Merge pull request #40 from pyrocko/v1.2.0-dev

Merge development branches

CHANGELOG.md

@@ -4,6 +4,21 @@ All notable changes to Grond will be documented in this file.
 
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 
+## Unreleased
+
+### Added
+- Added Covariance weighting from `SatelliteTarget` and `GNSSTarget`.
+- Added new Volume source (VLVD).
+- Added SVG as `grond plot` export format.
+
+### Changed
+- Consistent utilisation of `ModelHistory` for plots and results.
+- `MTLocationPlot` supports Gamma scaling (misfit^gamma).
+- `SatelliteTarget` speeds up bootstrapping by multi-threading.
+- `Envronment` can be initialised from `ProblemConfig`.
+- Improved: `grondown` downloading seismic wave forms.
+
+
 ## [1.2.0] 2019-02-19
 
 ### Added

setup.py

@@ -7,7 +7,7 @@
 from setuptools.command.install import install
 from setuptools.command.build_py import build_py
 
-version = '1.2.0'
+version = '1.3.0'
 
 
 class CustomInstallCommand(install):
@@ -111,6 +111,7 @@ def run(self):
         'grond.problems.double_dc',
         'grond.problems.rectangular',
         'grond.problems.volume_point',
+        'grond.problems.vlvd',
         'grond.optimisers',
         'grond.optimisers.highscore',
         'grond.analysers',

src/apps/grond.py

@@ -733,7 +733,11 @@ def setup(parser):
         parser.add_option(
             '--parallel', dest='nparallel', type=int, default=1,
             help='set number of events to process in parallel, '
-                 'If set to more than one, --status=quiet is implied.')
+                 'if set to more than one, --status=quiet is implied.')
+        parser.add_option(
+            '--threads', dest='nthreads', type=int, default=0,
+            help='set number of threads per process. This defaults to 0,'
+                 'which is using all available cores.')
 
     parser, options, args = cl_parse('go', args, setup)
 
@@ -749,7 +753,8 @@ def setup(parser):
             force=options.force,
             preserve=options.preserve,
             status=status,
-            nparallel=options.nparallel)
+            nparallel=options.nparallel,
+            nthreads=options.nthreads)
         if len(env.get_selected_event_names()) == 1:
             logger.info(CLIHints(
                 'go', rundir=env.get_rundir_path()))

src/core.py

@@ -141,6 +141,7 @@ def forward(rundir_or_config_path, event_names):
 def harvest(rundir, problem=None, nbest=10, force=False, weed=0):
 
     env = Environment([rundir])
+    optimiser = env.get_optimiser()
     nchains = env.get_optimiser().nchains
 
     if problem is None:
@@ -152,7 +153,6 @@ def harvest(rundir, problem=None, nbest=10, force=False, weed=0):
 
     logger.info('Harvesting problem "%s"...' % problem.name)
 
-    optimiser = load_optimiser_info(rundir)
     dumpdir = op.join(rundir, 'harvest')
     if op.exists(dumpdir):
         if force:
@@ -164,7 +164,7 @@ def harvest(rundir, problem=None, nbest=10, force=False, weed=0):
 
     ibests_list = []
     ibests = []
-    gms = problem.combine_misfits(misfits)
+    gms = bootstrap_misfits[:, 0]
     isort = num.argsort(gms)
 
     ibests_list.append(isort[:nbest])
@@ -442,10 +442,10 @@ def check(
 
 def go(environment,
        force=False, preserve=False,
-       nparallel=1, status='state'):
+       nparallel=1, status='state', nthreads=0):
 
     g_data = (environment, force, preserve,
-              status, nparallel)
+              status, nparallel, nthreads)
     g_state[id(g_data)] = g_data
 
     nevents = environment.nevents_selected
@@ -460,7 +460,7 @@ def go(environment,
 
 def process_event(ievent, g_data_id):
 
-    environment, force, preserve, status, nparallel = \
+    environment, force, preserve, status, nparallel, nthreads = \
         g_state[g_data_id]
 
     config = environment.get_config()
@@ -513,6 +513,8 @@ def process_event(ievent, g_data_id):
     config.change_basepath(basepath)
 
     optimiser = config.optimiser_config.get_optimiser()
+    optimiser.set_nthreads(nthreads)
+
     optimiser.init_bootstraps(problem)
     problem.dump_problem_info(rundir)
 
@@ -582,16 +584,15 @@ class ResultStats(Object):
     parameter_stats_list = List.T(ParameterStats.T())
 
 
-def make_stats(problem, xs, misfits, pnames=None):
-    gms = problem.combine_misfits(misfits)
+def make_stats(problem, models, gms, pnames=None):
     ibest = num.argmin(gms)
     rs = ResultStats(problem=problem)
     if pnames is None:
         pnames = problem.parameter_names
 
     for pname in pnames:
         iparam = problem.name_to_index(pname)
-        vs = problem.extract(xs, iparam)
+        vs = problem.extract(models, iparam)
         mi, p5, p16, median, p84, p95, ma = map(float, num.percentile(
             vs, [0., 5., 16., 50., 84., 95., 100.]))
 
@@ -676,9 +677,12 @@ def dump(x, gm, indices):
 
     header = None
     for rundir in rundirs:
-        problem, xs, misfits, bootstrap_misfits, _ = \
-            load_problem_info_and_data(
-                rundir, subset='harvest')
+        env = Environment(rundir)
+        history = env.get_history(subset='harvest')
+
+        problem = history.problem
+        models = history.models
+        misfits = history.get_primary_chain_misfits()
 
         if type == 'vector':
             pnames_take = pnames_clean or \
@@ -703,21 +707,25 @@ def dump(x, gm, indices):
             indices = None
 
         if what == 'best':
-            x_best, gm_best = stats.get_best_x_and_gm(problem, xs, misfits)
+            x_best = history.get_best_model()
+            gm_best = history.get_best_misfit()
             dump(x_best, gm_best, indices)
 
         elif what == 'mean':
-            x_mean, gm_mean = stats.get_mean_x_and_gm(problem, xs, misfits)
+            x_mean = history.get_mean_model()
+            gm_mean = history.get_mean_misfit(chain=0)
             dump(x_mean, gm_mean, indices)
 
         elif what == 'ensemble':
-            gms = problem.combine_misfits(misfits)
+            gms = misfits
             isort = num.argsort(gms)
             for i in isort:
-                dump(xs[i], gms[i], indices)
+                dump(models[i], gms[i], indices)
 
         elif what == 'stats':
-            rs = make_stats(problem, xs, misfits, pnames_clean)
+            rs = make_stats(problem, models,
+                            history.get_primary_chain_misfits(),
+                            pnames_clean)
             if shortform:
                 print(' ', format_stats(rs, pnames), file=out)
             else: