DDM RESPONSE_TIME Bug

[davidt0x]

Currently, on devel the RESPONSE_TIME output port of the integrator DDM appears to be bugged, but only in Python mode. The port mistakenly indicates that threshold was reached in a single step of integration. In compiled mode, the correct response time is returned. See example below.

import psyneulink as pnl

from psyneulink.core.globals.utilities import set_global_seed
set_global_seed(0)

my_ddm = pnl.DDM(
    function=pnl.DriftDiffusionIntegrator(
        starting_value=0,
        threshold=.1,
        noise=0.1,
        time_step_size=.1,
        non_decision_time=0
    ),
    reset_stateful_function_when=pnl.AtTrialStart(),
    output_ports=[pnl.DECISION_OUTCOME, pnl.RESPONSE_TIME],
    execute_until_finished=True,
    name="DDM"
)

comp = pnl.Composition()
comp.add_node(my_ddm)

comp.run(inputs=[.2])
# comp.run(inputs=[.2], execution_mode=pnl.ExecutionMode.LLVMRun)

print(comp.results)

Python Mode Output:

[[[1. ]
  [0.1]]]

Compiled Output:

[[[1. ]
  [0.5]]]

[jvesely]

This looks to be an issue with updating Parameter values.
With added tracking in the DriftDiffuctionIntegrator function:

        if not self.is_initializing:
            previous_value = adjusted_value
            print(context, "updating_time:", previous_time, "+", time_step_size)
            previous_time = previous_time + time_step_size

            self.parameters.previous_time._set(previous_time, context)
            print(context, "updated time:", self.parameters.previous_time.get(context), self._get_current_parameter_value('previous_time', context))

It shows that the value of previous_time gets updated, but the next iteration still observes 0.

<psyneulink.core.globals.context.Context object at 0x7fafabd16ae0> updating_time: [0.] + [0.1]
<psyneulink.core.globals.context.Context object at 0x7fafabd16ae0> updated time: [0.1] [0.1]

<psyneulink.core.globals.context.Context object at 0x7fafabd16ae0> updating_time: [0.] + [0.1]
<psyneulink.core.globals.context.Context object at 0x7fafabd16ae0> updated time: [0.1] [0.1]

<psyneulink.core.globals.context.Context object at 0x7fafabd16ae0> updating_time: [0.] + [0.1]
<psyneulink.core.globals.context.Context object at 0x7fafabd16ae0> updated time: [0.1] [0.1]

<psyneulink.core.globals.context.Context object at 0x7fafabd16ae0> updating_time: [0.] + [0.1]
<psyneulink.core.globals.context.Context object at 0x7fafabd16ae0> updated time: [0.1] [0.1]

<psyneulink.core.globals.context.Context object at 0x7fafabd16ae0> updating_time: [0.] + [0.1]
<psyneulink.core.globals.context.Context object at 0x7fafabd16ae0> updated time: [0.1] [0.1]

[jvesely]

The following patch fixes the issue for me:

diff --git a/psyneulink/core/components/functions/stateful/statefulfunction.py b/psyneulink/core/components/functions/stateful/statefulfunction.py
index 791faf5086..cadf64f13e 100644
--- a/psyneulink/core/components/functions/stateful/statefulfunction.py
+++ b/psyneulink/core/components/functions/stateful/statefulfunction.py
@@ -542,7 +542,7 @@ class StatefulFunction(Function_Base): #  --------------------------------------
                 continue
 
             param = getattr(self.parameters, attr)
-            param.set(kwargs[attr], context, override=True)
+            param.set(kwargs[attr].copy(), context, override=True)
             value.append(param._get(context))
 
         self.parameters.value.set(value, context, override=True)