# -*- coding: utf-8 -*-
# vim: tabstop=4 shiftwidth=4 softtabstop=4
#
# Copyright (C) 2015-2021 GEM Foundation
#
# OpenQuake is free software: you can redistribute it and/or modify it
# under the terms of the GNU Affero General Public License as published
# by the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# OpenQuake is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU Affero General Public License for more details.
#
# You should have received a copy of the GNU Affero General Public License
# along with OpenQuake. If not, see <http://www.gnu.org/licenses/>.
import ast
import os.path
import numbers
import operator
import functools
import itertools
import collections
import logging
import numpy
import pandas
from openquake.baselib.general import (
humansize, countby, AccumDict, CallableDict,
get_array, group_array, fast_agg)
from openquake.baselib.hdf5 import FLOAT, INT, get_shape_descr
from openquake.baselib.performance import performance_view
from openquake.baselib.python3compat import encode, decode
from openquake.hazardlib.gsim.base import ContextMaker
from openquake.commonlib import util, logictree
from openquake.risklib.scientific import losses_by_period, return_periods
from openquake.baselib.writers import build_header, scientificformat
from openquake.calculators.getters import get_rupture_getters
from openquake.calculators.extract import extract
F32 = numpy.float32
U32 = numpy.uint32
U8 = numpy.uint8
# a dictionary of views datastore -> array
view = CallableDict(keyfunc=lambda s: s.split(':', 1)[0])
# ########################## utility functions ############################## #
[docs]def dt(names):
"""
:param names: list or a string with space-separated names
:returns: a numpy structured dtype
"""
if isinstance(names, str):
names = names.split()
return numpy.dtype([(name, object) for name in names])
[docs]def text_table(data, header=None, fmt=None, ext='rst'):
"""
Build a .rst (or .org) table from a matrix or a DataFrame
>>> tbl = [['a', 1], ['b', 2]]
>>> print(text_table(tbl, header=['Name', 'Value']))
+------+-------+
| Name | Value |
+------+-------+
| a | 1 |
+------+-------+
| b | 2 |
+------+-------+
"""
assert ext in 'rst org', ext
if isinstance(data, pandas.DataFrame):
if data.index.name:
data = data.reset_index()
header = header or list(data.columns)
data = data.to_numpy()
if header is None and hasattr(data, '_fields'):
header = data._fields
try:
# see if data is a composite numpy array
data.dtype.fields
except AttributeError:
# not a composite array
header = header or ()
else:
if not header:
header = [col.split(':')[0] for col in build_header(data.dtype)]
if header:
col_sizes = [len(col) for col in header]
else:
col_sizes = [len(str(col)) for col in data[0]]
body = []
fmt = functools.partial(scientificformat, fmt=fmt) if fmt else form
for row in data:
tup = tuple(fmt(c) for c in row)
for (i, col) in enumerate(tup):
col_sizes[i] = max(col_sizes[i], len(col))
if len(tup) != len(col_sizes):
raise ValueError('The header has %d fields but the row %d fields!'
% (len(col_sizes), len(tup)))
body.append(tup)
wrap = '+-%s-+' if ext == 'rst' else '|-%s-|'
sepline = wrap % '-+-'.join('-' * size for size in col_sizes)
templ = '| %s |' % ' | '.join('%-{}s'.format(size) for size in col_sizes)
if header and ext == 'rst':
lines = [sepline, templ % tuple(header), sepline]
elif header and ext == 'org':
lines = [templ % tuple(header), sepline]
else:
lines = [sepline]
for row in body:
lines.append(templ % row)
if ext == 'rst':
lines.append(sepline)
return '\n'.join(lines)
[docs]@view.add('worst_sources')
def view_worst_sources(token, dstore):
"""
Returns the sources with worst weights
"""
if ':' in token:
step = int(token.split(':')[1])
else:
step = 1
data = dstore.read_df('source_data', 'src_id')
del data['nsites']
ser = data.groupby('taskno').ctimes.sum().sort_values().tail(1)
[[taskno, maxtime]] = ser.to_dict().items()
data = data[data.taskno == taskno]
print('Sources in the slowest task (%d seconds, weight=%d)'
% (maxtime, data['weight'].sum()))
data['slow_rate'] = data.ctimes / data.weight
df = data.sort_values('slow_rate', ascending=False)
return df[slice(None, None, step)]
[docs]@view.add('slow_sources')
def view_slow_sources(token, dstore, maxrows=20):
"""
Returns the slowest sources
"""
info = dstore['source_info']['source_id', 'code',
'calc_time', 'num_sites', 'eff_ruptures']
info = info[info['eff_ruptures'] > 0]
info.sort(order='calc_time')
data = numpy.zeros(len(info), dt(info.dtype.names))
for name in info.dtype.names:
data[name] = info[name]
return data[::-1][:maxrows]
[docs]@view.add('slow_ruptures')
def view_slow_ruptures(token, dstore, maxrows=25):
