teili.building_blocks package¶

Submodules¶

teili.building_blocks.building_block module¶

This module provides the parent class for (neuronal) building blocks.

Todo

Hierarchy of building_blocks.
There is a problem when 2 BBs of the same kind are added to another BB, as the names collide.
Monitor naming.
Important all Monitors of Building blocks have to be named and named uniquely! Otherwise they will not be found, when a Network is rerun in standalone mode after rebuild without recompile

class teili.building_blocks.building_block.BuildingBlock(*args, **kw)[source]¶

Bases: brian2.core.names.Nameable

This class is the parent class to all building blocks, e.g. WTA, SOM.

name¶

Name of the building_block population

Type: str, required

neuron_eq_builder¶

neuron class as imported from models/neuron_models

Type: class, optional

synapse_eq_builder¶

synapse class as imported from models/synapse_models

Type: class, optional

params¶

Dictionary containing all relevant parameters for each building block

Type: dictionary, optional

verbose¶

Flag to gain additional information

Type: bool, optional

groups¶

Keys to all neuron and synapse groups

Type: dictionary

monitors¶

Keys to all spike and state monitors

Type: dictionary

monitor¶

Flag to auto-generate spike and state monitors

Type: bool, optional

standalone_params¶

Dictionary for all parameters to create a standalone network

Type: dictionary

sub_blocks¶

Dictionary for all parent building blocks

Type: dictionary

input_groups¶

Dictionary containing all possible groups which are potential inputs

Type: dictionary

output_groups¶

Dictionary containing all possible groups which are potential outputs

Type: dictionary

hidden_groups¶

Dictionary containing all remaining groups which are neither inputs nor outputs

Type: dictionary

__iter__()[source]¶

this allows us to iterate over the BrianObjects and directly add the Block to a Network

Returns: Returns a dictionary wich contains all brian objects
Return type: TYPE

_set_tags(tags, target_group)[source]¶

This method allows the user to set a list of tags to a specific target group. Normally the tags are already assigned by each building block. So this method only adds convinience and a way to replace the default tags if this is needed by any user. Typically this should not be the user’s concern, that’s why it is private method.

Parameters

tags (dict) – A dictionary of tags {‘mismatch’ : False, ‘noise’ : False, ‘level’: 0 , ‘sign’: ‘None’, ‘target sign’: ‘None’, ‘num_inputs’ : 0, ‘bb_type’ : ‘None’, ‘group_type’ : ‘None’, ‘connection_type’ : ‘None’, }
target_group (str) – Name of group to set tags

get_groups(tags)[source]¶

Get all groups which have a certain set of tags

Parameters

tags (dict) – A dictionary of tags

Returns

List of all group objects which: share the same tags as specified.

Return type

target_dict (dict)

get_run_args()[source]¶

this collects the arguments to cpp main() for standalone run

Returns: Arguments which can be changed during/between runs
Return type: TYPE

get_tags(target_group)[source]¶

Get the currently set tags for a given group.

Parameters

target_group (str) – Name of group to get tags from or direct TeiliGroup

Returns

Dictionary containing all assigned _tags of provided: group

Return type

(dict)

property groups¶

This property will collect all available groups from the respective building block. The property follows a recursive strategy to collect all available groups. The intention is to easily update all available groups for stacked building blocks. NOTE Avoid any kind of loops between Building Blocks. Loops are forbidden as they lead to infinite collection of groups.

Returns: Dictionary containing all groups of all sub_blocks
Return type: tmp_groups (dict)

print_tags(target_group)[source]¶

Get the currently set tags for a given group.

Parameters: target_group (str) – Name of group to get tags from

teili.building_blocks.chain module¶

This is a simple syn-fire chain of neurons

teili.building_blocks.chain.chain_params¶

Dictionary of default parameters for syn-fire chain

Type: dict

Todo

Update docstrings

Not all Description of attributes are set. Please provide meaningful
docstrings

Example

To use the syn-fire chain building block in your simulation you need to create an object of the class by:

>>> from teili.building_blocks.chain import Chain
>>> my_bb = Chain(name='my_chain')

If you want to change the underlying neuron and synapse model you need to provide a different equation_builder class:

>>> from teili.models.neuron_models import DPI
>>> from teili.models.synapse_models import DPISyn
>>> my_bb = Chain(name='my_chain',
                  neuron_eq_builder=DPI,
                  synapse_eq_builder=DPISyn)

If you want to change the default parameters of your building block you need to define a dictionary, which you pass to the building_block

>>> chain_params = {'num_chains': 4,
                    'num_neurons_per_chain': 15,
                    'synChaCha1e_weight': 4,
                    'synInpCha1e_weight': 1,
                    'gChaGroup_refP': 1 * ms}
>>> my_bb = Chain(name='my_chain', block_params=chain_params)

class teili.building_blocks.chain.Chain(*args, **kw)[source]¶

Bases: teili.building_blocks.building_block.BuildingBlock

This is a simple syn-fire chain of neurons.

gChaGroup_refP¶

Parameter specifying the refractory period.

Type: str, optional

group¶

List of keys of neuron population.

Type: dict

inputGroup¶

SpikeGenerator obj. to stimulate syn-fire chain.

Type: SpikeGenerator

num_chains¶

Number of chains to generate.

Type: int, optional

num_neurons_per_chain¶

Number of neurons within one chain.

Type: int, optional

spikemon_cha¶

Description.

Type: brian2 SpikeMonitor obj.

spikemon_cha_inp¶

Description.

Type: brian2 SpikeMonitor obj.

standalone_params¶

Keys for all standalone parameters necessary for cpp code generation.

Type: dict

synapse¶

Description.

Type: TYPE

synChaCha1e_weight¶

Parameter specifying the recurrent weight.

Type: int, optional

synInpCha1e_weight¶

Parameter specifying the input weight.

Type: int, optional

plot(savedir=None)[source]¶

Simple function to plot recorded state and spikemonitors.

Parameters: savedir (str, optional) – Path to directory to save plot.
Returns: Returns figure.
Return type: matplotlib.pyplot object

teili.building_blocks.chain.gen_chain(groupname='Cha', neuron_eq_builder=<teili.models.neuron_models.ExpAdaptIF object>, synapse_eq_builder=<teili.models.synapse_models.ReversalSynV object>, num_chains=4, num_neurons_per_chain=15, num_inputs=1, synChaCha1e_weight=4, synInpCha1e_weight=1, gChaGroup_refP=1. * msecond, debug=False)[source]¶

Creates chains of neurons

Parameters

groupname (str, optional) – Base name for building block.
neuron_eq_builder (TYPE, optional) – Neuron equation builder object.
synapse_eq_builder (TYPE, optional) – Synapse equation builder object.
num_chains (int, optional) – Number of chains to generate.
num_neurons_per_chain (int, optional) – Number of neurons within one chain.
num_inputs (int, optional) – Number of inputs from different source populations.
synChaCha1e_weight (int, optional) – Parameter specifying the recurrent weight.
synInpCha1e_weight (int, optional) – Parameter specifying the input weight.
gChaGroup_refP (TYPE, optional) – Parameter specifying the refractory period.
debug (bool, optional) – Debug flag.

Returns

Keys to all neuron and synapse groups. Monitors (dictionary): Keys to all spike- and statemonitors. standalone_params (dictionary): Dictionary which holds all parameters to create a standalone network.

Return type

Groups (dictionary)

teili.building_blocks.reservoir module¶

This module provides a reservoir network, as described in Nicola and Clopath 2017.

teili.building_blocks.reservoir.reservoir_params¶

Dictionary of default parameters for reservoir.

Type: dict

Example

To use the Reservoir building block in your simulation you need to create an object of the class by:

>>> from teili.building_blocks.reservoir import Reservoir
>>> my_bb = Reservoir(name='my_reservoir')

If you want to change the underlying neuron and synapse model you need to provide a different equation_builder class:

>>> from teili.models.neuron_models import DPI as neuron_model
>>> from teili.models.synapse_models import DPISyn as synapse_model
>>> my_bb = Reservoir(name='my_reservoir',
                  neuron_eq_builder=DPI,
                  synapse_eq_builder=DPISyn)

If you want to change the default parameters of your building block you need to define a dictionary, which you pass to the building_block

>>> reservoir_params = {'weInpR': 1.5,
                        'weRInh': 1,
                        'wiInhR': -1,
                        'weRR': 0.5,
                        'sigm': 3,
                        'rpR': 0 * ms,
                        'rpInh': 0 * ms
                        }
>>> my_bb = Reservoir(name='my_reservoir', block_params=reservoir_params)

class teili.building_blocks.reservoir.Reservoir(*args, **kw)[source]¶

Bases: teili.building_blocks.building_block.BuildingBlock

A recurrent Neural Net implementing a Reservoir.

group¶

List of keys of neuron population.

Type: dict

input_group¶

SpikeGenerator object to stimulate Reservoir.

Type: SpikeGenerator

num_neurons¶

Size of Reservoir neuron population.

Type: int, optional

fraction_inh_neurons¶

Set to None to skip Dale’s principle.

Type: float, optional

spikemonR¶

A spikemonitor which monitors the activity of the reservoir population.

Type: brian2.SpikeMonitor object

standalone_params¶

Keys for all standalone parameters necessary for cpp code generation.

Type: dict

teili.building_blocks.reservoir.gen_reservoir(groupname, neuron_eq_builder=<class 'teili.models.neuron_models.Izhikevich'>, synapse_eq_builder=<class 'teili.models.synapse_models.DoubleExponential'>, neuron_model_params={'Cm': 250. * pfarad, 'Iconst': 0. * amp, 'Ileak': 0. * amp, 'Inoise': 0. * amp, 'VR': -60. * mvolt, 'VT': -20. * mvolt, 'a': 10. * hertz, 'b': 0. * siemens, 'c': -65. * mvolt, 'd': 200. * pamp, 'k': 2.5e-09 * metre ** -4 * kilogram ** -2 * second ** 6 * amp ** 3, 'refP': 1. * msecond}, weInpR=1.5, weRInh=1, wiInhR=-1, weRR=0.5, sigm=3, rpR=0. * second, rpInh=0. * second, num_neurons=64, Rconn_prob=None, adjecency_mtr=None, num_input_neurons=0, num_output_neurons=1, output_weights_init=[0], taud=0, taur=0, num_inputs=1, fraction_inh_neurons=0.2, spatial_kernel='kernel_mexican_1d', monitor=True, additional_statevars=[], debug=False)[source]¶

Generates a reservoir network.

Parameters

groupname (str, required) – Name of the Reservoir population.
neuron_eq_builder (class, optional) – neuron class as imported from models/neuron_models.
synapse_eq_builder (class, optional) – synapse class as imported from models/synapse_models.
weInpR (float, optional) – Excitatory synaptic weight between input SpikeGenerator and Reservoir neurons.
weRInh (int, optional) – Excitatory synaptic weight between Reservoir population and inhibitory interneuron.
wiInhR (TYPE, optional) – Inhibitory synaptic weight between inhibitory interneuron and Reservoir population.
weRR (float, optional) – Self-excitatory synaptic weight (Reservoir).
sigm (int, optional) – Standard deviation in number of neurons for Gaussian connectivity kernel.
rpR (float, optional) – Refractory period of Reservoir neurons.
rpInh (float, optional) – Refractory period of inhibitory neurons.
num_neurons (int, optional) – Size of Reservoir neuron population.
Rconn_prob (float, optional) – Float 0<p<1 to set connection probability within the reservoir
adjecency_mtr (numpy ndarray 3D of int and float, optional) – Uses the adjecency matrix to set connections in the reservoir
fraction_inh_neurons (int, optional) – Set to None to skip Dale’s principle.
cutoff (int, optional) – Radius of self-excitation.
num_inputs (int, optional) – Number of input currents to each neuron in the Reservoir.
num_input_neurons (int, optional) – Number of neurons in the input stage.
num_readout_neurons (int, optional) – Number of neurons in the readout stage.
spatial_kernel (str, optional) – None defaults to kernel_mexican_1d
monitor (bool, optional) – Flag to auto-generate spike and statemonitors.
additional_statevars (list, optional) – List of additional state variables which are not standard.
debug (bool, optional) – Flag to gain additional information.

