def test_reconstruct():
"""Test Tracker objects reconstruction"""
# test with one TrackerUD
with ExperimentController(*std_args, **std_kwargs) as ec:
tr = TrackerUD(ec, 1, 1, 3, 1, 5, np.inf, 3)
while not tr.stopped:
tr.respond(np.random.rand() < tr.x_current)
tracker = reconstruct_tracker(ec.data_fname)[0]
assert_true(tracker.stopped)
tracker.x_current
# test with one TrackerBinom
with ExperimentController(*std_args, **std_kwargs) as ec:
tr = TrackerBinom(ec, .05, .5, 10)
while not tr.stopped:
tr.respond(True)
tracker = reconstruct_tracker(ec.data_fname)[0]
assert_true(tracker.stopped)
tracker.x_current
# tracker not stopped
with ExperimentController(*std_args, **std_kwargs) as ec:
tr = TrackerUD(ec, 1, 1, 3, 1, 5, np.inf, 3)
tr.respond(np.random.rand() < tr.x_current)
assert_true(not tr.stopped)
assert_raises(ValueError, reconstruct_tracker, ec.data_fname)
# test with dealer
with ExperimentController(*std_args, **std_kwargs) as ec:
tr = [TrackerUD(ec, 1, 1, 3, 1, 5, np.inf, 3) for _ in range(3)]
td = TrackerDealer(ec, tr)
for _, x_current in td:
td.respond(np.random.rand() < x_current)
dealer = reconstruct_dealer(ec.data_fname)[0]
assert_true(all(td._x_history == dealer._x_history))
assert_true(all(td._tracker_history == dealer._tracker_history))
assert_true(all(td._response_history == dealer._response_history))
assert_true(td.shape == dealer.shape)
assert_true(td.trackers.shape == dealer.trackers.shape)
# no tracker/dealer in file
with ExperimentController(*std_args, **std_kwargs) as ec:
ec.identify_trial(ec_id='one', ttl_id=[0])
ec.start_stimulus()
ec.write_data_line('misc', 'trial one')
ec.stop()
ec.trial_ok()
ec.write_data_line('misc', 'end')
assert_raises(ValueError, reconstruct_tracker, ec.data_fname)
assert_raises(ValueError, reconstruct_dealer, ec.data_fname)
def test_tracker_ud():
"""Test TrackerUD"""
import matplotlib.pyplot as plt
tr = TrackerUD(callback, 3, 1, 1, 1, np.inf, 10, 1)
with ExperimentController('test', **std_kwargs) as ec:
tr = TrackerUD(ec, 3, 1, 1, 1, np.inf, 10, 1)
tr = TrackerUD(None, 3, 1, 1, 1, np.inf, 10, 1)
rand = np.random.RandomState(0)
while not tr.stopped:
tr.respond(rand.rand() < tr.x_current)
tr = TrackerUD(None, 3, 1, 1, 1, np.inf, 10, 1)
tr.threshold()
rand = np.random.RandomState(0)
while not tr.stopped:
tr.respond(rand.rand() < tr.x_current)
# test responding after stopped
assert_raises(RuntimeError, tr.respond, 0)
# all the properties better work
tr.up
tr.down
tr.step_size_up
tr.step_size_down
tr.stop_reversals
tr.stop_trials
tr.start_value
tr.x_min
tr.x_max
tr.stopped
tr.x
tr.responses
tr.n_trials
tr.n_reversals
tr.reversals
tr.reversal_inds
fig, ax, lines = tr.plot()
tr.plot_thresh(ax=ax)
tr.plot_thresh()
plt.close(fig)
ax = plt.axes()
fig, ax, lines = tr.plot(ax)
plt.close(fig)
tr.threshold()
tr.check_valid(2)
# bad callback type
assert_raises(TypeError, TrackerUD, 'foo', 3, 1, 1, 1, 10, np.inf, 1)
# test dynamic step size and error conditions
tr = TrackerUD(None, 3, 1, [1, 0.5], [1, 0.5], 10, np.inf, 1,
change_indices=[2])
tr.respond(True)
with warnings.catch_warnings(record=True) as w:
tr = TrackerUD(None, 1, 1, 0.75, 0.75, np.inf, 8, 1,
x_min=0, x_max=2)
