def run(self, commands):
self.setup_files["input.deck"] = util.decks.input_deck(solver=self.solver,
num_cells_x=36, num_cells_y=74, a_x=-1.55, b_x=1.45, a_y=-1.62, b_y=1.38,
t_final=1.03, sigma=1.111, init_cond = self.initcond,
gaussian_sigma=self.gaussian_sigma, num_mpi_partitions_x=self.xNodes,
num_mpi_partitions_y=self.yNodes)
with util.ResetFile(*(self.initial_paths + self.current_paths)):
util.right(self.log_path, "execution time:")
util.tee(self.log_path, " " * 10, wait=True)
super(RegressionTest, self).run(commands)
util.tee(self.log_path, "")
self.initial = util.formats.Composite(self.parser,
self.initial_path, self.xNodes, self.yNodes)
self.current = util.formats.Composite(self.parser,
self.current_path, self.xNodes, self.yNodes)
self.reference = self.parser(self.reference_path)
self.show_results()
return self.current
def convergence_test(target, source, env):
try:
util.tee(str(target[0]), "convergence tests:")
for s in SOLVER:
if s == "kinetic":
util.bullet(str(target[0]), "%s" % s, wait=False)
KineticConvergenceTest(str(target[0])).run([os.path.abspath(str(source[0]))], 5)
elif s == "moment":
for f in MOMENT_FILTER:
util.bullet(str(target[0]), "%s(%s)" % (s, f), wait=False)
MomentConvergenceTest(str(target[0]), f).run([os.path.abspath(str(source[0]))], 5)
elif s == "momopt":
for t in MOMOPT_TYPE:
util.bullet(str(target[0]), "%s(%s)" % (s, t), wait=False)
MomoptConvergenceTest(str(target[0]), t).run([os.path.abspath(str(source[0]))], 5)
except:
try:
os.unlink(str(target[0]))
except:
pass
raise
def regression_test(target, source, env):
try:
util.tee(str(target[0]), "regression tests:")
for p in [str(x) for x in source]:
if p.startswith("solver_mpi"):
commands = [["mpirun", "-np", "4", p]]
composite = True
else:
commands = [os.path.abspath(p)]
composite = False
for s in SOLVER:
for i in INITCOND:
if s == "kinetic":
util.bullet(str(target[0]), "%s,%s,%s" % (p, s, i), wait=False)
if composite:
KineticCompositeRegressionTest(str(target[0]), i).run(commands)
else:
KineticRegressionTest(str(target[0]), i).run(commands)
elif s == "moment":
for f in MOMENT_FILTER:
util.bullet(str(target[0]), "%s,%s,%s(%s)" % (p, s, i, f), wait=False)
if composite:
MomentCompositeRegressionTest(str(target[0]), f, i).run(commands)
else:
MomentRegressionTest(str(target[0]), f, i).run(commands)
elif s == "momopt":
for t in MOMOPT_TYPE:
util.bullet(str(target[0]), "%s,%s,%s(%s)" % (p, s, i, t), wait=False)
if composite:
MomoptCompositeRegressionTest(str(target[0]), t, i).run(commands)
else:
MomoptRegressionTest(str(target[0]), t, i).run(commands)
except:
try:
os.unlink(str(target[0]))
except:
pass
raise
def run(self, commands):
self.setup_files["input.deck"] = util.decks.input_deck(solver=self.solver,
num_cells_x=36, num_cells_y=74, a_x=-1.55, b_x=1.45, a_y=-1.62, b_y=1.38,
t_final=1.03, sigma=1.111, init_cond = self.initcond,
gaussian_sigma=self.gaussian_sigma)
with util.ResetFile(self.current_path, self.initial_path):
util.right(self.log_path, "execution time:")
util.tee(self.log_path, " " * 10, wait=True)
super(RegressionTest, self).run(commands)
util.tee(self.log_path, "")
self.initial = self.parser(self.initial_path)
self.current = self.parser(self.current_path)
