Merge pull request #38 from hoechenberger/rng

ENH: Allow passing a random seed via stim_selection_options

Author: hoechenberger

.appveyor.yml

@@ -21,7 +21,7 @@ install:
   - python setup.py build
 
   # Build & install sdist.
-  - python setup.py sdist
+  - python setup.py sdist --formats=zip
 #  - pip install --no-deps dist/questplus-*.zip
 #  - pip uninstall --yes questplus
 
@@ -30,7 +30,8 @@ install:
 #  - pip install --no-deps dist/questplus-*.whl
 #  - pip uninstall --yes questplus
 
-  - pip install .
+  - ps: Remove-Item –path dist, build –recurse
+  - pip install --no-deps .
 
 test_script:
   - py.test

.travis.yml

@@ -50,7 +50,7 @@ install:
 
   - rm -rf dist/ build/
 
-  - pip install .
+  - pip install --no-deps .
 
 script:
   - py.test

CHANGES.md

@@ -1,3 +1,10 @@
+v2019.2
+-------
+* Allow passing a random seed via `stim_selection_options` keyword
+  argument
+* Better handling of `stim_selection_options` defaults (now allows
+  to supply only a subset of options)
+
 v2019.1
 -------
 * Allow to pass priors for only some parameters

questplus/_constants.py

@@ -0,0 +1,3 @@
+DEFAULT_N = 4
+DEFAULT_MAX_CONSECUTIVE_REPS = 2
+DEFAULT_RANDOM_SEED = None

questplus/qp.py

@@ -66,8 +66,9 @@ def __init__(self, *,
             method specified via `stim_selection_method`. Currently, this can
             be used to specify the number of `n` stimuli that will yield the
             `n` smallest entropies if `stim_selection_method=min_n_entropy`,
-            and`max_consecutive_reps`, the number of times the same stimulus
-            can be presented consecutively.
+            and `max_consecutive_reps`, the number of times the same stimulus
+            can be presented consecutively. A random number generator seed
+            may be passed via `random_seed=12345`.
 
         param_estimation_method
             The method to use when deriving the final parameter estimate.
@@ -88,11 +89,33 @@ def __init__(self, *,
 
         self.stim_selection = stim_selection_method
 
-        if (self.stim_selection == 'min_n_entropy' and
-                stim_selection_options is None):
-            self.stim_selection_options = dict(n=4, max_consecutive_reps=2)
+        if self.stim_selection == 'min_n_entropy':
+            from ._constants import (DEFAULT_N, DEFAULT_RANDOM_SEED,
+                                     DEFAULT_MAX_CONSECUTIVE_REPS)
+
+            if stim_selection_options is None:
+                self.stim_selection_options = dict(
+                    n=DEFAULT_N,
+                    max_consecutive_reps=DEFAULT_MAX_CONSECUTIVE_REPS,
+                    random_seed=DEFAULT_RANDOM_SEED)
+            else:
+                self.stim_selection_options = stim_selection_options.copy()
+
+                if 'n' not in stim_selection_options:
+                    self.stim_selection_options['n'] = DEFAULT_N
+                if 'max_consecutive_reps' not in stim_selection_options:
+                    self.stim_selection_options['max_consecutive_reps'] = DEFAULT_MAX_CONSECUTIVE_REPS
+                if 'random_seed' not in stim_selection_options:
+                    self.stim_selection_options['random_seed'] = DEFAULT_RANDOM_SEED
+
+            del DEFAULT_N, DEFAULT_MAX_CONSECUTIVE_REPS, DEFAULT_RANDOM_SEED
+
+            seed = self.stim_selection_options['random_seed']
+            self._rng = np.random.RandomState(seed=seed)
+            del seed
         else:
             self.stim_selection_options = stim_selection_options
+            self._rng = None
 
         self.param_estimation_method = param_estimation_method
 
@@ -271,7 +294,7 @@ def next_stim(self) -> dict:
             while True:
                 # Randomly pick one index and retrieve its coordinates
                 # (stimulus parameters).
-                candidate_index = np.random.choice(indices)
+                candidate_index = self._rng.choice(indices)
                 coords = EH[candidate_index].coords
                 stim = {stim_property: stim_val.item()
                         for stim_property, stim_val in coords.items()}

questplus/tests/test_qp.py

@@ -2,6 +2,7 @@
 import scipy.stats
 import numpy as np
 from questplus.qp import QuestPlus, QuestPlusWeibull
+from questplus import _constants
 
 
 def test_threshold():
@@ -597,6 +598,66 @@ def test_prior_for_parameter_subset():
                                        'lower_asymptote']).sum())
 
 
+def test_stim_selection_options():
+    threshold = np.arange(-40, 0 + 1)
+    slope, guess, lapse = 3.5, 0.5, 0.02
+    contrasts = threshold.copy()
+
+    stim_domain = dict(intensity=contrasts)
+    param_domain = dict(threshold=threshold, slope=slope,
+                        lower_asymptote=guess, lapse_rate=lapse)
+    outcome_domain = dict(response=['Correct', 'Incorrect'])
+
+    f = 'weibull'
+    scale = 'dB'
+    stim_selection_method = 'min_n_entropy'
+    param_estimation_method = 'mode'
+
+    common_params = dict(stim_domain=stim_domain, param_domain=param_domain,
+                         outcome_domain=outcome_domain, func=f,
+                         stim_scale=scale,
+                         stim_selection_method=stim_selection_method,
+                         param_estimation_method=param_estimation_method)
+
+    stim_selection_options = None
+    q = QuestPlus(**common_params,
+                  stim_selection_options=stim_selection_options)
+    expected = dict(n=_constants.DEFAULT_N,
+                    max_consecutive_reps=_constants.DEFAULT_MAX_CONSECUTIVE_REPS,
+                    random_seed=_constants.DEFAULT_RANDOM_SEED)
+    assert expected == q.stim_selection_options
+
+    stim_selection_options = dict(n=5)
+    q = QuestPlus(**common_params,
+                  stim_selection_options=stim_selection_options)
+    expected = dict(n=5,
+                    max_consecutive_reps=_constants.DEFAULT_MAX_CONSECUTIVE_REPS,
+                    random_seed=_constants.DEFAULT_RANDOM_SEED)
+    assert expected == q.stim_selection_options
+
+    stim_selection_options = dict(max_consecutive_reps=4)
+    q = QuestPlus(**common_params,
+                  stim_selection_options=stim_selection_options)
+    expected = dict(n=_constants.DEFAULT_N,
+                    max_consecutive_reps=4,
+                    random_seed=_constants.DEFAULT_RANDOM_SEED)
+    assert expected == q.stim_selection_options
+
+    stim_selection_options = dict(random_seed=999)
+    q = QuestPlus(**common_params,
+                  stim_selection_options=stim_selection_options)
+    expected = dict(n=_constants.DEFAULT_N,
+                    max_consecutive_reps=_constants.DEFAULT_MAX_CONSECUTIVE_REPS,
+                    random_seed=999)
+    assert expected == q.stim_selection_options
+
+    stim_selection_options = dict(n=5, max_consecutive_reps=4, random_seed=999)
+    q = QuestPlus(**common_params,
+                  stim_selection_options=stim_selection_options)
+    expected = stim_selection_options.copy()
+    assert expected == q.stim_selection_options
+
+
 if __name__ == '__main__':
     test_threshold()
     test_threshold_slope()
@@ -609,3 +670,4 @@ def test_prior_for_parameter_subset():
     test_marginal_posterior()
     test_prior_for_unknown_parameter()
     test_prior_for_parameter_subset()
+    test_stim_selection_options()