Andrew Davison

Create conda package

2

and put on binstar

Suggestion: rename `IDMixin` to `ID`

6

I noticed when debugging a script using spyNNaker that objects of type "IDMixin" appear where I would expect to see type "ID". The name "IDMixin" is not intended to be...

RNG error with FixedTotalNumberConnector

3

This line: https://github.com/SpiNNakerManchester/sPyNNaker/blob/master/spynnaker/pyNN/models/neural_projections/connectors/multapse_connector.py#L68 calls the PyNN RNG object with a distribution name 'multinomial' and parameters in the form of a list. As of PyNN 0.8, the RNGs support only a...

Incorrect interface for FixedTotalNumberConnector

4

`FixedTotalNumberConnector` should take an argument `n`, for the number of connections. In spyNNaker8, `FixedTotalNumberConnector` is an alias for `MultapseConnector`, which has an argument `num_synapses`. I guess `FixedTotalNumberConnector` should instead sub-class...

Need to check that the dtypes in HDF5 files created by the `sonata` module match the spec at https://github.com/AllenInstitute/sonata/blob/master/docs/SONATA_DEVELOPER_GUIDE.md#network_config

Add Kobayashi neuron model

3

Kobayashi R, Tsubo Y and Shinomoto S (2009) Made-to-order spiking neuron model equipped with a multi-timescale adaptive threshold. Front. Comput. Neurosci. 3:9. http://dx.doi.org/10.3389/neuro.10.009.2009 `mat2_psc_exp` in NEST. Implementations for Brian and...

script hangs with mpirun

1

_Eloy Retamino commented:_ An issue was detected with latest https://github.com/NeuralEnsemble/PyNN/tree/nest-dev + NEST 2.18: running the script [braitenberg_distributed.py](https://gitlab.humanbrainproject.org/joint_infrastructure_coordination/Coordination/uploads/b427973c9a411950bc81515662f60481/braitenberg_distributed.py) works fine: python braitenberg_distributed.py running it with mpirun hangs (concretely after line 99...

Fix StepCurrentSource

2

The test `issue512` in `test_electrodes.py` has many conditional blocks for different backends. `step.times` and `step.amplitudes` should give identical results for all backends, even if the internal representations are different, so...