ctu.nengoros.modules.impl

Class DefaultNeuralModule

java.lang.Object

ctu.nengoros.util.sync.impl.SyncedUnit

ctu.nengoros.modules.impl.DefaultNeuralModule

All Implemented Interfaces:

Node, Probeable, Resettable, SimulationMode.ModeConfigurable, VisiblyMutable, NeuralModule, PeripheralsRegisteringNode, SyncedUnitInterface, java.io.Serializable, java.lang.Cloneable

public class DefaultNeuralModule
extends SyncedUnit
implements NeuralModule

Extend this Neuron in order to implement your own functionality, or use DefaultNeuralModule. The same as AbsAsynNeuralModule, but this supports synchronous communication too. This module supports synchronous or asynchronous communication with ROS ndoes. Asynchronous case: what comes onto terminations is sent to ROS node(s). What comes from ROS node(s) is passed to origins (do not care when). Synchronous case: by default: all origins (Decoders) are supposed to be synchronous: -each message sent discards "ready" state of all Decoders -simulator waits until all Decoders receive something If you need some decoder(s) asynchronous, call method CreateDecoder with synchronous set to false.

See Also:

Serialized Form

Nested Class Summary

Nested classes/interfaces inherited from interface ca.nengo.util.VisiblyMutable

VisiblyMutable.Event, VisiblyMutable.Listener, VisiblyMutable.NameChangeEvent, VisiblyMutable.NodeRemovedEvent

Field Summary

Fields

Modifier and Type	Field and Description
protected ModemContainer	mc
static java.lang.String	me
protected java.lang.String	myDocumentation
protected java.util.Map<java.lang.String,Encoder>	myEncoders
protected SimulationMode	myMode
protected java.lang.String	myName
protected int	myNumGPU
protected int	myNumJavaThreads
protected java.util.Map<java.lang.String,Origin>	myOrigins
protected java.util.Properties	myProperties
protected java.util.Map<java.lang.String,Termination>	myTerminations
protected float	myTime
protected boolean	myUseGPU
protected java.util.LinkedList<Encoder>	orderedEncoders
protected java.util.LinkedList<Origin>	orderedOrigins
protected java.util.LinkedList<Termination>	orderedTerminations

Fields inherited from class ctu.nengoros.util.sync.impl.SyncedUnit

synchronous

Constructor Summary

Constructors

Constructor and Description
DefaultNeuralModule(java.lang.String name, NodeGroup group) The NeuralModule which features BasicEncoders (with BasicMultiTerminations), BasicDecoders (which are synchronous by default, that means they wait for ROS message to be received), which is synchronous by default.
DefaultNeuralModule(java.lang.String name, NodeGroup group, boolean synchronous) The NeuralModule which features BasicEncoders (with BasicMultiTerminations), BasicDecoders (which are synchronous by default, that means they wait for ROS message to be received), which is synchronous by default.

