of {0}".format(cluster_prefix),model.lx, model.ly, model.lz, x_cluster)
x_cluster.write_cif('{0}opposite.cif'.format(cluster_prefix))
x_cluster.write_our_xyz('{0}opposite.xyz'.format(cluster_prefix))
print('{0}opposite.cif and {0}opposite.xyz contain {1} atoms.'.format(cluster_prefix, x_cluster.natoms))
if(False): # Further analysis
cn = 0.0
for atom in model.atoms:
cn += atom.cn
cn /= model.natoms
vpcn = 0.0
count = 0
for atom in ag.model.atoms:
if( atom.vp.type in cluster_types ):
vpcn += atom.cn
count += 1
vpcn /= count
natomsinVPclusters = allclusters.natoms # Number of atoms in VP clusters
nVPatoms = count # Number of VP atoms
numsepVPatoms = nVPatoms * vpcn # Number of atoms in VP clusters if all clusters were separated
maxnumatoms = model.natoms # Max number of atoms in VP clusters if all clusters were separated but still within the model size
print('Average CN is {0}'.format(cn))
print('Average CN of VP atoms is {0}'.format(vpcn))
print('# atoms in all clusters: {0}. # VP atoms * vpcn: {1}. # VP atoms: {2}'.format(natomsinVPclusters,numsepVPatoms,nVPatoms))
print('~ Number of VP that can fit in the model: {0}'.format(maxnumatoms/vpcn))
print('Ratio of: (# atoms involved in VP clusters)/(# atoms involved in VP clusters if all clusters were completely separated): {0}% <--- Therefore {1}% sharing.'.format(round(float(natomsinVPclusters)/(numsepVPatoms)*100.0,3),100.0-round(float(natomsinVPclusters)/(numsepVPatoms)*100.0,3)))
print('Ratio of: (# atoms involved in VP clusters)/(# atoms involved in VP clusters if all clusters were separated as much as possible within the model): {0}% <--- Therefore {1}% sharing.'.format(round(float(natomsinVPclusters)/min(numsepVPatoms,maxnumatoms)*100.0,3),100.0-round(float(natomsinVPclusters)/min(numsepVPatoms,maxnumatoms)*100.0,3) if numsepVPatoms < maxnumatoms else round(float(natomsinVPclusters)/min(numsepVPatoms,maxnumatoms)*100.0,3)))
vor_instance = Vor()
vor_instance.runall(modelfile,cutoff)
index = vor_instance.get_indexes()
vor_instance.set_atom_vp_indexes(model)
vp_dict = categorize_vor.load_param_file('/home/jjmaldonis/model_analysis/scripts/categorize_parameters_iso.txt')
atom_dict = categorize_vor.generate_atom_dict(index,vp_dict)
categorize_vor.set_atom_vp_types(model,vp_dict)
# Count the number of common neighbors in each of the VP
vp_atoms = []
for atom in model.atoms:
if(atom.vp.type in cluster_types):
vp_atoms.append(atom)
common_neighs = 0.0
atom_pairs = []
for atomi in vp_atoms:
for atomj in vp_atoms:
if(atomi != atomj):
if(atomi in atomj.neighs and [atomi,atomj] not in atom_pairs and [atomj,atomi] not in atom_pairs):
common_neighs += 1
atom_pairs.append([atomi,atomj])
#if(atomj in atomi.neighs): common_neighs += 0.5
for n in atomi.neighs:
if(n in atomj.neighs and [n,atomj] not in atom_pairs and [atomj,n] not in atom_pairs):
common_neighs += 1
atom_pairs.append([n,atomj])
#for n in atomj.neighs:
# if(n in atomi.neighs): common_neighs += 0.5
# Now common_neighs is the number of shared atoms
#print(common_neighs)
print('Percent shared based on common neighsbors: {0}'.format(100.0*common_neighs/natomsinVPclusters))