Revision 6c28dacde866ea096e45e6c3f1dd24dad5b3d395 authored by Ian Harry on 11 October 2019, 20:00:10 UTC, committed by GitHub on 11 October 2019, 20:00:10 UTC
1 parent 84e8c92
pycbc_make_inference_workflow
#!/usr/bin/env python
# Copyright (C) 2016 Christopher M. Biwer, Alexander Harvey Nitz
#
# This program is free software; you can redistribute it and/or modify it
# under the terms of the GNU General Public License as published by the
# Free Software Foundation; either version 3 of the License, or (at your
# option) any later version.
#
# This program 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 General
# Public License for more details.
#
# You should have received a copy of the GNU General Public License along
# with this program; if not, write to the Free Software Foundation, Inc.,
# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
""" Creates a DAX for a parameter estimation workflow.
"""
import argparse
import h5py
import logging
import os
import Pegasus.DAX3 as dax
import pycbc.workflow.minifollowups as mini
import pycbc.workflow.pegasus_workflow as wdax
import socket
import sys
from ligo import segments
from pycbc import results
from pycbc.results import layout
from pycbc.types import MultiDetOptionAction
from pycbc.types import MultiDetOptionAppendAction
from pycbc.workflow import configuration
from pycbc.workflow import core
from pycbc.workflow import datafind
from pycbc.workflow import plotting
from pycbc import __version__
# FIXME see https://github.com/gwastro/pycbc/issues/2349
import pycbc.workflow.inference_followups as inffu
def to_file(path, ifo=None):
""" Takes a str and returns a pycbc.workflow.pegasus_workflow.File
instance.
"""
fil = wdax.File(os.path.basename(path))
fil.ifo = ifo
path = os.path.abspath(path)
fil.PFN(path, "local")
return fil
def symlink_path(f, path):
""" Symlinks a path.
"""
if f is None:
return
try:
os.symlink(f.storage_path, os.path.join(path, f.name))
except OSError:
pass
# command line parser
parser = argparse.ArgumentParser(description=__doc__[1:])
# version option
parser.add_argument("--version", action="version", version=__version__,
help="Prints version information.")
# workflow options
parser.add_argument("--workflow-name", default="my_unamed_inference_run",
help="Name of the workflow to append in various places.")
parser.add_argument("--tags", nargs="+", default=[],
help="Tags to apend in various places.")
parser.add_argument("--output-dir", default=None,
help="Path to output directory.")
parser.add_argument("--output-map", default=None,
help="Path to output map file.")
parser.add_argument("--transformation-catalog", default=None,
help="Path to transformation catalog file.")
parser.add_argument("--output-file", default=None,
help="Path to DAX file.")
# input workflow file options
parser.add_argument("--bank-file", default=None,
help="HDF format template bank file.")
parser.add_argument("--statmap-file", default=None,
help="HDF format clustered coincident trigger "
"result file.")
parser.add_argument("--single-detector-triggers", nargs="+", default=None,
action=MultiDetOptionAction,
help="HDF format merged single detector trigger files.")
# analysis time option
# only required if not using input workflow file options
parser.add_argument("--gps-end-time", type=float, nargs="+", default=None,
help="Times to analyze. If given workflow files then "
"this option is ignored.")
# input configuration file options
parser.add_argument("--inference-config-file", type=str, nargs='+', required=True,
help="workflow.WorkflowConfigParser parsable file with "
"prior information.")
parser.add_argument("--prior-section", type=str, default="prior",
help="Name of the section in inference configuration file "
"that contains priors.")
# input frame files
parser.add_argument("--frame-files", nargs="+", default=None,
action=MultiDetOptionAppendAction,
help="GWF frame files to use.")