"""
Show the slowest ruptures
"""
fields = ['code', 'n_occ', 'mag', 'trt_smr']
rups = dstore['ruptures'][()][fields]
time = dstore['gmf_data/time_by_rup'][()]
arr = util.compose_arrays(rups, time)
arr = arr[arr['nsites'] > 0]
arr.sort(order='time')
return arr[-maxrows:]
[docs]@view.add('contents')
def view_contents(token, dstore):
"""
Returns the size of the contents of the datastore and its total size
"""
tot = (dstore.filename, humansize(os.path.getsize(dstore.filename)))
data = sorted((dstore.getsize(key), key) for key in dstore)
rows = [(key, humansize(nbytes)) for nbytes, key in data] + [tot]
return numpy.array(rows, dt('dataset size'))
[docs]def short_repr(lst):
if len(lst) <= 10:
return ' '.join(map(str, lst))
return '[%d rlzs]' % len(lst)
[docs]@view.add('full_lt')
def view_full_lt(token, dstore):
full_lt = dstore['full_lt']
num_paths = full_lt.get_num_potential_paths()
if not full_lt.num_samples and num_paths > 15000:
return '<%d realizations>' % num_paths
try:
rlzs_by_gsim_list = full_lt.get_rlzs_by_gsim_list(dstore['trt_smrs'])
except KeyError: # for scenario trt_smrs is missing
rlzs_by_gsim_list = [full_lt._rlzs_by_gsim(0)]
header = ['grp_id', 'gsim', 'rlzs']
rows = []
for grp_id, rbg in enumerate(rlzs_by_gsim_list):
for gsim, rlzs in rbg.items():
rows.append((grp_id, repr(str(gsim)), short_repr(rlzs)))
return numpy.array(rows, dt(header))
[docs]@view.add('weight_by_src')
def view_eff_ruptures(token, dstore):
info = dstore.read_df('source_info', 'source_id')
df = info.groupby('code').sum()
del df['grp_id'], df['trti']
return df
[docs]@view.add('short_source_info')
def view_short_source_info(token, dstore, maxrows=20):
return dstore['source_info'][:maxrows]
[docs]@view.add('params')
def view_params(token, dstore):
oq = dstore['oqparam']
params = ['calculation_mode', 'number_of_logic_tree_samples',
'maximum_distance', 'investigation_time',
'ses_per_logic_tree_path', 'truncation_level',
'rupture_mesh_spacing', 'complex_fault_mesh_spacing',
'width_of_mfd_bin', 'area_source_discretization',
'pointsource_distance',
'floating_x_step', 'floating_y_step',
'ground_motion_correlation_model', 'minimum_intensity',
'random_seed', 'master_seed', 'ses_seed']
if 'risk' in oq.calculation_mode:
params.append('avg_losses')
return numpy.array([(param, repr(getattr(oq, param, None)))
for param in params], dt('parameter value'))
[docs]def rst_links(*fnames):
links = []
for fname in fnames:
bname = os.path.basename(fname)
links.append("`%s <%s>`_" % (bname, bname))
return ' '.join(links)
[docs]def build_links(items):
out = []
for key, fname in items:
if isinstance(fname, dict):
for k, v in fname.items():
b = os.path.basename(v)
out.append(('reqv:' + k, "`%s <%s>`_" % (b, b)))
elif isinstance(fname, list):
out.append((key, rst_links(*fname)))
else:
out.append((key, rst_links(fname)))
return sorted(out)
def _humansize(literal):
dic = ast.literal_eval(decode(literal))
if isinstance(dic, dict):
items = sorted(dic.items(), key=operator.itemgetter(1), reverse=True)
lst = ['%s %s' % (k, humansize(v)) for k, v in items]
return ', '.join(lst)
else:
return str(dic)
[docs]@view.add('job_info')
def view_job_info(token, dstore):
"""
Determine the amount of data transferred from the controller node
to the workers and back in a classical calculation.
"""
data = []
task_info = dstore['task_info'][()]
task_sent = ast.literal_eval(decode(dstore['task_sent'][()]))
for task, dic in task_sent.items():
sent = sorted(dic.items(), key=operator.itemgetter(1), reverse=True)
sent = ['%s=%s' % (k, humansize(v)) for k, v in sent[:3]]
recv = get_array(task_info, taskname=encode(task))['received'].sum()
data.append((task, ' '.join(sent), humansize(recv)))
return numpy.array(data, dt('task sent received'))
[docs]@view.add('avglosses_data_transfer')
def avglosses_data_transfer(token, dstore):
"""
Determine the amount of average losses transferred from the workers to the
controller node in a risk calculation.
"""
oq = dstore['oqparam']
N = len(dstore['assetcol'])
R = dstore['full_lt'].get_num_rlzs()
L = len(dstore.get_attr('crm', 'loss_types'))
ct = oq.concurrent_tasks
size_bytes = N * R * L * 8 * ct # 8 byte floats
return (
'%d asset(s) x %d realization(s) x %d loss type(s) losses x '
'8 bytes x %d tasks = %s' % (N, R, L, ct, humansize(size_bytes)))
# for scenario_risk
[docs]@view.add('totlosses')
def view_totlosses(token, dstore):
"""
This is a debugging view. You can use it to check that the total
losses, i.e. the losses obtained by summing the average losses on
all assets are indeed equal to the aggregate losses. This is a
sanity check for the correctness of the implementation.