Returns

Keys to all neuron and synapse groups. Monitors (dictionary): Keys to all spike- and statemonitors. standalone_params (dictionary): Dictionary which holds all parameters to create a standalone network.

Return type

Groups (dictionary)

teili.building_blocks.sequence_learning module¶

This module provides a sequence learning building block. This building block can be used to learn sequences of items

teili.building_blocks.sequence_learning.sl_params¶

Dictionary of default parameters for reservoir.

Type: dict

Example

To use the Reservoir building block in your simulation you need to create an object of the class by doing:

>>> from teili.building_blocks.reservoir import Reservoir
>>> my_bb = Reservoir(name='my_sequence')

If you want to change the underlying neuron and synapse model you need to provide a different equation_builder class:

>>> from teili.models.neuron_models import DPI
>>> from teili.models.synapse_models import DPISyn
>>> my_bb = Reservoir(name='my_sequence',
                  neuron_eq_builder=DPI,
                  synapse_eq_builder=DPISyn)

If you want to change the default parameters of your building block you need to define a dictionary, which you pass to the building_block:

>>> sl_params = {'synInpOrd1e_weight': 1.3,
                 'synOrdMem1e_weight': 1.1,
                 'synMemOrd1e_weight': 0.16,
                 # local
                 'synOrdOrd1e_weight': 1.04,
                 'synMemMem1e_weight': 1.54,
                 # inhibitory
                 'synOrdOrd1i_weight': -1.95,
                 'synMemOrd1i_weight': -0.384,
                 'synCoSOrd1i_weight': -1.14,
                 'synResetOrd1i_weight': -1.44,
                 'synResetMem1i_weight': -2.6,
                 # refractory
                 'gOrdGroups_refP': 1.7 * ms,
                 'gMemGroups_refP': 2.3 * ms
                 }
>>> my_bb = Reservoir(name='my_sequence', block_params=sl_params)

class teili.building_blocks.sequence_learning.SequenceLearning(*args, **kw)[source]¶

Bases: teili.building_blocks.building_block.BuildingBlock

Sequence Learning Network.

cos_group¶

Condition of Satisfaction group.

Type: neuron group

group¶

List of keys of neuron population.

Type: dict

input_group¶

SpikeGenerator object to stimulate Reservoir.

Type: SpikeGenerator

reset_group¶

Reset group, to reset network after CoS is met.

Type: neuron group

standalone_params¶

Keys for all standalone parameters necessary for cpp code generation.

Type: dict

plot(duration=None)[source]¶

Simple plot for sequence learning network.

Returns: The window containing the plot.
Return type: pyqtgraph window

teili.building_blocks.sequence_learning.gen_sequence_learning(groupname='Seq', neuron_eq_builder=<class 'teili.models.neuron_models.ExpAdaptIF'>, synapse_eq_builder=<class 'teili.models.synapse_models.ReversalSynV'>, num_elements=4, num_neurons_per_group=8, synInpOrd1e_weight=1.3, synOrdMem1e_weight=1.1, synMemOrd1e_weight=0.16, synOrdOrd1e_weight=1.04, synMemMem1e_weight=1.54, synOrdOrd1i_weight=-1.95, synMemOrd1i_weight=-0.384, synCoSOrd1i_weight=-1.14, synResetOrd1i_weight=-1.44, synResetMem1i_weight=-2.6, gOrdGroups_refP=1.7 * msecond, gMemGroups_refP=2.3 * msecond, num_inputs=1, verbose=False)[source]¶

Create Sequence Learning Network after the model from Sandamirskaya and Schoener (2010).