# add option groups
configuration.add_workflow_command_line_group(parser)
# parser command line
opts = parser.parse_args()
# make data output directory
core.makedir(opts.output_dir)
# the file we'll store all output files in
filedir = 'output'
# log to terminal until we know where the path to log output file
log_format = "%(asctime)s:%(levelname)s : %(message)s"
logging.basicConfig(format=log_format, level=logging.INFO)
# create workflow and sub-workflows
container = core.Workflow(opts, opts.workflow_name)
workflow = core.Workflow(opts, opts.workflow_name + "-main")
finalize_workflow = core.Workflow(opts, opts.workflow_name + "-finalization")
# change working directory to the output
os.chdir(opts.output_dir)
# sections for output HTML pages
rdir = layout.SectionNumber("results",
["detector_sensitivity", "priors", "posteriors",
"samples", "workflow"])
# make results directories
core.makedir(rdir.base)
core.makedir(rdir["workflow"])
# create files for workflow log
log_file_txt = core.File(workflow.ifos, "workflow-log", workflow.analysis_time,
extension=".txt", directory=rdir["workflow"])
log_file_html = core.File(workflow.ifos, "WORKFLOW-LOG", workflow.analysis_time,
extension=".html", directory=rdir["workflow"])
# switch saving log to file
logging.basicConfig(format=log_format, level=logging.INFO,
filename=log_file_txt.storage_path, filemode="w")
log_file = logging.FileHandler(filename=log_file_txt.storage_path, mode="w")
log_file.setLevel(logging.INFO)
formatter = logging.Formatter(log_format)
log_file.setFormatter(formatter)
logging.getLogger("").addHandler(log_file)
logging.info("Created log file %s" % log_file_txt.storage_path)
# sanity check that user is either using workflow or command line
workflow_options = opts.bank_file != None \
and opts.statmap_file != None \
and opts.single_detector_triggers != None
if not workflow_options and not (opts.gps_end_time != None):
raise ValueError("Must use either workflow options or --gps-end-time")
# if using workflow files to find analysis times
if workflow_options:
# typecast str from command line to File instances
tmpltbank_file = to_file(opts.bank_file)
coinc_file = to_file(opts.statmap_file)
single_triggers = []
for ifo in opts.single_detector_triggers:
fname = opts.single_detector_triggers[ifo]
single_triggers.append(to_file(fname, ifo=ifo))
# get number of events to analyze
num_events = int(workflow.cp.get_opt_tags("workflow-inference",
"num-events", ""))
# get detection statistic from statmap file
# if less events that requested to analyze in config file
# then analyze all events
f = h5py.File(opts.statmap_file, "r")
stat = f["foreground/stat"][:]
if len(stat) < num_events:
num_events = len(stat)
# get index for sorted detection statistic
sorting = stat.argsort()[::-1]
# get times for loudest events
times = {
f.attrs["detector_1"]: f["foreground/time1"][:][sorting][0:num_events],
f.attrs["detector_2"]: f["foreground/time2"][:][sorting][0:num_events],
}
# else get analysis times from command line
else:
times = dict([(ifo,opts.gps_end_time) for ifo in workflow.ifos])
num_events = len(opts.gps_end_time)
# construct Executable for running sampler
inference_exe = core.Executable(workflow.cp, "inference",
ifos=workflow.ifos, out_dir=filedir)
# read inference configuration file
# typecast str from command line to File instances
cp = configuration.WorkflowConfigParser(opts.inference_config_file)
with open('inference.ini', 'w') as ff:
cp.write(ff)
config_file = to_file('inference.ini')
# get channel names
channel_names = {}
for ifo in workflow.ifos:
channel_names[ifo] = workflow.cp.get_opt_tags(
"workflow", "%s-channel-name" % ifo.lower(), "")
channel_names_str = " ".join([key + ":" + val for key, val in \
channel_names.items()])
# figure out what parameters user wants to plot from workflow configuration
plot_parameters = {}
for option in workflow.cp.options("workflow-inference"):
if option.startswith("plot-1d-"):
group = option.replace("plot-1d-", "").replace("-", "_")
plot_parameters[group] = workflow.cp.get_opt_tag(
"workflow", option, "inference").split(" ")
all_parameters = [param for group in plot_parameters.values()
for param in group]
# loop over number of loudest events to be analyzed
for num_event in range(num_events):
# set GPS times for reading in data around the event
avg_end_time = sum([end_times[num_event] \
for end_times in times.values()]) / len(times.keys())
seconds_before_time = int(workflow.cp.get_opt_tags(
"workflow-inference",
"data-seconds-before-trigger", ""))
seconds_after_time = int(workflow.cp.get_opt_tags(
"workflow-inference",
"data-seconds-after-trigger", ""))
gps_start_time = int(avg_end_time) - seconds_before_time
gps_end_time = int(avg_end_time) + seconds_after_time
# get dict of segments for each IFO
seg_dict = {ifo : segments.segmentlist([segments.segment(
gps_start_time, gps_end_time)]) for ifo in workflow.ifos}
# get frame files from command line or datafind server
frame_files = core.FileList([])
if opts.frame_files:
for ifo in workflow.ifos:
for path in opts.frame_files[ifo]:
frame_file = core.File(
ifo, "FRAME",
segments.segment(gps_start_time,
gps_end_time),
file_url="file://" + path)
frame_file.PFN(frame_file.storage_path, site="local")
frame_files.append(frame_file)
else:
frame_files, _, _, _ = datafind.setup_datafind_workflow(workflow, seg_dict,
"datafind")
# make node for running sampler
node = inference_exe.create_node()
node.add_opt("--instruments", " ".join(workflow.ifos))
node.add_opt("--gps-start-time", gps_start_time)
node.add_opt("--gps-end-time", gps_end_time)
if len(frame_files):
node.add_multiifo_input_list_opt("--frame-files", frame_files)
node.add_opt("--channel-name", channel_names_str)
node.add_input_opt("--config-file", config_file)
analysis_time = segments.segment(gps_start_time, gps_end_time)
inference_file = node.new_output_file_opt(analysis_time, ".hdf",
"--output-file",
tags=[str(num_event)])
# add node to workflow
workflow += node
# files that prints information about the search event
if workflow_options:
coinc_table_files = [mini.make_coinc_info(workflow, single_triggers,
tmpltbank_file, coinc_file,
num_event, rdir.base,
tags=opts.tags + [str(num_event)])]
# files for posteriors summary subsection
base = "posteriors"
post_table_files = inffu.make_inference_summary_table(
workflow, inference_file, rdir[base],
variable_params=all_parameters,
analysis_seg=analysis_time,
tags=opts.tags + [str(num_event)])
post_files = inffu.make_inference_posterior_plot(
workflow, inference_file, rdir[base],
parameters=all_parameters,
analysis_seg=analysis_time,
tags=opts.tags + [str(num_event)])
layout.single_layout(rdir[base], post_table_files + post_files)
# files for posteriors 1d_posteriors subsection
groups = plot_parameters.keys()
groups.sort()
for group in groups:
parameters = plot_parameters[group]
base = "posteriors/1d_{}_posteriors".format(group)
post_1d_files = inffu.make_inference_1d_posterior_plots(
workflow, inference_file, rdir[base],
parameters=parameters,
analysis_seg=analysis_time,
tags=opts.tags + [str(num_event), group])
layout.group_layout(rdir[base], post_1d_files)
if "inference_samples" in workflow.cp.options("executables"):
# files for samples summary subsection
base = "samples"
samples_files = []
for parameter in all_parameters:
samples_files += inffu.make_inference_samples_plot(
workflow, inference_file, rdir[base],
parameters=[parameter], analysis_seg=analysis_time,
tags=opts.tags + [str(num_event), parameter])
layout.group_layout(rdir[base], samples_files)
if "inference_rate" in workflow.cp.options("executables"):
# files for samples acceptance_rate subsection
base = "samples/acceptance_rate"
acceptance_files = inffu.make_inference_acceptance_rate_plot(
workflow, inference_file, rdir[base],
analysis_seg=analysis_time,
tags=opts.tags + [str(num_event)])
layout.single_layout(rdir[base], acceptance_files)
# files for detector_sensitivity summary subsection
base = "detector_sensitivity"
psd_file = plotting.make_spectrum_plot(workflow, [inference_file], rdir[base],
tags=opts.tags + [str(num_event)],
hdf_group="data")
layout.single_layout(rdir[base], [psd_file])
# files for priors summary section
base = "priors"
prior_files = []
for subsection in cp.get_subsections(opts.prior_section):
prior_files += inffu.make_inference_prior_plot(
workflow, config_file, rdir[base],
analysis_seg=workflow.analysis_time,
parameters=[subsection],
tags=opts.tags + [subsection])
layout.group_layout(rdir[base], prior_files)
# files for overall summary section
summ_files = coinc_table_files if workflow_options else []
summ_files = post_table_files + post_files
for summ_file in summ_files:
symlink_path(summ_file, rdir.base)
layout.single_layout(rdir.base, summ_files)
# create versioning HTML pages
results.create_versioning_page(rdir["workflow/version"], container.cp)
# create node for making HTML pages
plotting.make_results_web_page(finalize_workflow,
os.path.join(os.getcwd(), rdir.base))
# add sub-workflows to workflow
container += workflow
container += finalize_workflow
# make finalize sub-workflow depend on main sub-workflow
dep = dax.Dependency(parent=workflow.as_job, child=finalize_workflow.as_job)
container._adag.addDependency(dep)
# write dax
outputfile = "{}.dax".format(opts.workflow_name) if opts.output_file == None else opts.output_file
container.save(filename=outputfile, output_map_path=opts.output_map,
transformation_catalog_path=opts.transformation_catalog)
# save workflow configuration file
base = rdir["workflow/configuration"]
core.makedir(base)
wf_ini = workflow.save_config("workflow.ini", base, container.cp)
layout.single_layout(base, wf_ini)
# save prior configuration file
base = rdir["priors/configuration"]
core.makedir(base)
prior_ini = workflow.save_config("priors.ini", base, cp)
layout.single_layout(base, prior_ini)
# close the log and flush to the html file
logging.shutdown()
with open (log_file_txt.storage_path, "r") as log_file:
log_data = log_file.read()
log_str = """
<p>Workflow generation script created workflow in output directory: %s</p>
<p>Workflow name is: %s</p>
<p>Workflow generation script run on host: %s</p>
<pre>%s</pre>
""" % (os.getcwd(), opts.workflow_name, socket.gethostname(), log_data)
kwds = {"title" : "Workflow Generation Log",
"caption" : "Log of the workflow script %s" % sys.argv[0],
"cmd" : " ".join(sys.argv)}
results.save_fig_with_metadata(log_str, log_file_html.storage_path, **kwds)
layout.single_layout(rdir["workflow"], ([log_file_html]))
![swh spinner](/static/img/swh-spinner.gif)
Computing file changes ...