"""
oq = dstore['oqparam']
tot_losses = dstore['avg_losses-rlzs'][()].sum(axis=0)
return text_table(tot_losses.view(oq.loss_dt(F32)), fmt='%.6E')
[docs]def alt_to_many_columns(alt, loss_types):
# convert an risk_by_event in the format
# (event_id, agg_id, loss_id, loss) =>
# (event_id, agg_id, structural, nonstructural, ...)
dic = dict(event_id=[])
for ln in loss_types:
dic[ln] = []
for (eid, kid), df in alt.groupby(['event_id', 'agg_id']):
dic['event_id'].append(eid)
arr = numpy.zeros(len(loss_types))
arr[df.loss_id.to_numpy()] = df.loss.to_numpy()
for li, ln in enumerate(loss_types):
dic[ln].append(arr[li])
return pandas.DataFrame(dic)
def _portfolio_loss(dstore):
oq = dstore['oqparam']
R = dstore['full_lt'].get_num_rlzs()
K = dstore['risk_by_event'].attrs.get('K', 0)
alt = dstore.read_df('risk_by_event', 'agg_id', dict(agg_id=K))
df = alt_to_many_columns(alt, oq.loss_types)
eids = df.pop('event_id').to_numpy()
loss = df.to_numpy()
rlzs = dstore['events']['rlz_id'][eids]
L = loss.shape[1]
data = numpy.zeros((R, L))
for row, rlz in zip(loss, rlzs):
data[rlz] += row
return data
[docs]@view.add('portfolio_losses')
def view_portfolio_losses(token, dstore):
"""
The losses for the full portfolio, for each realization and loss type,
extracted from the event loss table.
"""
oq = dstore['oqparam']
loss_dt = oq.loss_dt()
data = _portfolio_loss(dstore).view(loss_dt)[:, 0] # shape R
rlzids = [str(r) for r in range(len(data))]
array = util.compose_arrays(numpy.array(rlzids), data, 'rlz_id')
# this is very sensitive to rounding errors, so I am using a low precision
return text_table(array, fmt='%.5E')
[docs]@view.add('portfolio_loss')
def view_portfolio_loss(token, dstore):
"""
The mean portfolio loss for each loss type,
extracted from the event loss table.
"""
oq = dstore['oqparam']
R = dstore['full_lt'].get_num_rlzs()
K = dstore['risk_by_event'].attrs.get('K', 0)
alt_df = dstore.read_df('risk_by_event', 'agg_id', dict(agg_id=K))
weights = dstore['weights'][:]
rlzs = dstore['events']['rlz_id']
E = len(rlzs)
ws = weights[rlzs]
avgs = []
for li, ln in enumerate(oq.loss_types):
df = alt_df[alt_df.loss_id == li]
eids = df.pop('event_id').to_numpy()
[docs] avgs.append(ws[eids] @ df.loss.to_numpy() / ws.sum() * E / R)
return text_table([['avg'] + avgs], ['loss'] + oq.loss_types)
@view.add('portfolio_dmgdist')
def portfolio_dmgdist(token, dstore):
"""
The portfolio damages extracted from the first realization of damages-rlzs
"""
oq = dstore['oqparam']
dstates = ['no_damage'] + oq.limit_states
D = len(dstates)
arr = dstore['damages-rlzs'][:, 0, :, :D].sum(axis=0) # shape (L, D)
tbl = numpy.zeros(len(arr), dt(['loss_type', 'total'] + dstates))
tbl['loss_type'] = oq.loss_types
tbl['total'] = arr.sum(axis=1)
for dsi, ds in enumerate(dstates):
tbl[ds] = arr[:, dsi]
return tbl
[docs]@view.add('portfolio_damage')
def view_portfolio_damage(token, dstore):
"""
The mean full portfolio damage for each loss type,
extracted from the average damages
"""
if 'aggcurves' in dstore: # event_based_damage
K = dstore.get_attr('risk_by_event', 'K')
df = dstore.read_df('aggcurves', sel=dict(agg_id=K, return_period=0))
lnames = numpy.array(dstore['oqparam'].loss_types)
df['loss_type'] = lnames[df.loss_id.to_numpy()]
del df['loss_id']
del df['agg_id']
del df['return_period']
return df.set_index('loss_type')
# dimensions assets, stat, loss_types, dmg_state
if 'damages-stats' in dstore:
attrs = get_shape_descr(dstore['damages-stats'].attrs['json'])
arr = dstore.sel('damages-stats', stat='mean').sum(axis=(0, 1))
else:
attrs = get_shape_descr(dstore['damages-rlzs'].attrs['json'])
arr = dstore.sel('damages-rlzs', rlz=0).sum(axis=(0, 1))
rows = [(lt,) + tuple(row) for lt, row in zip(attrs['loss_type'], arr)]
return numpy.array(rows, dt(['loss_type'] + list(attrs['dmg_state'])))
[docs]def sum_table(records):
"""
Used to compute summaries. The records are assumed to have numeric
fields, except the first field which is ignored, since it typically
contains a label. Here is an example:
>>> sum_table([('a', 1), ('b', 2)])
['total', 3]
"""
size = len(records[0])
result = [None] * size
firstrec = records[0]
for i in range(size):
if isinstance(firstrec[i], (numbers.Number, numpy.ndarray)):
result[i] = sum(rec[i] for rec in records)
else:
result[i] = 'total'
return result
[docs]@view.add('exposure_info')
def view_exposure_info(token, dstore):
"""
Display info about the exposure model
"""
assetcol = dstore['assetcol/array'][:]
taxonomies = sorted(set(dstore['assetcol'].taxonomies))
data = [('#assets', len(assetcol)),
('#taxonomies', len(taxonomies))]
return text_table(data) + '\n\n' + view_assets_by_site(token, dstore)
[docs]@view.add('ruptures_events')
def view_ruptures_events(token, dstore):
num_ruptures = len(dstore['ruptures'])
num_events = len(dstore['events'])
events_by_rlz = countby(dstore['events'][()], 'rlz_id')
mult = round(num_events / num_ruptures, 3)
lst = [('Total number of ruptures', num_ruptures),
('Total number of events', num_events),
('Rupture multiplicity', mult),
('Events by rlz', events_by_rlz.values())]
return text_table(lst)
[docs]@view.add('fullreport')
def view_fullreport(token, dstore):
"""
Display an .rst report about the computation
"""
# avoid circular imports
from openquake.calculators.reportwriter import ReportWriter
return ReportWriter(dstore).make_report()
[docs]def stats(name, array, *extras):
"""
Returns statistics from an array of numbers.