if REGENERATE:
try:
os.makedirs(os.path.dirname(self.reference_path))
except OSError:
pass
shutil.copyfile(self.current_path, self.reference_path)
self.reference = self.parser(self.reference_path)
self.show_results()
return self.current
def run(self,input,**k):
c = self.config
cmd = '%s -l %g -u %g ' % (svm_scale_pathname, c.lower, c.upper)
if c.y_lower:
cmd += "-y %g %g " % (c.y_lower, c.y_upper)
if k['scale info input']:
cmd += '-r %s ' % k['scale info input'].pathname
filename = tempfile.mktemp()
sio = LibsvmScaleFile(filename,autodel=True)
cmd += '-s %s ' % filename
cmd += input.pathname
log = self.callback.log
log('# Running %s\n' % cmd)
t = util.tee()
util.run_command(cmd,t,log)
return {'output': LibsvmInputFile(t.get_filename(),autodel=True),
'scale info output': sio}
def run(self,**k):
input = k['training data']
if k['optional parameter']:
self.config.__dict__.update(k['optional parameter'].get_data())
filename = tempfile.mktemp()
cmd = self.make_command(input.pathname, filename)
log = self.callback.log
log('# Running %s\n' % cmd)
t = util.tee(log)
util.run_command(cmd,t,t)
if self.config.fold == 0:
m = LibsvmModelFile(filename,autodel=True)
else:
m = None # no model file while doing cv
return {'stdout/stderr': TextFile(t.get_filename(), autodel=True),
'model': m }
def run(self, input, **k):
c = self.config
cmd = '%s -l %g -u %g ' % (svm_scale_pathname, c.lower, c.upper)
if c.y_lower:
cmd += "-y %g %g " % (c.y_lower, c.y_upper)
if k['scale info input']:
cmd += '-r %s ' % k['scale info input'].pathname
filename = tempfile.mktemp()
sio = LibsvmScaleFile(filename, autodel=True)
cmd += '-s %s ' % filename
cmd += input.pathname
log = self.callback.log
log('# Running %s\n' % cmd)
t = util.tee()
util.run_command(cmd, t, log)
return {
'output': LibsvmInputFile(t.get_filename(), autodel=True),
'scale info output': sio
}
def run(self, **k):
input = k['training data']
if k['optional parameter']:
self.config.__dict__.update(k['optional parameter'].get_data())
filename = tempfile.mktemp()
cmd = self.make_command(input.pathname, filename)
log = self.callback.log
log('# Running %s\n' % cmd)
t = util.tee(log)
util.run_command(cmd, t, t)
if self.config.fold == 0:
m = LibsvmModelFile(filename, autodel=True)
else:
m = None # no model file while doing cv
return {
'stdout/stderr': TextFile(t.get_filename(), autodel=True),
'model': m
}
def run(self,**k):
input = k['training data']
out_filename = tempfile.mktemp()
png_filename = tempfile.mktemp()
c = self.config
cmd = "%s -svmtrain %s " % (grid_pathname, svm_train_pathname)
cmd += "-gnuplot %s -out %s " % (gnuplot_pathname, out_filename)
cmd += "-png %s " % png_filename
cmd += "-log2c %s,%s,%s " % (c.c_begin, c.c_end, c.c_step)
cmd += "-log2g %s,%s,%s " % (c.g_begin, c.g_end, c.g_step)
cmd += input.pathname
log = self.callback.log
log('# Running %s\n' % cmd)
t = util.tee(log)
util.run_command(cmd,t,t)
best_c, best_g, best_rate = map(float,open(t.get_filename()).readlines()[-1].split())
return {'contour': ImageFile(png_filename,autodel=True),
'result': TextFile(out_filename,autodel=True),
'stdout/stderr': TextFile(t.get_filename(),autodel=True),
'best parameter': ConfigData(cost=best_c,gamma=best_g)}