def dump_data(self, directory, time_stamp=False):
"""Save the data in directory. The files are prefixed with
config.timestamp in HH.MM.SS_ format if time_stamp is True."""
path = directory
tables = []
if not os.access(path, os.W_OK):
config.LOGGER.warning('data directory: ' + path +
' is not writable')
return
else:
if not os.access(path, os.F_OK):
os.mkdir(path)
if time_stamp:
prefix = config.timestamp.strftime('%H.%M.%S') + '_'
else:
prefix = ''
for table_id in self.data.children():
table = moose.Table(table_id)
tables.append(table)
file_name = prefix + table.name + '.plot'
file_path = os.path.join(path, file_name)
if table.stepMode == 3:
ts = numpy.linspace(0, self.simtime, len(table))
numpy.savetxt(file_path, zip(ts, numpy.asarray(table)))
else:
table.dumpFile(file_path)
config.LOGGER.info('Dumped data in %s' % (file_path))
return tables
def loop_h5_alx(SEQS, CDTS, TMPS, NUMS, h5file, ppty, tpostproc_group, alx_kwargs):
arrayname_pattern = alx_kwargs["arraynamepattern"]
for seq in SEQS:
alx_dd = calc_alx_dd(seq)
for cdt in CDTS:
arrayname = arrayname_pattern.format(seq=seq, cdt=cdt)
if tpostproc_group.__contains__(arrayname):
print "{0} HAS ALREADY EXISTED".format(arrayname)
pass
else:
denominator, xcoln, ycoln = alx_dd[ppty] # xcol, ycol name, respectively
# pf is used to identify the tables in ogd
pfpattern = alx_kwargs["pfpattern"]
# collect tables in ogd
tables = []
for num in NUMS:
pf = pfpattern.format(seq=seq, cdt=cdt, num=num)
if h5file.root.ogd.__contains__(pf):
tables.append(h5file.getNode(h5file.root.ogd, pf))
else:
print "{0} doesn't exist, you know this? Right!".format(pf)
# need to compare the axes first and at the same time get the
# minimum length of all the data first, all could be done by
# get_minimum_length function
min_len, xaxis_ref = get_minimum_length(tables, xcoln)
# starting to collect vales along the y axis
ave_ds = [] # ave distances
for table in tables:
distance = table.read(field=ycoln)[:min_len]
ave_ds.append(distance)
ave_ds = np.array(ave_ds) / denominator
alx_ave = np.average(ave_ds, axis=0) # summing along the 0th axis
# alx_std = np.std(ave_ds, axis=0)
from scipy import stats
alx_std = stats.sem(ave_ds, axis=0)
# if np.array(alx_index_ref) in previously code, comparing two arrays raise complexity
# alx_index = np.array(alx_index_ref)
# transpose to be in consistend with tables in ogd
alx_result = np.array([xaxis_ref, alx_ave, alx_std]).transpose()
h5file.createArray(
tpostproc_group,
arrayname,
alx_result,
title=(
"average along the x axis over all replicas" "column 0, 1, 2 are x axis, ave, std, respectively"
),
)
print "{0} IS DONE".format(arrayname)
h5file.close()
def loadData(self, symbolList, start, end):
tables = []
for vtSymbol in symbolList:
data = self.cache.get(vtSymbol, start, end)
if len(data):
data["vtSymbol"] = vtSymbol
data["symbol"], data["exchange"] = vtSymbol.split(":", 1)
tables.append(data)
if len(tables):
frame = pd.concat(tables,
ignore_index=True).sort_values("datetime")
return frame
else:
return pd.DataFrame()
def find_tables(h5file, table_names, group_name=''):
tables = []
for table_name in table_names:
tables.append(find_table(h5file, table_name, group_name))
return tables
