braintools.input.Transformed

class braintools.input.Transformed(input_obj, func)

Custom transformation applied to an input.

Applies an arbitrary function to transform an input, enabling custom nonlinearities and processing.

Parameters:

• input_obj (Input) – The input to transform.

• func (Callable) – Function to apply to the array. Should accept and return an array-like object.

Examples

>>> import jax.numpy as jnp
>>> sine = Sinusoidal(1.0, 10 * u.Hz, 200 * u.ms)
>>>
>>> # Half-wave rectification
>>> rectified = Transformed(sine, lambda x: jnp.maximum(x, 0))
>>>
>>> # Squaring (frequency doubling)
>>> squared = Transformed(sine, lambda x: x ** 2)
>>>
>>> # Sigmoid nonlinearity
>>> sigmoid = Transformed(sine,
...     lambda x: 1 / (1 + jnp.exp(-10 * x)))
>>>
>>> # Usually created via apply() method
>>> transformed = sine.apply(lambda x: jnp.abs(x))

__init__(input_obj, func)

Initialize transformed input.

Parameters:

• input_obj (Input) – The input to transform.

• func (Callable) – Function to apply to the array.

Methods

__init__(input_obj, func)	Initialize transformed input.
apply(func)	Apply a custom function to the input.
clip([min_val, max_val])	Clip the input values to a range.
generate()	Generate the transformed input.
repeat(n_times)	Repeat the input pattern n times.
scale(factor)	Scale the input by a factor.
shift(time_shift)	Shift the input in time.
smooth(tau)	Apply exponential smoothing to the input.

Attributes

dt	Get the time step from global environment.
n_steps	Get the number of time steps.
shape	Get the shape of the input array.