:param name: a descriptive string
:returns: (name, mean, rel_std, min, max, len) + extras
"""
avg = numpy.mean(array)
std = 'nan' if len(array) == 1 else '%d%%' % (numpy.std(array) / avg * 100)
max_ = numpy.max(array)
return (name, len(array), avg, std, numpy.min(array), max_) + extras
[docs]@view.add('num_units')
def view_num_units(token, dstore):
"""
Display the number of units by taxonomy
"""
taxo = dstore['assetcol/tagcol/taxonomy'][()]
counts = collections.Counter()
for asset in dstore['assetcol']:
counts[taxo[asset['taxonomy']]] += asset['value-number']
data = sorted(counts.items())
data.append(('*ALL*', sum(d[1] for d in data)))
return numpy.array(data, dt('taxonomy num_units'))
[docs]@view.add('assets_by_site')
def view_assets_by_site(token, dstore):
"""
Display statistical information about the distribution of the assets
"""
taxonomies = dstore['assetcol/tagcol/taxonomy'][()]
assets_by_site = dstore['assetcol'].assets_by_site()
data = ['taxonomy num_sites mean stddev min max num_assets'.split()]
num_assets = AccumDict()
for assets in assets_by_site:
num_assets += {k: [len(v)] for k, v in group_array(
assets, 'taxonomy').items()}
for taxo in sorted(num_assets):
val = numpy.array(num_assets[taxo])
data.append(stats(taxonomies[taxo], val, val.sum()))
if len(num_assets) > 1: # more than one taxonomy, add a summary
n_assets = numpy.array([len(assets) for assets in assets_by_site])
data.append(stats('*ALL*', n_assets, n_assets.sum()))
return text_table(data)
[docs]@view.add('required_params_per_trt')
def view_required_params_per_trt(token, dstore):
"""
Display the parameters needed by each tectonic region type
"""
full_lt = dstore['full_lt']
tbl = []
for trt in full_lt.trts:
gsims = full_lt.gsim_lt.values[trt]
maker = ContextMaker(trt, gsims, {'imtls': {}})
distances = sorted(maker.REQUIRES_DISTANCES)
siteparams = sorted(maker.REQUIRES_SITES_PARAMETERS)
ruptparams = sorted(maker.REQUIRES_RUPTURE_PARAMETERS)
tbl.append((trt, ' '.join(map(repr, map(repr, gsims))),
distances, siteparams, ruptparams))
return text_table(
tbl, header='trt_smr gsims distances siteparams ruptparams'.split(),
fmt=scientificformat)
[docs]@view.add('task_info')
def view_task_info(token, dstore):
"""
Display statistical information about the tasks performance.
It is possible to get full information about a specific task
with a command like this one, for a classical calculation::
$ oq show task_info:classical
"""
task_info = dstore['task_info']
task_info.refresh()
args = token.split(':')[1:] # called as task_info:task_name
if args:
[task] = args
array = get_array(task_info[()], taskname=task.encode('utf8'))
rduration = array['duration'] / array['weight']
data = util.compose_arrays(rduration, array, 'rduration')
data.sort(order='duration')
return data
data = []
for task, arr in group_array(task_info[()], 'taskname').items():
val = arr['duration']
if len(val):
data.append(stats(task, val, val.max() / val.mean()))
if not data:
return 'Not available'
return numpy.array(
data, dt('operation-duration counts mean stddev min max slowfac'))
[docs]@view.add('task_durations')
def view_task_durations(token, dstore):
"""
Display the raw task durations. Here is an example of usage::
$ oq show task_durations:classical
"""
task = token.split(':')[1] # called as task_duration:task_name
array = get_array(dstore['task_info'][()], taskname=task)['duration']
return '\n'.join(map(str, array))
[docs]@view.add('task')
def view_task_hazard(token, dstore):
"""
Display info about a given task. Here are a few examples of usage::
$ oq show task:classical:0 # the fastest task
$ oq show task:classical:-1 # the slowest task
"""
_, name, index = token.split(':')
if 'source_data' not in dstore:
return 'Missing source_data'
data = get_array(dstore['task_info'][()], taskname=encode(name))
if len(data) == 0:
raise RuntimeError('No task_info for %s' % name)
data.sort(order='duration')
rec = data[int(index)]
taskno = rec['task_no']
sdata = dstore.read_df('source_data', 'taskno')
eff_ruptures = sdata.loc[taskno].nrupts.sum()
eff_sites = sdata.loc[taskno].nsites.sum()
res = ('taskno=%d, eff_ruptures=%d, eff_sites=%d, weight=%d, duration=%d s'
% (taskno, eff_ruptures, eff_sites, rec['weight'], rec['duration']))
return res
[docs]@view.add('source_data')
def view_source_data(token, dstore):
"""
Display info about a given task. Here is an example::
$ oq show source_data:42
"""
if ':' not in token:
return dstore.read_df(token, 'src_id')
_, taskno = token.split(':')
taskno = int(taskno)
