def spike_loc(cl, vl, trigger_tm, target_cl,key=None):
''' Return the indices of vl['xs'] and vl['ys']
that correspond to spikes with cluster label
target_cl.
Note: vl['xs'] and vl['ys'] have repeated values.
We must return exactly one index for each unique
location.'''
cached_results = (try_cache(key) if key is not None else None)
if cached_results is not None:
logging.info('Got spike cluster %i from Cache.cache.',target_cl)
return cached_results
if target_cl == 1:
logging.warning('Are you SURE you want to find cluster 1?')
y = raw_input()
if y in ['n', 'N']:
return np.NAN
# Get the times when a cluster in label 'target_cl' occurs
# If target_cl is None, then find the iterations for everything
st = cl['Time'][cl['Label']==target_cl]
if st.shape == (0,):
logging.info('No clusters with label %i. Quitting...', target_cl)
return np.NAN
# Ask user if he wants to waste time on an excessively large dataset
'''
logging.info('%i pts in cluster %i',len(st),target_cl)
if 1.0*len(st)/len(cl['Label']) > .05:
logging.warning('Are you SURE you want to proceed?')
return np.NAN
y = raw_input()
if y in ['n', 'N']:
return np.NAN'''
# Get the vl indices corresponding to times in st
spk_i = np.array(list(match_cl_to_vl(st, vl, trigger_tm)))
#import pdb; pdb.set_trace()
# Delete NaN values
spk_i = spk_i[~np.isnan(spk_i)].astype(int)
# Determine speed
# Matlab rounds speed to 6 or 7 decimals, so do the same for consistency
speed = matround(np.sqrt(vl['vxs']**2+vl['vys']**2),decimals=6)[spk_i]
# Only leave spikes when rat is running faster than 2 in /sec
spk_i = spk_i[(speed > 2) & (speed < 40) ]
if key is not None: store_in_cache(key, spk_i)
return np.unique(spk_i)
def match_cl_to_vl(times, vl, trigger_tm):
''' Return a generator with the iteration for each
element in array 'times' '''
times = times.astype(np.float64)
times /= (cl_rate * 24 * 60 * 60)
times += trigger_tm
# Clean up the virmenLog iterations numbers by removing the
# nan ones and linearly interpolating between them
f = np.nonzero(~np.isnan(vl['Iter num']))[0]
y = matround(np.interp(range(len(vl['Iter num'])), f, vl['Iter num'][f]),
0)
y = y.astype(int)
# Iterations are Matlab indices, so they begin at 1
# Adjust them to fit Python
y -= 1
# Determine the iteration number at which each spike occurred
# This is a real time killer
"""
for ndx in range(len(times)):
try:
# Find iteration preceeding spike
f = np.nonzero(vl['Iter time'] < times[ndx])[0][-1] # Take the latimes one
# Make sure there is an iteration following the spike
np.nonzero(vl['Iter time'] > times[ndx])[0][0]
yield y[f]
except:
# If nonzero is empty, the spike occurred before the first
# or after the last spike
yield np.NAN"""
for ndx in range(len(times)):
try:
# Find iteration preceeding spike
f = np.nonzero(
vl['Iter time'] < times[ndx])[0][-1] # Take the latimes one
yield y[f]
except:
# If nonzero is empty, the spike occurred before the first
# or after the last spike
yield np.NAN
def match_cl_to_vl(times, vl, trigger_tm):
''' Return a generator with the iteration for each
element in array 'times' '''
times = times.astype(np.float64)
times /= (cl_rate*24*60*60)
times += trigger_tm
# Clean up the virmenLog iterations numbers by removing the
# nan ones and linearly interpolating between them
f = np.nonzero(~np.isnan(vl['Iter num']))[0]
y = matround(np.interp(range(len(vl['Iter num'])), f, vl['Iter num'][f]),0)
y = y.astype(int)
# Iterations are Matlab indices, so they begin at 1
# Adjust them to fit Python
y -= 1
# Determine the iteration number at which each spike occurred
# This is a real time killer
"""
for ndx in range(len(times)):
try:
# Find iteration preceeding spike
f = np.nonzero(vl['Iter time'] < times[ndx])[0][-1] # Take the latimes one
# Make sure there is an iteration following the spike
np.nonzero(vl['Iter time'] > times[ndx])[0][0]
yield y[f]
except:
# If nonzero is empty, the spike occurred before the first
# or after the last spike
yield np.NAN"""
for ndx in range(len(times)):
try:
# Find iteration preceeding spike
f = np.nonzero(vl['Iter time'] < times[ndx])[0][-1] # Take the latimes one
yield y[f]
except:
# If nonzero is empty, the spike occurred before the first
# or after the last spike
yield np.NAN