skneuromsi.utils.neural_tools module

Tools for scikit-neuromsi neural models.

skneuromsi.utils.neural_tools.calculate_neural_distance(neurons, position_j, position_k)

Computes the distance between the pre-synaptic and post-synaptic neurons.

Designed for neurons encoding an external 1D space. It assumes that neurons are connected in a circular structure so that every neuron receives the same number of lateral connections.

Parameters:

• neurons (int) – The number of neurons.

• position_j (float) – Position of the post-synaptic neuron

• position_k (float) – Position of the pre-synaptic neuron

Returns:

Distance between the two neurons in model units.

Return type:

numpy.float64

skneuromsi.utils.neural_tools.calculate_lateral_synapses(neurons, excitation_loc, inhibition_loc, excitation_scale, inhibition_scale, dtype=<class 'numpy.float32'>)

Computes the values of lateral synapses.

Designed for a group of recurrently connected neurons following a “Mexican hat” distribution (a central excitatory zone surrounded by an inhibitory annulus) calculated as a substraction of two Gaussians.

Parameters:

• neurons (int) – The number of neurons.

• excitation_loc (float) – Loc of the excitatory Gaussian function.

• inhibition_loc (float) – Loc of the inhibitory Gaussian function.

• excitation_scale (float) – Scale of the excitatory Gaussian function.

• inhibition_scale (float) – Scale of the excitatory Gaussian function.

• dtype (numpy class) – Type of the array to store the values.

Returns:

The values of lateral synapses.

Return type:

numpy.array

skneuromsi.utils.neural_tools.calculate_inter_areal_synapses(neurons, weight, sigma, dtype=<class 'numpy.float32'>)

Computes the values of inter-areal synapses.

Designed for two connected group of neurons following a Gaussian function. It assumes symmetrical connectivity and the same number of neurons in both groups.

Parameters:

• neurons (int) – The number of neurons in each group.

• weight (float) – The highest level of synaptic efficacy.

• sigma (float) – The width of the Gaussian function.

• dtype (numpy class) – Type of the array to store the values.

Returns:

The value of inter-areal synapses.

Return type:

numpy.array

skneuromsi.utils.neural_tools.prune_synapses(synapses_weight_matrix, pruning_threshold)

Prunes neural connections.

Pruning is implemented by assigning zero to those synapses values below a given threshold.

Parameters:

• synapses_weight_matrix (numpy.array) – Array containing the values of the synapses.

• pruning_threshold (float) – Threshold value of the pruning procedure.

Returns:

The value of the external stimuli for each neuron.

Return type:

numpy.array

skneuromsi.utils.neural_tools.calculate_stimuli_input(neurons, intensity, *, scale, loc, dtype=<class 'numpy.float32'>)

Computes the values of stimuli as a spatial Gaussian function.

The Gaussian accounts for the uncertainty in the detection of stimuli.

Parameters:

• neurons (int) – The number of neurons to be stimulated.

• intensity (float) – The strength of the external stimulus.

• scale (float) – Scale of the Gaussian function.

• loc (float) – Loc of the Gaussian function.

• dtype (numpy class) – Type of the array to store the values.

Returns:

The value of the external stimuli for each neuron.

Return type:

numpy.array

skneuromsi.utils.neural_tools.create_unimodal_stimuli_matrix(neurons, stimuli, stimuli_duration, onset, simulation_length, time_res, dt, stimuli_n=1, soa=None, dtype=<class 'numpy.float32'>)

Creates the matrix of a unimodal stimuli for each neuron at each timepoint.

Supports multiple stimuli.

Parameters:

• neurons (int) – The number of neurons to be stimulated.

• stimuli (numpy.array) – Array with the values of the stimuli for each neuron.

• stimuli_duration (float) – Duration of the stimuli in model time units.

• onset (float) – Onset of the unimodal stimuli in model time units.

• simulation_length (float) – Total duration of the model run in model time units.

• time_res (float) – Temporal resolution of the model.

• dt (float) – Model integrator dt.

• stimuli_n (int) – Number of unimodal stimuli.

• soa (int) – Stimuli onset asynchrony. Relevant for more than 1 stimuli. Must be higher than stimulus_duration.

• dtype (numpy class) – Type of the array to store the values.

Returns:

The value of the external stimuli for each neuron.

Return type:

numpy.array

Raises:

ValueError – If the total duration of all stimuli plus the total duration of the inter-stimulus intervals (i.e., stimuli_duration * stimuli_n + soa * (stimuli_n - 1)) exceeds the simulation_length.

skneuromsi.utils.neural_tools.compute_latency(time, latency)

Computes the latency-adjusted time in the simulation.

Latency is computed by subtracting the given latency from the current time, ensuring that the result is not negative.

Parameters:

• time (float) – The current time in the simulation.

• latency (float) – The latency to subtract from the current time.

Returns:

The latency-adjusted time. If the result of the subtraction is negative, returns 0 instead.

Return type:

float

Notes

This method is used to adjust the time for processing delays in the simulation. It ensures that the computed latency does not result in a negative time value, which could be invalid for certain operations.