Etienne Ackermann

This way, we should be able to pass interpenetrate Primary and SecondaryThresholds

Requesting multiple epochs, including some that don't exist, returns empty EpochArrays

1

```python npl.plot(signal[:,[0,2,3]]) ``` works, and so does ```python npl.plot(signal[[0,2,3],:]) ``` but ```python npl.plot(signal[[0,1,30],:]) ``` does not, and instead of "ignoring" the 30th epoch (which doesn't exist), it returns an empty...

support adjacent epochs with the same timestamps

2

`test = nel.AnalogSignalArray([1,1,1,5,5,5], tdata=[1,2,3,3,4,5], support=nel.EpochArray([[1,3],[3,5]]))` how can we force test[0] to correspond to the first 3 samples, and test[1] to correspond to the last 3 samples? This seems highly non-trivial...

Finish spatial information calculations

1

Currently, both spatial information (in bits) and sparsity (in percent) exist in nelpy and can easily be computed on a `TuningCurve1D` object. However, according to Markus et al. (1994) we...

decoding error with xval for single epoch BST

1

Currently we train and evaluate using the events in BST. If BST only has one event, then it could make sense to split the data into chunks ourselves. Actually, it...

need to write comprehensive tests for data-dependencies

2

As an example, in `AnalogSignalArray`, we have ```python ... def __init__(...): ... self._ydata = ydata ... ``` So what happens if we instantiate an object, e.g., `my_asa = nel.AnalogSignalArray(my_ydata)` and...

better handling of plotting **kwargs

1

instead of only catching some kwargs, we should update a dict

create object for decoded output

2

container for posteriors, that would also contain boundaries, methods for mean_pth and mode_pth, and would greatly facilitate plotting posteriors, with spike rasters and/or LFP and other AnalogSignalArrays on top of,...

Currently, we plot in numerical order by default. I think it makes much more sense to plot in object-order instead, especially if it's a new plot. If there are existing...