Method Summary

Methods

Modifier and Type	Method and Description
void	addChangeListener(VisiblyMutable.Listener listener)
protected void	addEncoder(Encoder e)
void	addOrigin(Origin o) This is used by Decoders to add their origin to the NeuralModule, should not be used from any other place.
void	addTermination(Termination t) This is used by Encoders to add their termination to the NeuralModule, should not be used from any other place.
DefaultNeuralModule	clone()
void	createDecoder(java.lang.String topicName, java.lang.String dataType)
void	createDecoder(java.lang.String topicName, java.lang.String dataType, boolean synchronous) Decoder with number of primitive data types given by the ROS message type.
void	createDecoder(java.lang.String topicName, java.lang.String dataType, int dimensionSize)
void	createDecoder(java.lang.String topicName, java.lang.String dataType, int[] dimensionSizes) Add the Decoder to the Module, that is: Subscribe to given ROS topic with expected type of data and expected dimensionality. Add Decoder to Node in the Nengo with corresponding dimensionality.
void	createDecoder(java.lang.String topicName, java.lang.String dataType, int[] dimensionSizes, boolean synchronous) This method should add decoder (that is Nengo origin to this NeuralModule).
void	createDecoder(java.lang.String topicName, java.lang.String dataType, int dimensionSize, boolean synchronous)
void	createEncoder(java.lang.String topicName, java.lang.String dataType) Create encoder where the dimensionality of message data is determined by data type (e.g.
void	createEncoder(java.lang.String topicName, java.lang.String dataType, int dimensionSize)
void	createEncoder(java.lang.String topicName, java.lang.String dataType, int[] dimensionSizes) This method adds encoder, that is owner of one or multiple Terminations to this NeuralModule.
Node[]	getChildren()
java.lang.String	getDocumentation()
TimeSeries	getHistory(java.lang.String key) this is made for Probeable and enables GUI to add a probe (and collect data) I made every (RealValued) Termination and Origin probeable
SimulationMode	getMode()
MultiTermination	getMultiTermination(java.lang.String name) This NeuralModule supports connecting multiple "inputs" to one Encoder.
java.lang.String	getName()
Origin	getOrigin(java.lang.String name)
Origin[]	getOrigins()
Termination	getTermination(java.lang.String name)
Termination[]	getTerminations()
protected void	init(java.lang.String name, NodeGroup group)
java.util.Properties	listStates() also for Probeable..
Termination	newTerminationFor(java.lang.String name) In order to simplify connecting new inputs to the Encoder (MultiTermination in general), this method can be use to create new Terminations.
Termination	newTerminationFor(java.lang.String name, float weight) The same as the NeuralModule.newTerminationFor(String), but here the weight can be specified.
Termination	newTerminationFor(java.lang.String name, java.lang.Float[] weights) The same as the NeuralModule.newTerminationFor(String), but here the weight for each dimension can be specified.
void	notifyAboutDeletion() This method tells the Node (network) that is being deleted from Nengo, Node can inform its children (child Nodes) that are being deleted, these nodes can e.g.
void	printOriginNames()
void	printTerminationNames()
void	removeChangeListener(VisiblyMutable.Listener listener)
void	reset(boolean randomize)
void	run(float startTime, float endTime) Runs the Node (including all its components), updating internal state and outputs as needed.
void	setDocumentation(java.lang.String text)
void	setMode(SimulationMode mode) Sets the object to run in either the given mode or the closest mode that it supports (all ModeConfigurables must support SimulationMode.DEFAULT, and must default to this mode).
void	setName(java.lang.String name)
void	setSynchronous(boolean synchronous) Turn synchronous communication on/off.
java.lang.String	toScript(java.util.HashMap<java.lang.String,java.lang.Object> scriptData)

Methods inherited from class ctu.nengoros.util.sync.impl.SyncedUnit

addChild, discardChildsReady, isReady, removeChild, setReady

Methods inherited from class java.lang.Object

equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

me

public static final java.lang.String me

See Also:

Constant Field Values

myTime

protected float myTime

myProperties

protected java.util.Properties myProperties

myOrigins

protected java.util.Map<java.lang.String,Origin> myOrigins

orderedOrigins

protected java.util.LinkedList<Origin> orderedOrigins

myTerminations

protected java.util.Map<java.lang.String,Termination> myTerminations

orderedTerminations

protected java.util.LinkedList<Termination> orderedTerminations

myEncoders

protected java.util.Map<java.lang.String,Encoder> myEncoders

orderedEncoders

protected java.util.LinkedList<Encoder> orderedEncoders

myDocumentation

protected java.lang.String myDocumentation

myName

protected java.lang.String myName

myNumGPU

protected int myNumGPU

myNumJavaThreads

protected int myNumJavaThreads

myUseGPU

protected boolean myUseGPU

myMode

protected SimulationMode myMode

mc

protected ModemContainer mc

Constructor Detail

DefaultNeuralModule

public DefaultNeuralModule(java.lang.String name,
                   NodeGroup group)
                    throws ConnectionException

The NeuralModule which features BasicEncoders (with BasicMultiTerminations), BasicDecoders (which are synchronous by default, that means they wait for ROS message to be received), which is synchronous by default.

The simulator waits for all synchronous components at each sim. step

Parameters:

name - name of the Module

group - NodeGroup which belongs to this Module (group can have one or multiple ROS nodes and one Modem for translating communication between ROS and Nengo)

Throws:

ConnectionException - group of running ROS nodes should already contain the Modem, if no Modem is found, ROS communication will not be available

See Also:

Modem, SyncedUnit, BasicDecoder, BasicEncoder

DefaultNeuralModule

public DefaultNeuralModule(java.lang.String name,
                   NodeGroup group,
                   boolean synchronous)
                    throws ConnectionException

The NeuralModule which features BasicEncoders (with BasicMultiTerminations), BasicDecoders (which are synchronous by default, that means they wait for ROS message to be received), which is synchronous by default.

