Datasets:
benchflow
/
frontierphysics-pr139-evidence

Dataset card Data Studio Files
xet
 Community
text
string
::test_arrays_are_well_formed_and_on_the_supplied_grid PASSED [ 50%]
::test_reported_forward_histories_match_independent_solver FAILED [ 60%]
::test_blind_sensor_residuals_are_at_the_noise_floor FAILED [ 70%]
::test_flux_histories_match_withheld_injections PASSED [ 80%]
::test_flux_energy_and_peak_are_recovered_and_self_consistent PASSED [ 90%]
::test_summary_is_consistent PASSED [100%]
=================================== FAILURES ===================================
____________ test_calibration_within_declared_bounds_and_recovered _____________
def test_calibration_within_declared_bounds_and_recovered():
calibration = reported_calibration()
vector = np.array(
[
calibration["conductivity_scale"],
calibration["depth_offset_m"],
*calibration["sensor_time_constants_s"],
],
dtype=float,
)
bounds = CONFIG["calibration_bounds"]
lower = np.array(
[
bounds["conductivity_scale"][0],
bounds["depth_offset_m"][0],
bounds["sensor_time_constants_s"][0][0],
bounds["sensor_time_constants_s"][1][0],
]
)
upper = np.array(
[
bounds["conductivity_scale"][1],
bounds["depth_offset_m"][1],
bounds["sensor_time_constants_s"][0][1],
bounds["sensor_time_constants_s"][1][1],
]
)
assert np.all(vector >= lower) and np.all(vector <= upper)
assert abs(vector[0] - CAL_TRUE[0]) <= 0.05, "conductivity scale not calibrated"
assert abs(vector[1] - CAL_TRUE[1]) <= 5.0e-5, "sensor depth offset not calibrated"
> assert abs(vector[2] - CAL_TRUE[2]) <= 0.018, "near-probe time constant not calibrated"
E AssertionError: near-probe time constant not calibrated
E assert 0.026175878027679858 <= 0.018
E + where 0.026175878027679858 = abs((0.09817587802767985 - 0.072))
/verifier/test_outputs.py:161: AssertionError
_______________ test_calibration_reproduces_all_known_flux_shots _______________
def test_calibration_reproduces_all_known_flux_shots():
calibration = reported_calibration()
t_s = CALIBRATION_DATA["t_s"].astype(float)
measured = CALIBRATION_DATA["temperature_K"].astype(float)
known_flux = CALIBRATION_DATA["known_flux_W_m2"].astype(float)
noise = CALIBRATION_DATA["noise_std_K"].astype(float)
predictions = []
for q in known_flux:
prediction, _ = simulate(t_s, q, CONFIG, calibration, nx=83, cfl=0.31)
predictions.append(prediction)
predictions = np.asarray(predictions)
rms = float(np.sqrt(np.mean((predictions - measured) ** 2)))
assert rms <= CALIBRATION_RMS_RATIO_MAX * float(np.mean(noise))
> assert abs(float(calibration["calibration_rms_K"]) - rms) <= 0.75 * float(
np.mean(noise)
)
E assert 0.0482322620574334 <= (0.75 * 0.0575)
E + where 0.0482322620574334 = abs((0.07289191383040161 - 0.12112417588783501))
E + where 0.07289191383040161 = float(0.07289191383040161)
E + and 0.0575 = float(0.0575)
E + where 0.0575 = <function mean at 0x7c775158a030>(array([0.05 , 0.065]))
E + where <function mean at 0x7c775158a030> = np.mean
/verifier/test_outputs.py:178: AssertionError
___________ test_reported_forward_histories_match_independent_solver ___________
def test_reported_forward_histories_match_independent_solver():
for index, shot_id in enumerate(SHOT_IDS):
_, _, _, surface, predicted = shot_entry(shot_id)
reference_sensor, reference_surface = reference_forward(index)
> assert np.sqrt(np.mean((predicted - reference_sensor) ** 2)) <= 0.12, (
f"{shot_id}: predicted_sensor_K is not from the reported model and flux"
)
E AssertionError: blind_01: predicted_sensor_K is not from the reported model and flux
E assert 0.13865697392003495 <= 0.12
E + where 0.13865697392003495 = <ufunc 'sqrt'>(0.01922575641666125)
E + where <ufunc 'sqrt'> = np.sqrt
E + and 0.01922575641666125 = <function mean at 0x7c775158a030>(((array([[293.15 , 293.15004079, 293.1501771 , 293.15045331,\n 293.15089747, 293.15152417, 293.15233897, 293.15334207,\n 293.15453073, 293.15590073, 293.15751029, 293.15961827,\n 293.16267989, 293.16725864, 293.173...
E + where <function mean at 0x7c775158a030> = np.mean
/verifier/test_outputs.py:204: AssertionError
______________ test_blind_sensor_residuals_are_at_the_noise_floor ______________
def test_blind_sensor_residuals_are_at_the_noise_floor():
measured = BLIND_DATA["temperature_K"].astype(float)
for index, shot_id in enumerate(SHOT_IDS):
entry, _, _, _, _ = shot_entry(shot_id)
prediction, _ = reference_forward(index)
rms = float(np.sqrt(np.mean((prediction - measured[index]) ** 2)))
ratio = rms / float(np.mean(NOISE))
assert FORWARD_RMS_RATIO[0] <= ratio <= FORWARD_RMS_RATIO[1], (
f"{shot_id}: sensor residual/noise ratio {ratio:.3f} is outside "