if 'source_data' not in dstore:
return 'Missing source_data'
df = dstore.read_df('source_data', 'src_id', sel={'taskno': taskno})
del df['taskno']
df['slowrate'] = df['ctimes'] / df['weight']
return df.sort_values('ctimes')
[docs]@view.add('task_ebrisk')
def view_task_ebrisk(token, dstore):
"""
Display info about ebrisk tasks:
$ oq show task_ebrisk:-1 # the slowest task
"""
idx = int(token.split(':')[1])
task_info = get_array(dstore['task_info'][()], taskname=b'ebrisk')
task_info.sort(order='duration')
info = task_info[idx]
times = get_array(dstore['gmf_info'][()], task_no=info['task_no'])
extra = times[['nsites', 'gmfbytes', 'dt']]
ds = dstore.parent if dstore.parent else dstore
rups = ds['ruptures']['id', 'code', 'n_occ', 'mag'][times['rup_id']]
codeset = set('code_%d' % code for code in numpy.unique(rups['code']))
tbl = text_table(util.compose_arrays(rups, extra))
codes = ['%s: %s' % it for it in ds.getitem('ruptures').attrs.items()
if it[0] in codeset]
msg = '%s\n%s\nHazard time for task %d: %d of %d s, ' % (
tbl, '\n'.join(codes), info['task_no'], extra['dt'].sum(),
info['duration'])
msg += 'gmfbytes=%s, w=%d' % (
humansize(extra['gmfbytes'].sum()),
(rups['n_occ'] * extra['nsites']).sum())
return msg
[docs]@view.add('global_hazard')
def view_global_hazard(token, dstore):
"""
Display the global hazard for the calculation. This is used for
debugging purposes when comparing the results of two
calculations.
"""
imtls = dstore['oqparam'].imtls
arr = dstore.sel('hcurves-stats', stat='mean') # shape N, S, M, L
res = tuple(arr.mean(axis=(0, 1, 3))) # length M
return numpy.array([res], dt(imtls))
[docs]@view.add('global_hmaps')
def view_global_hmaps(token, dstore):
"""
Display the global hazard maps for the calculation. They are
used for debugging purposes when comparing the results of two
calculations. They are the mean over the sites of the mean hazard
maps.
"""
oq = dstore['oqparam']
dt = numpy.dtype([('%s-%s' % (imt, poe), F32)
for imt in oq.imtls for poe in oq.poes])
hmaps = dstore.sel('hmaps-stats', stat='mean')
means = hmaps.mean(axis=(0, 1)) # shape M, P
return numpy.array([tuple(means.flatten())], dt)
[docs]@view.add('global_gmfs')
def view_global_gmfs(token, dstore):
"""
Display GMFs on the first IMT averaged on everything for debugging purposes
"""
imtls = dstore['oqparam'].imtls
row = [dstore[f'gmf_data/gmv_{m}'][:].mean(axis=0)
for m in range(len(imtls))]
return text_table([row], header=imtls)
[docs]@view.add('gmf')
def view_gmf(token, dstore):
"""
Display a mean gmf for debugging purposes
"""
df = dstore.read_df('gmf_data', 'sid')
gmf = df.groupby(df.index).mean()
return str(gmf)
[docs]def binning_error(values, eids, nbins=10):
"""
:param values: E values
:param eids: E integer event indices
:returns: std/mean for the sums of the values
Group the values in nbins depending on the eids and returns the
variability of the sums relative to the mean.
"""
df = pandas.DataFrame({'val': values}, eids)
res = df.groupby(eids % nbins).val.sum()
return res.std() / res.mean()
[docs]@view.add('extreme_gmvs')
def view_extreme_gmvs(token, dstore):
"""
Display table of extreme GMVs with fields (eid, gmv_0, sid, rlz. rup)
"""
if ':' in token:
maxgmv = float(token.split(':')[1])
else:
maxgmv = 10 # 10g is default value defining extreme GMVs
imt0 = list(dstore['oqparam'].imtls)[0]
eids = dstore['gmf_data/eid'][:]
gmvs = dstore['gmf_data/gmv_0'][:]
sids = dstore['gmf_data/sid'][:]
msg = ''
err = binning_error(gmvs, eids)
if err > .05:
msg += ('Your results are expected to have a large dependency '
'from the rupture seed: %d%%' % (err * 100))
if imt0.startswith(('PGA', 'SA(')):
gmpe = GmpeExtractor(dstore)
df = pandas.DataFrame({'gmv_0': gmvs, 'sid': sids}, eids)
extreme_df = df[df.gmv_0 > maxgmv].rename(
columns={'gmv_0': imt0})
ev = dstore['events'][()][extreme_df.index]
extreme_df['rlz'] = ev['rlz_id']
extreme_df['rup'] = ev['rup_id']
trt_smrs = dstore['ruptures']['trt_smr'][extreme_df.rup]
extreme_df['gmpe'] = gmpe.extract(trt_smrs, ev['rlz_id'])
exdf = extreme_df.sort_values(imt0).groupby('sid').head(1)
if len(exdf):
msg += ('\nThere are extreme GMVs, run `oq show extreme_gmvs:%s`'
'to see them' % maxgmv)
if ':' in token:
msg += '\n%s' % exdf.set_index('rup')
return msg
return msg + '\nCould not extract extreme GMVs for ' + imt0
[docs]@view.add('mean_disagg')
def view_mean_disagg(token, dstore):
"""
Display mean quantities for the disaggregation. Useful for checking
differences between two calculations.