The simulator waits for all synchronous components at each sim. step

Parameters:

name - name of the Module

group - NodeGroup which belongs to this Module (group can have one or multiple ROS nodes and one Modem for translating communication between ROS and Nengo)

synchronous - the Nengo simulator waits each time step for all synchronous Modules

Throws:

ConnectionException - group of running ROS nodes should already contain the Modem, if no Modem is found, ROS communication will not be available

See Also:

Modem, SyncedUnit, BasicDecoder, BasicEncoder

Method Detail

init

protected void init(java.lang.String name,
        NodeGroup group)
             throws ConnectionException

Throws:

ConnectionException

setSynchronous

public void setSynchronous(boolean synchronous)

Description copied from interface: NeuralModule

Turn synchronous communication on/off. Node is synchronous by default. If the synchronous communication is turned on, the Nengo simulator waits for messages from all synchronous Decoders of the Module each simulation step.

If the Module is set to asynchronous mode: the simulation does not wait for any of its childs. In the synchronous mode, only synchronous Decoders can block the simulation

.

Specified by:

setSynchronous in interface NeuralModule

Specified by:

setSynchronous in interface SyncedUnitInterface

Overrides:

setSynchronous in class SyncedUnit

createDecoder

public void createDecoder(java.lang.String topicName,
                 java.lang.String dataType,
                 int[] dimensionSizes,
                 boolean synchronous)

This method should add decoder (that is Nengo origin to this NeuralModule). Each Decoder is registered as a child of thus SynchedUnit, so if the decoder is set to be synchronous, the simulator will wait for new values on it each sim. step. Note: units are left default, TODO: add units to make use of other ROS message types

Specified by:

createDecoder in interface NeuralModule

Parameters:

topicName - name of the origin and ROS topic

dimensionSizes - dimensions of decoded data

synchronous - whether the decoder will be synchronous (see above)

dataType - type of data

See Also:

rosjava_core doc

createDecoder

public void createDecoder(java.lang.String topicName,
                 java.lang.String dataType,
                 int dimensionSize,
                 boolean synchronous)

Specified by:

createDecoder in interface NeuralModule

createDecoder

public void createDecoder(java.lang.String topicName,
                 java.lang.String dataType,
                 boolean synchronous)

Decoder with number of primitive data types given by the ROS message type.

Specified by:

createDecoder in interface NeuralModule

createDecoder

public void createDecoder(java.lang.String topicName,
                 java.lang.String dataType,
                 int[] dimensionSizes)

Description copied from interface: NeuralModule

Add the Decoder to the Module, that is:

• Subscribe to given ROS topic with expected type of data and expected dimensionality.
• Add Decoder to Node in the Nengo with corresponding dimensionality. The Decoder converts ROS messages to array of float values expected by the Nengo RealOutput.
• The decoder is allowed to add own Origin to this Node. This Origin then produces data received by the ROS subscriber

Note: the name of ROS node is specified by the Modem instance.

Specified by:

createDecoder in interface NeuralModule

Parameters:

topicName - name of the ROS topic

dataType - type of data

dimensionSizes - list of dimension sizes (best support only for 1D vectors)

See Also:

rosjava_core doc

createDecoder

public void createDecoder(java.lang.String topicName,
                 java.lang.String dataType,
                 int dimensionSize)

Specified by:

createDecoder in interface NeuralModule

createDecoder

public void createDecoder(java.lang.String topicName,
                 java.lang.String dataType)

Specified by:

createDecoder in interface NeuralModule

createEncoder

public void createEncoder(java.lang.String topicName,
                 java.lang.String dataType,
                 int[] dimensionSizes)

This method adds encoder, that is owner of one or multiple Terminations to this NeuralModule. Note: units are left default, TODO: add units to make use of other ROS message types

Specified by:

createEncoder in interface NeuralModule

Parameters:

topicName - name of termination and the corresponding ROS topic

dimension -

See Also:

rosjava_core doc

createEncoder

public void createEncoder(java.lang.String topicName,
                 java.lang.String dataType,
                 int dimensionSize)

Specified by:

createEncoder in interface NeuralModule

createEncoder

public void createEncoder(java.lang.String topicName,
                 java.lang.String dataType)

Create encoder where the dimensionality of message data is determined by data type (e.g. turtlesim/Velocity=2)

Specified by:

createEncoder in interface NeuralModule

run

public void run(float startTime,
       float endTime)
         throws SimulationException

Description copied from interface: Node

Runs the Node (including all its components), updating internal state and outputs as needed. Runs should be short (eg 1ms), because inputs can not be changed during a run, and outputs will only be communicated to other Nodes after a run.

