Implementing build_tensor and warp_tensor in warping.py

[gviejo]

Here is the proposed behavior :

Simple trial-based tensor :

import pynapple as nap
import numpy as np
group = nap.TsGroup({0:nap.Ts(t=np.arange(0, 100))})
ep = nap.IntervalSet(start=np.arange(20, 100, 20), end=np.arange(20, 100, 20) + np.arange(2, 10, 2))
tensor = nap.build_tensor(group, ep, binsize=1)
print(tensor)
    array([[[ 1.,  1., nan, nan, nan, nan, nan, nan],
            [ 1.,  1.,  1.,  1., nan, nan, nan, nan],
            [ 1.,  1.,  1.,  1.,  1.,  1., nan, nan],
            [ 1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.]]])

Time-warped trial based tensor :

import pynapple as nap
import numpy as np
group = nap.TsGroup({0:nap.Ts(t=np.arange(0, 100))})
ep = nap.IntervalSet(start=np.arange(20, 100, 20), end=np.arange(20, 100, 20) + np.arange(2, 10, 2))
tensor = nap.warp_tensor(group, ep, num_bin=10)
print(tensor)
    array([[[1., 0., 0., 0., 0., 1., 0., 0., 0., 0.],
            [1., 0., 1., 0., 0., 1., 0., 1., 0., 0.],
            [1., 1., 0., 1., 0., 1., 1., 0., 1., 0.],
            [1., 1., 1., 1., 0., 1., 1., 1., 1., 0.]]])

Both functions works for any time series.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Files with missing lines	Coverage Δ
pynapple/core/base_class.py	97.60% <100.00%> (ø)
pynapple/core/ts_group.py	92.57% <100.00%> (-0.29%)	⬇️
pynapple/process/__init__.py	100.00% <100.00%> (ø)
pynapple/process/perievent.py	100.00% <100.00%> (+5.20%)	⬆️
pynapple/process/warping.py	100.00% <100.00%> (ø)

[sjvenditto]

I have some thoughts about the naming conventions used here:

1. Having warp_tensor as a separate function from build_tensor feels a little misleading since warp_tensor also "builds" a tensor.

My thought is to have a central function build_tensor with a method parameter that can be "pad" or "warp", that takes method_kwargs, so something like:

def build_tensor(input, ep, method, **method_kwargs)

and that way both ways of building a tensor can be accessed in a centralized location. In this way, the current build_tensor would be renamed to pad_tensor, and keep the kwargs definitions in the docstring. warp_tensor could remain as-is

2. Similar to numpy functions like np.reshape, I think it would be useful to define an object method to_tensor that calls build_tensor

What do you think about this?