"""
N, M, P, Z = dstore['hmap4'].shape
tbl = []
kd = {key: dset[:] for key, dset in sorted(dstore['disagg'].items())}
oq = dstore['oqparam']
for s in range(N):
for m, imt in enumerate(oq.imtls):
for p in range(P):
row = ['%s-sid-%d-poe-%s' % (imt, s, p)]
for k, d in kd.items():
row.append(d[s, m, p].mean())
tbl.append(tuple(row))
return numpy.array(sorted(tbl), dt(['key'] + list(kd)))
[docs]@view.add('disagg_times')
def view_disagg_times(token, dstore):
"""
Display slow tasks for disaggregation
"""
data = dstore['disagg_task'][:]
info = dstore.read_df('task_info', 'taskname').loc[b'compute_disagg']
tbl = []
for duration, task_no in zip(info['duration'], info['task_no']):
tbl.append((duration, task_no) + tuple(data[task_no]))
header = ('duration', 'task_no') + data.dtype.names
return numpy.array(sorted(tbl), dt(header))
[docs]@view.add('bad_ruptures')
def view_bad_ruptures(token, dstore):
"""
Display the ruptures degenerating to a point
"""
data = dstore['ruptures']['id', 'code', 'mag',
'minlon', 'maxlon', 'minlat', 'maxlat']
bad = data[numpy.logical_and(data['minlon'] == data['maxlon'],
data['minlat'] == data['maxlat'])]
return bad
Source = collections.namedtuple(
'Source', 'source_id code num_ruptures checksum')
[docs]@view.add('disagg_by_grp')
def view_disagg_by_grp(token, dstore):
"""
Show the source groups contributing the most to the highest IML
"""
data = dstore['disagg_by_grp'][()]
data.sort(order='avg_poe')
return data[::-1]
[docs]@view.add('gmvs_to_hazard')
def view_gmvs_to_hazard(token, dstore):
"""
Show the number of GMFs over the highest IML
"""
args = token.split(':')[1:] # called as view_gmvs_to_hazard:sid
if not args:
sid = 0
elif len(args) == 1: # only sid specified
sid = int(args[0])
assert sid in dstore['sitecol'].sids
oq = dstore['oqparam']
num_ses = oq.ses_per_logic_tree_path
data = dstore.read_df('gmf_data', 'sid').loc[sid]
tbl = []
for imti, (imt, imls) in enumerate(oq.imtls.items()):
gmv = data['gmv_%d' % imti].to_numpy()
for iml in imls:
# same algorithm as in _gmvs_to_haz_curve
exceeding = numpy.sum(gmv >= iml)
poe = 1 - numpy.exp(- exceeding / num_ses)
tbl.append((sid, imt, iml, exceeding, poe))
return numpy.array(tbl, dt('sid imt iml num_exceeding poe'))
[docs]@view.add('gmvs')
def view_gmvs(token, dstore):
"""
Show the GMVs on a given site ID
"""
sid = int(token.split(':')[1]) # called as view_gmvs:sid
assert sid in dstore['sitecol'].sids
data = dstore.read_df('gmf_data', 'sid')
return data.loc[sid]
[docs]@view.add('events_by_mag')
def view_events_by_mag(token, dstore):
"""
Show how many events there are for each magnitude
"""
rups = dstore['ruptures'][()]
num_evs = fast_agg(dstore['events']['rup_id'])
counts = {}
for mag, grp in group_array(rups, 'mag').items():
counts[mag] = sum(num_evs[rup_id] for rup_id in grp['id'])
return numpy.array(list(counts.items()), dt('mag num_events'))
[docs]@view.add('ebrups_by_mag')
def view_ebrups_by_mag(token, dstore):
"""
Show how many event based ruptures there are for each magnitude
"""
mags = dstore['ruptures']['mag']
uniq, counts = numpy.unique(mags, return_counts=True)
return text_table(zip(uniq, counts), ['mag', 'num_ruptures'])
[docs]@view.add('maximum_intensity')
def view_maximum_intensity(token, dstore):
"""
Show intensities at minimum and maximum distance for the highest magnitude
"""
effect = extract(dstore, 'effect')
data = zip(dstore['full_lt'].trts, effect[-1, -1], effect[-1, 0])
return text_table(data, ['trt', 'intensity1', 'intensity2'])
[docs]@view.add('extreme_sites')
def view_extreme(token, dstore):
"""
Show sites where the mean hazard map reaches maximum values
"""
mean = dstore.sel('hmaps-stats', stat='mean')[:, 0, 0, -1] # shape N1MP
site_ids, = numpy.where(mean == mean.max())
return dstore['sitecol'][site_ids]
[docs]@view.add('zero_losses')
def view_zero_losses(token, dstore):
"""
Sanity check on avg_losses and avg_gmf
"""
R = len(dstore['weights'])
oq = dstore['oqparam']
avg_gmf = dstore['avg_gmf'][0]
asset_df = dstore.read_df('assetcol/array', 'site_id')
for col in asset_df.columns:
if not col.startswith('value-'):
del asset_df[col]
values_df = asset_df.groupby(asset_df.index).sum()
avglosses = dstore['avg_losses-rlzs'][:].sum(axis=1) / R # shape (A, L)
dic = dict(site_id=dstore['assetcol']['site_id'])