Specified by:

run in interface Node

Parameters:

startTime - simulation time at which running starts (s)

endTime - simulation time at which running ends (s)

Throws:

SimulationException - if a problem is encountered while trying to run

reset

public void reset(boolean randomize)

Specified by:

reset in interface Resettable

Parameters:

randomize - True indicates that the object should be reset to a randomly selected initial state (the object must be aware of the distribution from which to draw from). False indicates that the object should be reset to a fixed initial state (which it must also know). Some objects may not support randomization of the initial state, in which case a fixed state will be used in either case.

setMode

public void setMode(SimulationMode mode)

Description copied from interface: SimulationMode.ModeConfigurable

Sets the object to run in either the given mode or the closest mode that it supports (all ModeConfigurables must support SimulationMode.DEFAULT, and must default to this mode).

Specified by:

setMode in interface SimulationMode.ModeConfigurable

Parameters:

mode - SimulationMode in which it is desired that the object runs.

getMode

public SimulationMode getMode()

Specified by:

getMode in interface SimulationMode.ModeConfigurable

Returns:

The SimulationMode in which the object is running

addChangeListener

public void addChangeListener(VisiblyMutable.Listener listener)

Specified by:

addChangeListener in interface VisiblyMutable

Parameters:

listener - Listener to add

removeChangeListener

public void removeChangeListener(VisiblyMutable.Listener listener)

Specified by:

removeChangeListener in interface VisiblyMutable

Parameters:

listener - Listener to remove

getHistory

public TimeSeries getHistory(java.lang.String key)
                      throws SimulationException

this is made for Probeable and enables GUI to add a probe (and collect data) I made every (RealValued) Termination and Origin probeable

Specified by:

getHistory in interface Probeable

Parameters:

key - A state variable name

Returns:

History of values for the named state variable. The history must cover the most recent network time step, and no more. There should be no overlap in the time points returned for different steps.

Throws:

SimulationException - if the Probeable does not have the requested state

listStates

public java.util.Properties listStates()

also for Probeable..

Specified by:

listStates in interface Probeable

Returns:

List of state variable names, eg "V", and associated descriptions eg "membrane potential (mV)"

getName

public java.lang.String getName()

Specified by:

getName in interface Node

Specified by:

getName in interface SyncedUnitInterface

Overrides:

getName in class SyncedUnit

Returns:

Name of Node (must be unique in a Network)

setName

public void setName(java.lang.String name)
             throws StructuralException

Specified by:

setName in interface Node

Specified by:

setName in interface SyncedUnitInterface

Overrides:

setName in class SyncedUnit

Parameters:

name - The new name

Throws:

StructuralException - if name already exists?

getOrigins

public Origin[] getOrigins()

Specified by:

getOrigins in interface Node

Returns:

Sets of ouput channels (eg spiking outputs, gap junctional outputs, etc.)

getOrigin

public Origin getOrigin(java.lang.String name)
                 throws StructuralException

Specified by:

getOrigin in interface Node

Parameters:

name - Name of an Origin on this Node

Returns:

The named Origin if it exists

Throws:

StructuralException - if the named Origin does not exist

getTerminations

public Termination[] getTerminations()

Specified by:

getTerminations in interface Node

Returns:

Sets of input channels (these have the same dimension as corresponding Origins to which they are connected).

getTermination

public Termination getTermination(java.lang.String name)
                           throws StructuralException

Specified by:

getTermination in interface Node

Parameters:

name - Name of a Termination onto this Node

Returns:

The named Termination if it exists

Throws:

StructuralException - if the named Termination does not exist

getDocumentation

public java.lang.String getDocumentation()

Specified by:

getDocumentation in interface Node

Returns:

User-specified documentation for the Node, if any

setDocumentation

public void setDocumentation(java.lang.String text)

Specified by:

setDocumentation in interface Node

Parameters:

text - New user-specified documentation for the Node

clone

public DefaultNeuralModule clone()

Specified by:

clone in interface Node

Overrides:

clone in class java.lang.Object

Returns:

An independent copy of the Node

notifyAboutDeletion

public void notifyAboutDeletion()

Description copied from interface: Node

This method tells the Node (network) that is being deleted from Nengo, Node can inform its children (child Nodes) that are being deleted, these nodes can e.g. stop their ROS backends, their ROS nodes etc.. Note that it is expected that node returns from this method when everything that should be done before his deletion is done. Also, Node does not have to react to this (as in pure Nengo version). ///my @author Jaroslav Vitku

Specified by:

notifyAboutDeletion in interface Node

getChildren

public Node[] getChildren()

Specified by:

getChildren in interface Node

toScript

public java.lang.String toScript(java.util.HashMap<java.lang.String,java.lang.Object> scriptData)
                          throws ScriptGenException

Specified by:

toScript in interface Node

Parameters:

scriptData - Map of class parent and prefix data for generating python script

Returns:

Python script for generating the node

Throws:

ScriptGenException - if the node cannot be generated in script

addOrigin

public void addOrigin(Origin o)
               throws StructuralException

Description copied from interface: PeripheralsRegisteringNode

This is used by Decoders to add their origin to the NeuralModule, should not be used from any other place.

