TEST add-nest-example-scripts

[HarshKhilawala]

Balanced Neuron Example
IAF neuron example with current generator

Resultant inhibitory frequency differs from NEST example for Balanced Neuron Example

[HarshKhilawala]

@apdavison Added another script as per this: Link

[coveralls]

coverage: 63.264% (-0.07%) from 63.329%
when pulling 56362b0 on HarshKhilawala:test-add-balanced-neuron
into 1a1ce5c on NeuralEnsemble:master.

[apdavison]

The comments from #801 also apply to the additional examples here:

• All example scripts should work with the nest, neuron and brian2 backends.
• The backend to use should be specified on the command line, and retrieved using the get_simulator() function from pyNN.utility. See examples/small_network.py for an example of this.
• Use save(figure_filename) instead of show()
• The top-level docstring should follow the example in examples/small_network.py

[apdavison]

I suggest moving the network creation outside the function, as is done in the PyNEST implementation. Inside the function, you just need to change the input rate, use sim.run(...), then neuron.get_data(clear=True) (the clear=True argument removes the data from the population after you've retrieved them)

[HarshKhilawala]

@apdavison 'sim.reset()' doesn't work with NEST. This was one of the primary reason as to implement inside the function call.

Sim.run() will continue the simulation from previous checkpoint which is opposite to what I desired for.

[HarshKhilawala]

For balanced_neuron_example, my results still differ from native NEST example? Is it fine? Is it due to the core implementation being reason for two different resultant value?

[HarshKhilawala]

@apdavison So for one_neuron example, do you mean I develop three separate scripts each for 3 simulators? But Isn't this example a native NEST example?

If that is something you meant, Do I need to go through simple network example from each simulator and then implement script to support each of these simulators?

While adding examples to PyNN was just an approach to get familiarity with PyNN, if this is the case, I may not be able to complete the Primary Project which was to add e-props support to PyNN.

Would you please clarify a bit as I might have misunderstood?

[HarshKhilawala]

I just updated one_neuron_example.py as per your suggestion. Can you please confirm if it is 'merge ready example'? Is this what you meant when you said to have support for 3 backends?