for lti, lname in enumerate(oq.loss_types):
dic[lname] = avglosses[:, lti]
losses_df = pandas.DataFrame(dic).groupby('site_id').sum()
sids = losses_df.index.to_numpy()
avg_gmf = avg_gmf[sids]
nonzero_gmf = (avg_gmf > oq.min_iml).any(axis=1)
nonzero_losses = (losses_df > 0).to_numpy().any(axis=1)
bad, = numpy.where(nonzero_gmf != nonzero_losses)
# this happens in scenario_risk/case_shakemap and case_3
msg = 'Site #%d is suspicious:\navg_gmf=%s\navg_loss=%s\nvalues=%s'
for idx in bad:
sid = sids[idx]
logging.warning(msg, sid, dict(zip(oq.all_imts(), avg_gmf[sid])),
_get(losses_df, sid), _get(values_df, sid))
return bad
def _get(df, sid):
return df.loc[sid].to_dict()
[docs]@view.add('gsim_for_event')
def view_gsim_for_event(token, dstore):
"""
Display the GSIM used when computing the GMF for the given event:
$ oq show gsim_for_event:123 -1
[BooreAtkinson2008]
"""
eid = int(token.split(':')[1])
full_lt = dstore['full_lt']
rup_id, rlz_id = dstore['events'][eid][['rup_id', 'rlz_id']]
trt_smr = dstore['ruptures'][rup_id]['trt_smr']
trti = trt_smr // len(full_lt.sm_rlzs)
gsim = full_lt.get_realizations()[rlz_id].gsim_rlz.value[trti]
return gsim
[docs]@view.add('event_loss_table')
def view_event_loss_table(token, dstore):
"""
Display the top 20 losses of the event loss table for the first loss type
$ oq show event_loss_table
"""
K = dstore['risk_by_event'].attrs.get('K', 0)
df = dstore.read_df('risk_by_event', 'event_id',
dict(agg_id=K, loss_id=0))
df['std'] = numpy.sqrt(df.variance)
df.sort_values('loss', ascending=False, inplace=True)
del df['agg_id']
del df['loss_id']
del df['variance']
return df[:20]
[docs]@view.add('delta_loss')
def view_delta_loss(token, dstore):
"""
Estimate the stocastic error on the loss curve by splitting the events
in odd and even. Example:
$ oq show delta_loss # consider the first loss type
"""
if ':' in token:
_, li = token.split(':')
li = int(li)
else:
li = 0
oq = dstore['oqparam']
efftime = oq.investigation_time * oq.ses_per_logic_tree_path * len(
dstore['weights'])
num_events = len(dstore['events'])
num_events0 = num_events // 2 + (num_events % 2)
num_events1 = num_events // 2
periods = return_periods(efftime, num_events)[1:-1]
K = dstore['risk_by_event'].attrs.get('K', 0)
df = dstore.read_df('risk_by_event', 'event_id',
dict(agg_id=K, loss_id=li))
if len(df) == 0: # for instance no fatalities
return {'delta': numpy.zeros(1)}
mod2 = df.index % 2
losses0 = df['loss'][mod2 == 0]
losses1 = df['loss'][mod2 == 1]
c0 = losses_by_period(losses0, periods, num_events0, efftime / 2)
c1 = losses_by_period(losses1, periods, num_events1, efftime / 2)
ok = (c0 != 0) & (c1 != 0)
c0 = c0[ok]
c1 = c1[ok]
losses = losses_by_period(df['loss'], periods, num_events, efftime)[ok]
dic = dict(loss=losses, even=c0, odd=c1,
delta=numpy.abs(c0 - c1) / (c0 + c1))
return pandas.DataFrame(dic, periods[ok])
[docs]def to_str(arr):
return ' '.join(map(str, numpy.unique(arr)))
[docs]@view.add('composite_source_model')
def view_composite_source_model(token, dstore):
"""
Show the structure of the CompositeSourceModel in terms of grp_id
"""
lst = []
n = len(dstore['full_lt'].sm_rlzs)
trt_smrs = dstore['trt_smrs'][:]
for grp_id, df in dstore.read_df('source_info').groupby('grp_id'):
trts, sm_rlzs = numpy.divmod(trt_smrs[grp_id], n)
lst.append((str(grp_id), to_str(trts), to_str(sm_rlzs), len(df)))
return numpy.array(lst, dt('grp_id trt smrs num_sources'))
[docs]@view.add('branches')
def view_branches(token, dstore):
"""
Show info about the branches in the logic tree
"""
full_lt = dstore['full_lt']
smlt = full_lt.source_model_lt
gslt = full_lt.gsim_lt
tbl = []
for k, v in full_lt.source_model_lt.shortener.items():
tbl.append((k, v, smlt.branches[k].value))
gsims = sum(gslt.values.values(), [])
if len(gslt.shortener) < len(gsims): # possible for engine < 3.13
raise ValueError(
'There were duplicated branch IDs in the gsim logic tree %s'
% gslt.filename)
for g, (k, v) in enumerate(gslt.shortener.items()):
tbl.append((k, v, str(gsims[g]).replace('\n', r'\n')))
return numpy.array(tbl, dt('branch_id abbrev uvalue'))
[docs]@view.add('rlz')
def view_rlz(token, dstore):
"""
Show info about a given realization in the logic tree
Example:
$ oq show rlz:0 -1
"""
