nengo.Node Docs

[IgnacioRubioScola]

Hi,
I am learning how to use nengo and I think that the use of implicit time argument should be explained in the documentation of Node because it is non-trivial. I found these cases in a solved issue (#212), i don't know if it is completely correct, but help me to understand.

# Case 1: constant
n = nengo.Node([0,0,0])   
nengo.Connection(n, b, fiter=None)

# Case 2: time-varying
def my_function(t):
    return np.sin(t)
n = nengo.Node(my_function) 
nengo.Connection(n, b, filter=None)

# Case 3: time-and-state-varying
def my_function(t, x):
   return np.sin(t*x)
n = nengo.Node(my_function, dimensions=1)  
nengo.Connection(a, n, filter=None)
nengo.Connection(n, b, filter=None)

Thank you!

[hunse]

Thanks @IgnacioRubioScola, this is a good suggestion to add these details to the documentation. We've added this issue to our backlog, and plan to do this as part of the Nengo 4.0 release.

I think you've got the general idea correct. The one thing I noticed is that you've used dimensions instead of the correct size_in to set the size of your Node (we do this because a Node can have an input size that is different from its output size).

To summarize, node functions always take the simulation time t as the first input. If the node has a size_in != None then the function must also have a second argument x, which is the vector coming in to the node at the current time (the size of the vector equals size_in).