Parameters

groupname (str, optional) – Base name for building block.
neuron_eq_builder (teili.models.builder obj, optional) – Neuron equation builder object.
synapse_eq_builder (teili.models.builder obj, optional) – Synapse equation builder object.
num_elements (int, optional) – Number of elements in the sequence.
num_neurons_per_group (int, optional) – Number of neurons used to remember each item.
synInpOrd1e_weight (float, optional) – Parameter specifying the input weight.
synOrdMem1e_weight (float, optional) – Parameter specifying the ordinary to memory weight.
synMemOrd1e_weight (float, optional) – Parameter specifying the memory to ordinary weight.
synOrdOrd1e_weight (float, optional) – Parameter specifying the recurrent weight (ord).
synMemMem1e_weight (float, optional) – Parameter specifying the recurrent weight (memory).
synOrdOrd1i_weight (TYPE, optional) – Parameter specifying the recurrent inhibitory weight.
synMemOrd1i_weight (TYPE, optional) – Parameter specifying the memory to ordinary inhibitory weight.
synCoSOrd1i_weight (TYPE, optional) – Parameter specifying the inhibitory weight from cos to ord.
synResetOrd1i_weight (TYPE, optional) – Parameter specifying the the inhibitory weight from reset to ord.
synResetMem1i_weight (TYPE, optional) – Parameter specifying the the inhibitory weight from reset cos to memory.
gOrdGroups_refP (TYPE, optional) – Parameter specifying the refractory period.
gMemGroups_refP (TYPE, optional) – Parameter specifying the refractory period.
num_inputs (int, optional) – Number of inputs from different source populations.
debug (bool, optional) – Debug flag.

Returns

Keys to all neuron and synapse groups. Monitors (dictionary): Keys to all spike- and statemonitors. standalone_params (dictionary): Dictionary which holds all parameters to create a standalone network.

Return type

Groups (dictionary)

teili.building_blocks.sequence_learning.plot_sequence_learning(Monitors, duration=None)[source]¶

A simple matplotlib wrapper function to plot network activity.

Parameters: Monitors (building_block.monitors) – Dictionary containing all monitors created by gen_sequence_learning().
Returns: Matplotlib figure.
Return type: plt.fig

teili.building_blocks.wta module¶

This module provides different Winner-Takes_all (WTA) circuits.

Beside different dimensionality of the WTA, i.e 1D & 2D, you can select different spatial connectivity and neuron and synapse models.

teili.building_blocks.wta.wta_params¶

Dictionary of default parameters for wta.

Type: dict

Todo

Generalize for n dimensions

Example

To use the WTA building block in your simulation you need to create an object of the class by doing:

>>> from teili.building_blocks.wta import WTA
>>> my_bb = WTA(name='my_wta')

If you want to change the underlying neuron and synapse model you need to provide a different equation_builder class:

>>> from teili.models.neuron_models import ExpAdaptIF
>>> from teili.models.synapse_models import ReversalSynV
>>> my_bb = WTA(name='my_wta',
                neuron_eq_builder=ExpAdaptIF,
                synapse_eq_builder=ReversalSynV)

If you want to change the default parameters of your building block you need to define a dictionary, which you pass to the building_block:

>>> wta_params = {'we_inp_exc': 1.5,
                'we_exc_inh': 1,
                'wi_inh_exc': -1,
                'we_exc_exc': 0.5,
                'wi_inh_inh': -1,
                'sigm': 3,
                'rp_exc': 3 * ms,
                'rp_inh': 1 * ms,
                'ei_connection_probability': 1,
                'ie_connection_probability': 1,
                'ii_connection_probability': 0
                }
    >>> my_bb = WTA(name='my_wta', block_params=wta_params)

class teili.building_blocks.wta.WTA(*args, **kw)[source]¶

Bases: teili.building_blocks.building_block.BuildingBlock

A 1 or 2D square Winner-Takes_all (WTA) Building block.

dimensions¶

Specifies if 1 or 2 dimensional WTA is created.