_, rlz_id = token.split(':')
full_lt = dstore['full_lt']
rlz = full_lt.get_realizations()[int(rlz_id)]
smlt = full_lt.source_model_lt
gslt = full_lt.gsim_lt
tbl = []
for bset, brid in zip(smlt.branchsets, rlz.sm_lt_path):
tbl.append((bset.uncertainty_type, smlt.branches[brid].value))
for trt, value in zip(sorted(gslt.bsetdict), rlz.gsim_rlz.value):
tbl.append((trt, value))
return numpy.array(tbl, dt('uncertainty_type uvalue'))
[docs]@view.add('branchsets')
def view_branchsets(token, dstore):
"""
Show the branchsets in the logic tree
"""
flt = dstore['full_lt']
clt = logictree.compose(flt.gsim_lt, flt.source_model_lt)
return text_table(enumerate(map(repr, clt.branchsets)),
header=['bsno', 'bset'], ext='org')
[docs]@view.add('rupture')
def view_rupture(token, dstore):
"""
Show a rupture with its geometry
"""
rup_id = int(token.split(':')[1])
slc = slice(rup_id, rup_id + 1)
dicts = []
for rgetter in get_rupture_getters(dstore, slc=slc):
dicts.append(rgetter.get_rupdict())
return str(dicts)
[docs]@view.add('event_rates')
def view_event_rates(token, dstore):
"""
Show the number of events per realization multiplied by risk_time/eff_time
"""
oq = dstore['oqparam']
R = dstore['full_lt'].get_num_rlzs()
if oq.calculation_mode != 'event_based_damage':
return numpy.ones(R)
time_ratio = (oq.risk_investigation_time or oq.investigation_time) / (
oq.ses_per_logic_tree_path * oq.investigation_time)
if oq.collect_rlzs:
return numpy.array([len(dstore['events']) * time_ratio / R])
else:
rlzs = dstore['events']['rlz_id']
return numpy.bincount(rlzs, minlength=R) * time_ratio
[docs]def tup2str(tups):
return ['_'.join(map(str, t)) for t in tups]
[docs]@view.add('sum')
def view_sum(token, dstore):
"""
Show the sum of an array of shape (A, R, L, ...) on the first axis
"""
_, arrayname = token.split(':') # called as sum:damages-rlzs
dset = dstore[arrayname]
A, R, L, *D = dset.shape
cols = ['RL'] + tup2str(itertools.product(*[range(d) for d in D]))
arr = dset[:].sum(axis=0) # shape R, L, *D
z = numpy.zeros(R * L, dt(cols))
for r, ar in enumerate(arr):
for li, a in enumerate(ar):
a = a.flatten()
for c, col in enumerate(cols):
z[r * L + li][col] = a[c-1] if c > 0 else (r, li)
return z
[docs]@view.add('agg_id')
def view_agg_id(token, dstore):
"""
Show the available aggregations
"""
dfa = dstore.read_df('agg_keys')
keys = [col for col in dfa.columns if not col.endswith('_')]
df = dfa[keys]
totdf = pandas.DataFrame({key: ['*total*'] for key in keys})
concat = pandas.concat([df, totdf], ignore_index=True)
concat.index.name = 'agg_id'
return concat
[docs]@view.add('mean_perils')
def view_mean_perils(token, dstore):
"""
For instance `oq show mean_perils`
"""
oq = dstore['oqparam']
pdcols = dstore.get_attr('gmf_data', '__pdcolumns__').split()
perils = [col for col in pdcols[2:] if not col.startswith('gmv_')]
N = len(dstore['sitecol/sids'])
sid = dstore['gmf_data/sid'][:]
out = numpy.zeros(N, [(per, float) for per in perils])
if oq.number_of_logic_tree_samples:
E = len(dstore['events'])
for peril in perils:
out[peril] = fast_agg(sid, dstore['gmf_data/' + peril][:]) / E
else:
rlz_weights = dstore['weights'][:]
ev_weights = rlz_weights[dstore['events']['rlz_id']]
totw = ev_weights.sum() # num_gmfs
for peril in perils:
data = dstore['gmf_data/' + peril][:]
weights = ev_weights[dstore['gmf_data/eid'][:]]
out[peril] = fast_agg(sid, data * weights) / totw
return out
[docs]@view.add('src_groups')
def view_src_groups(token, dstore):
"""
Show the hazard contribution of each source group
"""
disagg = dstore['disagg_by_grp'][:]
contrib = disagg['avg_poe'] / disagg['avg_poe'].sum()
source_info = dstore['source_info'][:]
tbl = []
for grp_id, rows in group_array(source_info, 'grp_id').items():
srcs = decode(rows['source_id'])
if len(srcs) > 2:
text = ' '.join(srcs[:2]) + ' ...'
else:
text = ' '.join(srcs)
tbl.append((grp_id, contrib[grp_id], text))
tbl.sort(key=operator.itemgetter(1), reverse=True)
return text_table(tbl, header=['grp_id', 'contrib', 'sources'],
ext='org')
[docs]@view.add('rup_stats')
def view_rup_stats(token, dstore):
"""
Show the statistics of event based ruptures
"""
rups = dstore['ruptures'][:]
out = [stats(f, rups[f]) for f in 'mag n_occ'.split()]
return numpy.array(out, dt('kind counts mean stddev min max'))