responses = [True, True, True, False, False, False, False, True, False]
for r in responses: # run long enough to encounter change_indices
tr.respond(r)
assert_equal(len(w), 1)
assert(tr.check_valid(1)) # make sure checking validity is good
assert(not tr.check_valid(3))
assert_raises(ValueError, tr.threshold, 1)
tr.threshold(3)
# run tests with ignore too--should generate warnings, but no error
with warnings.catch_warnings(record=True) as w:
tr = TrackerUD(None, 1, 1, 0.75, 0.25, np.inf, 7, 1,
x_min=0, x_max=2, repeat_limit='ignore')
responses = [False, True, False, False, True, True, False, True]
for r in responses: # run long enough to encounter change_indices
tr.respond(r)
assert_equal(len(w), 1)
tr.threshold(0)
# bad stop_trials
assert_raises(ValueError, TrackerUD, None, 3, 1, 1, 1, 10, 'foo', 1)
# bad stop_reversals
assert_raises(ValueError, TrackerUD, None, 3, 1, 1, 1, 'foo', 10, 1)
# change_indices too long
assert_raises(ValueError, TrackerUD, None, 3, 1, [1, 0.5], [1, 0.5], 10,
np.inf, 1, change_indices=[1, 2])
# step_size_up length mismatch
assert_raises(ValueError, TrackerUD, None, 3, 1, [1], [1, 0.5], 10,
np.inf, 1, change_indices=[2])
# step_size_down length mismatch
assert_raises(ValueError, TrackerUD, None, 3, 1, [1, 0.5], [1], 10,
np.inf, 1, change_indices=[2])
# bad change_rule
assert_raises(ValueError, TrackerUD, None, 3, 1, [1, 0.5], [1, 0.5], 10,
np.inf, 1, change_indices=[2], change_rule='foo')
# no change_indices (i.e. change_indices=None)
assert_raises(ValueError, TrackerUD, None, 3, 1, [1, 0.5], [1, 0.5], 10,
np.inf, 1)
# start_value scalar type checking
assert_raises(TypeError, TrackerUD, None, 3, 1, [1, 0.5], [1, 0.5], 10,
np.inf, [9, 5], change_indices=[2])
assert_raises(TypeError, TrackerUD, None, 3, 1, [1, 0.5], [1, 0.5], 10,
np.inf, None, change_indices=[2])
# test with multiple change_indices
tr = TrackerUD(None, 3, 1, [3, 2, 1], [3, 2, 1], 10, np.inf, 1,
change_indices=[2, 4], change_rule='reversals')
max_lag = 2
pace_rule = 'reversals'
rng_dealer = np.random.RandomState(4) # random seed for selecting trial type
##############################################################################
# Initializing and Running Trackers
# ---------------------------------
# The two trackers in this example use all of the same parameters except for
# the start value and then are passed into the dealer. After the dealer is
# created, the type of trial with the start value above or below the true
# threshold (returned as an index) and trial level for that trial can be
# acquired.
# initialize two tracker objects--one for each start value
tr_ud = [
TrackerUD(callback, up, down, step_size_up, step_size_down, stop_reversals,
stop_trials, sv, change_indices, change_rule, x_min, x_max)
for sv in start_value
]
# initialize TrackerDealer object
td = TrackerDealer(callback, tr_ud, max_lag, pace_rule, rng_dealer)
# Initialize human state
rng_human = np.random.RandomState(1) # random seed for modeled subject
for _, level in td:
# Get information of which trial type is next and what the level is at
# that time from TrackerDealer
td.respond(rng_human.rand() < sigmoid(
level - true_thresh, lower=chance, slope=slope))