Merge branch 'devel'

* devel:
  Feat/projections/masked mapping projection (#775)
  adding exceptions to get_current_function_param and get_current_mechanism_param - not intended for looking up 'variable'
  Feat/projections/masked mapping projection (#771)
  • LCA (#770)
  • LCA (#769)
  Refactor/mechanism/transfermechanism (#768)
  Fix default_variable bug (#767)
  Feat/inputstate/combine (#766)
  Refactor/context/source (#765)
  Defaults: ensure that function attr appears when examining Defaults
  Rewrite Function instantiation:
  Component: create fewer instances during init when function arg is a class
  LinearMatrix: make keyword method static to reflect how it is used
  refactor/rename _variable_not_specified to be more generic, using enum
  Defaults: add enum to denote flexibility of assignments
  Component: add class method to get new param class defaults
  Defaults: add ClassDefaults.function to several classes, and use it as fallback
  utilities: add function to detect if obj is an instance or a subclass of a class
  Component: override __deepcopy__ method to use shared items
  parsing: add method to parse function variable from variable
  imports: pycharm-optimize to avoid unnecessary circularities
  Params: copy dict/ROODs nested in params arguments to avoid side effects
  Refactor/context/set and test (#762)
  fixing bug in looking up target values for learning which caused first target value to be repeated
  LinearCombination, Reduce: handle negative exponents during init (avoid zero division)
  Functions: simplify noise validation
  adding a note to the ddm plot() documentation to clarify that the plot does not correspond to the actual results of the mechanism
  fixing drift diffusion integrator bug - returned threshold even when it crossed the negative threshold
  Fix/misc/misc (#759)
  PredictionErrorMechanism: correct infinite recursion in _execute
  Docs/context/context (#757)
  • Mechanism (#756)
  Refactor/mechanisms/ execute (#754)
  LearningAuxilliary: rename to LearningAuxiliary (correct double l typo)
  scheduling: decouple Scheduler and Condition, allow multiple execution_ids
  utilites: get method to prune unused args from a function from composition branch
  Feat/mechanism/input target lable dicts (#752)
  testing: Resolve leftover merge conflicts, fixes #747
  testing: correct pytest setup ovewrite, losing some settings
  Feat/mechanism/input target label dicts (#751)
  Scheduling: fix bug where termination conditions persisted across calls to run
  tests,function/LinearCombination: Add tests with absent parameters
  test,function/LinearCombination: Rename second test function to prevent overwriting results
  • Context (#746)
  Refactor/context/deprecate init status (#745)
  Refactor/context/structured (#744)
  Fix/function/stability (#743)
  • IntegratorMechanism (#742)

README.rst

@@ -159,7 +159,7 @@ Contributors
 * **Peter Johnson**, Princeton Neuroscience Institute, Princeton University
 * **Justin Junge**, Department of Psychology, Princeton University
 * **Kristen Manning**, Princeton Neuroscience Institute, Princeton University
-* **Kevin Mantel**, Princeton Neuroscience Institute, Princeton University
+* **Katherine Mantel**, Princeton Neuroscience Institute, Princeton University
 * **Markus Spitzer**, Princeton Neuroscience Institute, Princeton University
 * **Jan Vesely**, Department of Computer Science, Rutgers University
 * **Changyan Wang**, Princeton Neuroscience Institute, Princeton University