Type: int, optional

num_neurons¶

Size of WTA neuron population.

Type: int, optional

spike_gen¶

SpikeGenerator group of the BB

Type: brian2.Spikegenerator obj.

spikemon_exc¶

A spike monitor which monitors the activity of the WTA population.

Type: brian2.SpikeMonitor obj.

standalone_params¶

Dictionary of parameters to be changed after standalone code generation.

Type: dict

teili.building_blocks.wta.gen1dWTA(groupname, neuron_eq_builder=<class 'teili.models.neuron_models.DPI'>, synapse_eq_builder=<class 'teili.models.synapse_models.DPISyn'>, we_inp_exc=1.5, we_exc_inh=1, wi_inh_inh=-1, wi_inh_exc=-1, we_exc_exc=0.5, sigm=3, rp_exc=3. * msecond, rp_inh=1. * msecond, num_neurons=64, num_inh_neurons=5, num_input_neurons=None, num_inputs=1, num_inh_inputs=2, cutoff=10, spatial_kernel='kernel_gauss_1d', ei_connection_probability=1, ie_connection_probability=1, ii_connection_probability=0, additional_statevars=[], monitor=True, verbose=False)[source]¶

Creates a 1D WTA population of neurons, including the inhibitory interneuron population

Parameters

groupname (str, required) – Name of the WTA population.
neuron_eq_builder (class, optional) – neuron class as imported from models/neuron_models.
synapse_eq_builder (class, optional) – synapse class as imported from models/synapse_models.
we_inp_exc (float, optional) – Excitatory synaptic weight between input SpikeGenerator and WTA neurons.
we_exc_inh (float, optional) – Excitatory synaptic weight between WTA population and inhibitory interneuron.
wi_inh_inh (float, optional) – Inhibitory synaptic weight between interneurons.
wi_inh_exc (float, optional) – Inhibitory synaptic weight between inhibitory interneuron and WTA population.
we_exc_exc (float, optional) – Self-excitatory synaptic weight (WTA).
sigm (int, optional) – Standard deviation in number of neurons for Gaussian connectivity kernel.
rp_exc (float, optional) – Refractory period of WTA neurons.
rp_inh (float, optional) – Refractory period of inhibitory neurons.
num_neurons (int, optional) – Size of WTA neuron population.
num_inh_neurons (int, optional) – Size of inhibitory interneuron population.
num_input_neurons (int, optional) – Size of input population. If None, equal to size of WTA population.
num_inputs (int, optional) – Number of input currents to WTA.
num_inh_inputs (int, optional) – Number of input currents to the inhibitory group.
cutoff (int, optional) – Radius of self-excitation.
spatial_kernel (str, optional) – Connectivity kernel for lateral connectivity. Default is ‘kernel_gauss_1d’. See tools.synaptic_kernel for more detail.
ei_connection_probability (float, optional) – WTA to interneuron connectivity probability.
ie_connection_probability (float, optional) – Interneuron to WTA connectivity probability
ii_connection_probability (float, optional) – Interneuron to Interneuron connectivity probability.
additional_statevars (list, optional) – List of additional state variables which are not standard.
monitor (bool, optional) – Flag to auto-generate spike and state monitors.
verbose (bool, optional) – Flag to gain additional information.

Returns

Keys to all neuron and synapse groups. monitors (dictionary): Keys to all spike and state monitors. standalone_params (dictionary): Dictionary which holds all

parameters to create a standalone network.

