Any int or float cell or connection parameter should be replaceable by a RandomDistribution object, or by a list/array of ints/floats

[apdavison]

To allow easy specification of neuronal diversity, we should allow specification of int/float parameters using a pyNN.random.RandomDistribution object. The value for an individual cell/connection would be picked from this distribution.

---

Imported from Trac ticket:4

Opened: 2007-04-24 13:15:54
Last modified: 2010-12-09 20:59:03
Component: unspecified
Priority: major
Owner: Pierre
Reporter: apdavison

[apdavison]

apdavison wrote on 2007-11-06 00:06:47:

This can currently be done quite easily using the rset() method, so although it's a nice idea, it's far from urgent.

[apdavison]

Pierre wrote on 2008-01-18 00:03:41:

Now implemented for all the connector objects. Weights and Delays can be passed either as int/float or RandomDistribution objects. We need therefore to harmonize their conversions according to the simulators. Same with cell parameters. rset() will do the job, but we have to better check the conversion when the argument is a random distribution object

[apdavison]

apdavison wrote on 2010-03-02 11:16:42:

similarly, it should be possible to use an array/list of values, as in tset().

[apdavison]

Implemented in the parameter_space branch, now merged to master, will be released in PyNN 0.8. Note that Population.tset() and Population.rset() deprecated. Population.set() now accepts RandomDistribution objects or arrays as parameter values.