========================
PyNN 0.8.2 release notes
========================

6th December 2016

Welcome to PyNN 0.8.2!


New spike sources
-----------------

Two new spike source models were added, with implementations for the NEST and NEURON backends:
:class:`SpikeSourceGamma` (spikes follow a gamma process) and :class:`SpikeSourcePoissonRefractory`
(inter-spike intervals are drawn from an exponential distribution as for a Poisson process,
but there is a fixed refractory period after each spike during which no spike can occur).

Other changes
-------------

* Changed the :func:`save_positions()` format from ``id x y z`` to ``index x y z`` to make it simulator independent.
* Added histograms to the :mod:`utility.plotting` module.
* Added a `multiple_synapses` flag to :func:`Projection.get(..., format="array")` to control how
  synaptic parameters are combined when there are multiple connections between pairs of neurons.
  Until now, parameters were summed, which made sense for weights but not for delays.
  We have adopted the Brian_ approach of adding an argument ``multiple_synapses`` which is one
  of ``{'last', 'first', 'sum', 'min', 'max'}``. The default is ``sum``.
* `Assorted bug fixes`_



.. _Brian: http://briansimulator.org
.. _`Assorted bug fixes`: https://github.com/NeuralEnsemble/PyNN/issues?q=is%3Aclosed+milestone%3A0.8.2