Return type

_groups (dictionary)

teili.building_blocks.wta.gen2dWTA(groupname, neuron_eq_builder=<class 'teili.models.neuron_models.DPI'>, synapse_eq_builder=<class 'teili.models.synapse_models.DPISyn'>, we_inp_exc=1.5, we_exc_inh=1, wi_inh_inh=-1, wi_inh_exc=-1, we_exc_exc=2.0, sigm=2.5, rp_exc=3. * msecond, rp_inh=1. * msecond, num_neurons=64, num_inh_neurons=5, num_input_neurons=None, num_inputs=1, num_inh_inputs=2, cutoff=10, spatial_kernel='kernel_gauss_2d', ei_connection_probability=1.0, ie_connection_probability=1.0, ii_connection_probability=0.1, additional_statevars=[], monitor=True, verbose=False)[source]¶

Creates a 2D square WTA population of neurons, including the inhibitory interneuron population

Parameters

groupname (str, required) – Name of the WTA population.
neuron_eq_builder (class, optional) – neuron class as imported from models/neuron_models.
synapse_eq_builder (class, optional) – synapse class as imported from models/synapse_models.
we_inp_exc (float, optional) – Excitatory synaptic weight between input SpikeGenerator and WTA neurons.
we_exc_inh (int, optional) – Excitatory synaptic weight between WTA population and inhibitory interneuron.
wi_inh_inh (int, optional) – Self-inhibitory weight of the interneuron population.
wi_inh_exc (TYPE, optional) – Inhibitory synaptic weight between inhibitory interneuron and WTA population.
we_exc_exc (float, optional) – Self-excitatory synaptic weight (WTA).
sigm (int, optional) – Standard deviation in number of neurons for Gaussian connectivity kernel.
rp_exc (float, optional) – Refractory period of WTA neurons.
rp_inh (float, optional) – Refractory period of inhibitory neurons.
num_neurons (int, optional) – Size of WTA neuron population.
num_inh_neurons (int, optional) – Size of inhibitory interneuron population.
num_input_neurons (int, optional) – Size of input population. If None, equal to size of WTA population.
num_inputs (int, optional) – Number of input currents to WTA.
num_inh_inputs (int, optional) – Number of input currents to the inhibitory group.
cutoff (int, optional) – Radius of self-excitation.
spatial_kernel (str, optional) – Description
ei_connection_probability (float, optional) – WTA to interneuron connectivity probability.
ie_connection_probability (float, optional) – Interneuron to excitory neuron connectivity probability.
ii_connection_probability (float, optional) – Interneuron to interneuron neuron connectivity probability.
additional_statevars (list, optional) – List of additional state variables which are not standard.
monitor (bool, optional) – Flag to auto-generate spike and statemonitors.
verbose (bool, optional) – Flag to gain additional information.

Returns

Keys to all neuron and synapse groups. monitors (dictionary): Keys to all spike and state monitors. standalone_params (dictionary): Dictionary which holds all

parameters to create a standalone network.

Return type

_groups (dictionary)

teili.building_blocks.wta.set_wta_tags(self, _groups)[source]¶

Sets default tags to a WTA network

Parameters: _groups (dictionary) – Keys to all neuron and synapse groups.

No Longer Returned:

Tags will be added to all _groups passed. They follow this structure:

tags = {‘mismatch’(bool, 0/1): ‘noise : (bool, 0/1) ‘level’: int ‘sign’: str (exc/inh/None) ‘target sign’: str (exc/inh/None) ‘num_inputs’ : int (0 if not Neuron group), ‘bb_type’ : str (WTA/3-WAY), ‘group_type’ : str (Neuron/Connection/ SpikeGen) ‘connection_type’ : str (rec/lateral/fb/ff/None)

}

teili.building_blocks package¶

Submodules¶

teili.building_blocks.building_block module¶

teili.building_blocks.chain module¶

teili.building_blocks.reservoir module¶

teili.building_blocks.sequence_learning module¶

teili.building_blocks.wta module¶

Module contents¶