def check_restarted_in_middle(restart_outdir, restart_point=20):
last_no_soln = "soln%i.dat" % (restart_point - 1)
first_soln = "soln%i.dat" % restart_point
ok = (not os.path.isfile(pjoin(restart_outdir, last_no_soln))) \
and (os.path.isfile(pjoin(restart_outdir, first_soln)))
if ok:
mm.okprint("Restarted at correct point in", restart_outdir)
else:
mm.badprint("Failed or restarted at wrong point in", restart_outdir)
return ok
def check_ndt_less_than(data, max_ndt, identifier=None):
if identifier is None:
identifier = _default_label(data)
ndt = len(data['dts'])
passed = ndt < max_ndt
if not passed:
mm.badprint("FAILED in ", identifier)
mm.badprint("n steps is", ndt, "which is more than", max_ndt)
else:
mm.okprint("n steps", ndt, "ok in", identifier)
return passed
def check_m_length(data, tol=1e-8, identifier=None):
"""Check that abs(|m| - 1) < tol."""
if identifier is None:
identifier = _default_label(data)
max_length_err = max(map(abs, data['m_length_error_means']))
length_test = max_length_err < tol
if not length_test:
mm.badprint("FAILED in ", identifier,
"with |m| error of", max_length_err)
else:
mm.okprint("|m| max of", max_length_err, "ok in", identifier)
return length_test
def check_error_norm(data, tol=1e-5, identifier=None):
"""Check that max error norm < tol."""
if identifier is None:
identifier = _default_label(data)
max_err_norm = max(map(abs, data['error_norms']))
norm_test = max_err_norm < tol
if not norm_test:
mm.badprint("FAILED in", identifier)
mm.badprint("with max error norm of", max_err_norm)
else:
mm.okprint("max error norm of", max_err_norm, "ok in", identifier)
return norm_test
def check_solns_match(main_dir="Validation", compare_dir="validata", **kwargs):
matches = []
for fname in os.listdir(main_dir):
if ("soln" in fname) and (".dat" in fname):
match, _, _ = fpdiff(pjoin(main_dir, fname),
pjoin(compare_dir, fname), **kwargs)
matches.append(match)
if len(matches) == 0:
mm.badprint("No files in", main_dir)
return False
ok = all(matches)
if ok:
mm.okprint("Files match in", main_dir, compare_dir)
else:
mm.badprint("Files don't match in", main_dir, compare_dir)
return ok
def try_run(args, dt, dir_naming_args, root_outdir,
maxallowederror = 0.01, maxallowedangle = sp.pi/4):
args.update({'-dt' : dt})
# Try to run it
err_code, outdir = mm._run(args, root_outdir, dir_naming_args,
quiet=True)
# Parse output files
data = mm.parse_run(outdir)
if data is not None:
errs = data['m_length_error_means']
assert errs[0] >= 0
maxerror = max(errs)
angles = data['max_angle_errors']
maxangle = max(angles)
maxtime = data['times'][-1]
nsteps = len(data['times'])
else:
maxerror = sp.inf
maxangle = sp.inf
maxtime = 0
nsteps = 0
success = (err_code == 0 and maxerror < maxallowederror
and maxangle < maxallowedangle)
if success:
mm.okprint("Succedded in", data['-outdir'])
else:
mm.badprint(pjoin(data['-outdir'], "stdout:1:1:"), "FAILED")
return success, data
def check_error_norm_relative(data, tol=1e-5, identifier=None):
"""Check that max error norm < tol * maximum_trace_value."""
if identifier is None:
identifier = _default_label(data)
max_err_norm = max(map(abs, data['error_norms']))
# Get biggest value from list of list of trace values
max_value = max(it.chain(*data['trace_values']))
# If max < 1 then use 1.0 as multiplier
norm_test = max_err_norm < tol * max(max_value, 1.0)
if not norm_test:
mm.badprint("FAILED in", identifier)
mm.badprint("with max error norm of", max_err_norm)
else:
mm.okprint("max error norm of", max_err_norm, "ok in", identifier)
return norm_test
def check_convergence(datasets, expected_rate, tol=0.2):
"""For a set of convergence test check that the rate of convergence as dt
and h -> 0 is as expected. dt and h are linked so it is enough to
test only for dt.
"""
rate = mm.convergence_rate(datasets, error_norm_norm=max)
# Check if near to expected value
ok = abs(rate - expected_rate) < tol
# Print messages, identify by first dataset for simplicity.
dataid = _default_label(datasets[0])
if ok:
mm.okprint("convergence rate", rate, "ok in", dataid)
else:
mm.badprint("convergence rate", rate,
"but expected", expected_rate, "+/-", tol,
"in", dataid)
# ??ds check goodness of fit?
return ok
def check_mean_m_matches(data1, data2, tol=1e-4):
id1 = _default_label(data1)
id2 = _default_label(data2)
for a, b in zip(data1['mean_mxs'], data2['mean_mxs']):
if abs(a - b) > tol:
mm.badprint("Failed mx comparison in", id1, id2, "error of", abs(a - b))
return False
for a, b in zip(data1['mean_mys'], data2['mean_mys']):
if abs(a - b) > tol:
mm.badprint("Failed my comparison in", id1, id2, "error of", abs(a - b))
return False
for a, b in zip(data1['mean_mzs'], data2['mean_mzs']):
if abs(a - b) > tol:
mm.badprint("Failed mz comparison in", id1, id2, "error of", abs(a - b))
return False
mm.okprint("m comparison ok in", id1, id2)
return True
def check_dicts_match(data, exact, keys, **kwargs):
"""Given two dicts and a list of keys check that those values match.
??ds actually temp hack: use the second value of the array given by
data['k']...
"""
def fpcheck(v1, v2, rtol=1e-8, zero=1e-14):
"""Check two floats are the same: if both small then true, otherwise check
relative difference.
"""
if v1 < zero and v2 < zero:
return True
else:
return (abs((v1 - v2)/v1) < rtol)
dataid = _default_label(data)
ok = []
for k in keys:
try:
a = fpcheck(exact[k], data[k][1], **kwargs)
ok.append(a)
# Print messages
if a:
mm.okprint(k, "comparison ok in", dataid)
else:
mm.badprint("Failed", k, "comparison with values of",
data[k][1], exact[k], "in", dataid)
except KeyError:
mm.badprint("Failed: missing key", k, "in", dataid)
ok.append(False)
return all(ok)