Scripts/Aida.py

@@ -0,0 +1,137 @@
+
+# coding: utf-8
+
+# In[1]:
+
+
+import psyneulink as pnl
+import numpy as np
+
+
+# In[2]:
+
+
+# ECin = pnl.KWTA(size=8, function=pnl.Linear)
+# DG = pnl.KWTA(size=400, function=pnl.Linear)
+# CA3 = pnl.KWTA(size=80, function=pnl.Linear)
+# CA1 = pnl.KWTA(size=100, function=pnl.Linear)
+# ECout = pnl.KWTA(size=8, function=pnl.Linear)
+ECin = pnl.TransferMechanism(size=8, function=pnl.Linear(), name='ECin')
+DG = pnl.TransferMechanism(size=400, function=pnl.Logistic(), name='DG')
+CA3 = pnl.TransferMechanism(size=80, function=pnl.Logistic(), name='CA3')
+CA1 = pnl.TransferMechanism(size=100, function=pnl.Linear(), name='CA1')
+ECout = pnl.TransferMechanism(size=8, function=pnl.Logistic(), name='ECout')
+
+
+# In[3]:
+
+
+def make_mask(in_features, out_features, connectivity):
+    mask = np.zeros((in_features, out_features))
+    rand = np.random.random(mask.shape)
+    idxs = np.where(rand < connectivity)
+    mask[idxs[0], idxs[1]] = 1
+    return mask
+
+def make_mat(in_features, out_features, lo, high, mask):
+    w = np.random.uniform(lo ,high, size=(in_features, out_features))
+    w = mask * w
+    return w
+
+
+# In[4]:
+
+
+ECin_s, ECout_s, DG_s, CA3_s, CA1_s = 8, 8, 400, 80, 100
+
+
+# In[5]:
+
+
+mask_ECin_DG = make_mask(ECin_s, DG_s, 0.25)
+mask_DG_CA3 = make_mask(DG_s, CA3_s, 0.05)
+mask_ECin_CA3 = make_mask(ECin_s, CA3_s, 0.25)
+
+mat_ECin_DG = make_mat(ECin_s, DG_s, 0.25, 0.75, mask_ECin_DG)
+mat_DG_CA3 = make_mat(DG_s, CA3_s, 0.89, 0.91, mask_DG_CA3)
+mat_ECin_CA3 = make_mat(ECin_s, CA3_s, 0.25, 0.75, mask_ECin_CA3)
+
+mat_CA3_CA1 = make_mat(CA3_s, CA1_s, 0.25, 0.75, np.ones((CA3_s, CA1_s)))
+mat_CA1_ECout = make_mat(CA1_s, ECout_s, 0.25, 0.75, np.ones((CA1_s, ECout_s)))
+mat_ECin_CA1 = make_mat(ECin_s, CA1_s, 0.25, 0.75, np.ones((ECin_s, CA1_s)))
+
+
+# In[6]:
+
+
+ECin_to_DG=pnl.MappingProjection(matrix=mat_ECin_DG)
+DG_to_CA3=pnl.MappingProjection(matrix=mat_DG_CA3)
+ECin_to_CA3=pnl.MappingProjection(matrix=mat_ECin_CA3)
+CA3_to_CA1=pnl.MappingProjection(matrix=mat_CA3_CA1)
+CA1_to_ECout=pnl.MappingProjection(sender=CA1, receiver=ECout, matrix=mat_CA1_ECout)
+ECin_to_CA1=pnl.MappingProjection(sender=ECin, receiver=CA1, matrix=mat_ECin_CA1)
+
+
+# In[7]:
+
+
+proc_ECin_DG = pnl.Process(pathway=[ECin, ECin_to_DG, DG], learning=pnl.ENABLED, learning_rate=0.2)
+proc_ECin_CA3 = pnl.Process(pathway=[ECin, ECin_to_CA3, CA3], learning=pnl.ENABLED, learning_rate=0.2)
+proc_DG_CA3 = pnl.Process(pathway=[DG, DG_to_CA3, CA3], learning=pnl.ENABLED, learning_rate=0)
+proc_CA3_CA1 = pnl.Process(pathway=[CA3, CA3_to_CA1, CA1], learning=pnl.ENABLED, learning_rate=0.05)
+proc_CA1_ECout = pnl.Process(pathway=[CA1, ECout], learning=pnl.ENABLED, learning_rate=0.02)
+proc_ECin_CA1 = pnl.Process(pathway=[ECin, CA1], learning_rate=0.02)
+
+
+# In[8]:
+
+
+TSP = pnl.System(processes=[proc_ECin_DG, proc_ECin_CA3, proc_DG_CA3, proc_CA3_CA1, proc_CA1_ECout])
+# MSP = pnl.System(processes=[proc_ECin_CA1, proc_CA1_ECout])
+
+
+# In[9]:
+
+
+TSP.show_graph()
+assert True
+
+# In[10]:
+
+
+## Method for making input
+def statistical():
+    chars = ['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I']
+    sequence = ''
+    letters = range(8)
+    starters = range(0, 8, 2)
+    enders = range(1, 8, 2)
+
+    ## minus phase
+    idx = np.random.randint(len(starters))
+    s = starters[idx]
+    e = enders[idx]
+    minus_input, minus_target = np.zeros((8)), np.zeros((8))
+    minus_input[s] = 1.0
+    minus_target[e] = 1.0
+    minus_target[s] = 0.9
+    sequence += chars[s]
+    sequence += chars[e]
+
+    ## plus phase
+    plus_input, plus_target = minus_target, np.zeros((8))
+    plus_target[s] = 1
+    plus_target[e] = 1
+
+    return (minus_input, minus_target, plus_input, plus_target)
+
+
+# In[ ]:
+
+
+epochs = 100
+for epoch in range(epochs):
+    minus_x, minus_y, plus_x, plus_y = statistical()
+    TSP.run(inputs={ECin:minus_x}, targets={ECout:minus_y})
+    ## Running the above line of code causes weights to get too large
+

Scripts/Examples/EVC-Gratton.py

@@ -145,8 +145,8 @@
 
 # Show graph of system (with control components)
 # mySystem.show_graph(show_dimensions=pnl.ALL, show_projection_labels=True)
-# mySystem.show_graph(show_control=True, show_projection_labels=True)
-mySystem.show_graph(show_control=True, show_mechanism_structure=True, show_headers=False)
+mySystem.show_graph(show_control=True, show_projection_labels=False)
+# mySystem.show_graph(show_control=True, show_mechanism_structure=True, show_headers=False)
 
 # configure EVC components
 mySystem.controller.control_signals[0].intensity_cost_function = pnl.Exponential(rate=0.8046).function
@@ -193,7 +193,7 @@
 
 mySystem.controller.reportOutputPref = True
 
-Flanker_Rep.set_log_conditions((pnl.SLOPE, pnl.ContextStatus.CONTROL))
+Flanker_Rep.set_log_conditions((pnl.SLOPE, pnl.ContextFlags.CONTROL))
 
 mySystem.run(
     num_trials=nTrials,