Specified by:

addOrigin in interface PeripheralsRegisteringNode

Parameters:

o - Origin to be added

Throws:

StructuralException - is thrown if there already is orig. with this name

addTermination

public void addTermination(Termination t)
                    throws StructuralException

Description copied from interface: PeripheralsRegisteringNode

This is used by Encoders to add their termination to the NeuralModule, should not be used from any other place. Data from this termination are read by encoder, encoded into ROS message format and sent to ROS net.

Specified by:

addTermination in interface PeripheralsRegisteringNode

Parameters:

t - Termination to be added to the parent

Throws:

StructuralException - is thrown if there already is term. with this name

getMultiTermination

public MultiTermination getMultiTermination(java.lang.String name)
                                     throws StructuralException

Description copied from interface: NeuralModule

This NeuralModule supports connecting multiple "inputs" to one Encoder. It is accomplished by the fact that each Encoder has MultiTermination, where the MultiTermination can have several own Terminations, whose values are combined into one value.

The user is allowed to create new Terminations for Encoder. In order to do this, the Encoders MultiTermination has to be available. The MultiTermination is generated with identical name as the Encoder. This method provides access to the Encoders MultiTermination by its name.

Specified by:

getMultiTermination in interface NeuralModule

Parameters:

name - name of the Encoder==MultiTermination

Returns:

MultiTermination if found

Throws:

StructuralException - if no Encoder of such name found

addEncoder

protected void addEncoder(Encoder e)
                   throws StructuralException

Throws:

StructuralException

printTerminationNames

public void printTerminationNames()

Specified by:

printTerminationNames in interface PeripheralsRegisteringNode

printOriginNames

public void printOriginNames()

Specified by:

printOriginNames in interface PeripheralsRegisteringNode

newTerminationFor

public Termination newTerminationFor(java.lang.String name)
                              throws StructuralException

Description copied from interface: NeuralModule

In order to simplify connecting new inputs to the Encoder (MultiTermination in general), this method can be use to create new Terminations. It receives name of the Encoded ROS topic (name of MultiTermination), crates new Termination with default weight (1), registers it to own MultiTermination and this NauralModule.

The MultiTermination in DefaultNeuralModule computes weighted sum of all its Terminations. The resulting value is then encoded and sent as a ROS message on the topic which corresponds to the Encoders name.

Specified by:

newTerminationFor in interface NeuralModule

Parameters:

name - name of MultiTermination to connect newly created Termination (this equals to the name of ROS topic used by Encoder)

Returns:

newly created Termination

Throws:

StructuralException - if the Termination could not be created (e.g. Encoder not found, name is used etc.) Encoder.getMultiTermination() MultiTermination.addTermination()

newTerminationFor

public Termination newTerminationFor(java.lang.String name,
                            float weight)
                              throws StructuralException

Description copied from interface: NeuralModule

The same as the NeuralModule.newTerminationFor(String), but here the weight can be specified.

Specified by:

newTerminationFor in interface NeuralModule

Parameters:

name - name of the ROS topic == name of the Encoder == name of the parent MultiTermination

weight - terminations input is weighted

Returns:

newly created Termination

Throws:

StructuralException - if the Termination could not be created

See Also:

{@link ctu.nengoros.comm.rosBackend.encoders.multiTermination.MultiTermination#addTermination(float)}

newTerminationFor

public Termination newTerminationFor(java.lang.String name,
                            java.lang.Float[] weights)
                              throws StructuralException

Description copied from interface: NeuralModule

The same as the NeuralModule.newTerminationFor(String), but here the weight for each dimension can be specified.

Specified by:

newTerminationFor in interface NeuralModule

Parameters:

name - name of the Encoder

weights - array of weights, for each dimension one weight

Returns:

newly created Termination

Throws:

StructuralException - exception if the Termination could not be created (e.g. dimension of weight array provided is incorrect)

See Also:

{@link ctu.nengoros.comm.rosBackend.encoders.multiTermination.MultiTermination#addTermination(Float[])}