def test_rank_models():
RE = RunEngine()
# Create accurate fit
motor = SynAxis(name='motor')
det = SynSignal(name='centroid',
func=lambda: 5 * motor.read()['motor']['value'] + 2)
fit1 = LinearFit('centroid', 'motor', update_every=None, name='Accurate')
RE(scan([det], motor, -1, 1, 50), fit1)
# Create inaccurate fit
det2 = SynSignal(name='centroid',
func=lambda: 25 * motor.read()['motor']['value'] + 2)
fit2 = LinearFit('centroid', 'motor', update_every=None, name='Inaccurate')
RE(scan([det2], motor, -1, 1, 50), fit2)
# Create inaccurate fit
det3 = SynSignal(name='centroid',
func=lambda: 12 * motor.read()['motor']['value'] + 2)
fit3 = LinearFit('centroid',
'motor',
update_every=None,
name='Midly Inaccurate')
RE(scan([det3], motor, -1, 1, 50), fit3)
# Rank models
ranking = rank_models([fit2, fit1, fit3], target=22, x=4)
assert ranking[0] == fit1
assert ranking[1] == fit3
assert ranking[2] == fit2
def test_walk_to_pixel(RE, one_bounce_system):
logger.debug("test_walk_to_pixel")
_, mot, det = one_bounce_system
##########################
# Test on simple devices #
##########################
simple_motor = SynAxis(name='motor')
simple_det = SynSignal(name='det',
func=lambda: 5*simple_motor.read()['motor']['value']
+ 2)
# Naive step
plan = run_wrapper(walk_to_pixel(simple_det, simple_motor, 200, 0,
first_step=1e-6,
tolerance=10, average=None,
target_fields=['det', 'motor'],
max_steps=3))
RE(plan)
assert np.isclose(simple_det.read()['det']['value'], 200, atol=1)
simple_motor.set(0.)
# Gradient
simple_motor = SynAxis(name='motor')
simple_det = SynSignal(name='det',
func=lambda: 5*simple_motor.read()['motor']['value']
+ 2)
plan = run_wrapper(walk_to_pixel(simple_det, simple_motor, 200, 0,
gradient=1.6842e+06,
tolerance=10, average=None,
target_fields=['det', 'motor'],
max_steps=3))
RE(plan)
assert np.isclose(simple_det.read()['det']['value'], 200, atol=1)
##########################
# Test on full model #
##########################
# Naive step
cent = 'detector_stats2_centroid_x'
plan = run_wrapper(walk_to_pixel(det, mot, 200, 0, first_step=1e-6,
tolerance=10, average=None,
target_fields=[cent, 'sim_alpha'],
max_steps=3))
RE(plan)
assert np.isclose(det.read()[det.name + "_" + cent]['value'], 200, atol=1)
mot.set(0.)
# Gradient
plan = run_wrapper(walk_to_pixel(det, mot, 200, 0, gradient=1.6842e+06,
tolerance=10, average=None,
target_fields=[cent, 'sim_alpha'],
max_steps=3))
RE(plan)
assert np.isclose(det.read()[det.name + "_" + cent]['value'], 200, atol=10)
def test_multi_fit():
RE = RunEngine()
# Expected values of fit
expected = {'x0': 5, 'x1': 4, 'x2': 3}
m1 = SynAxis(name='m1')
m2 = SynAxis(name='m2')
det = SynSignal(name='centroid',
func=lambda: 5 + 4 * m1.read()['m1']['value'] + 3 * m2.
read()['m2']['value'])
# Assemble fitting callback
cb = MultiPitchFit('centroid', ('m1', 'm2'), update_every=None)
RE(outer_product_scan([det], m1, -1, 1, 10, m2, -1, 1, 10, False), cb)
# Check accuracy of fit
logger.debug(cb.result.fit_report())
for k, v in expected.items():
assert np.allclose(cb.result.values[k], v, atol=1e-6)
# Check we create an accurate estimate
assert np.allclose(cb.eval(a0=5, a1=10), 55, atol=1e-5)
assert np.allclose(cb.backsolve(55, a1=10)['a0'], 5, atol=1e-5)
assert np.allclose(cb.backsolve(55, a0=5)['a1'], 10, atol=1e-5)
def test_linear_fit():
RE = RunEngine()
# Expected values of fit
expected = {'slope': 5, 'intercept': 2}
motor = SynAxis(name='motor')
det = SynSignal(name='centroid',
func=lambda: 5 * motor.read()['motor']['value'] + 2)
# Assemble fitting callback
cb = LinearFit('centroid', 'motor', update_every=None)
RE(scan([det], motor, -1, 1, 50), cb)
# Check accuracy of fit
for k, v in expected.items():
assert np.allclose(cb.result.values[k], v, atol=1e-6)
# Check we create an accurate estimate
assert np.allclose(cb.eval(x=10), 52, atol=1e-5)
assert np.allclose(cb.eval(motor=10), 52, atol=1e-5)
assert np.allclose(cb.backsolve(52)['x'], 10, atol=1e-5)