Hi!
I am trying to understand Nengo’s connections in this example of a simple harmonic oscillator. Following this example, I created an ensemble with 10 neurons representing 2 dimensions. I also manually fixed tuning curves for every neuron, which is shown in the figure below.
To understand connections in Nengo, I took the encoding matrix (which is 10 by 2) and directed connection weights (10 by 10), and put it in another network, whose structure is shown below.
In this network, instead of setting an ensemble with 10 neurons representing 2 dimensions, I created an ensemble with 10 neurons representing 10 dimensions. The mapping between the neurons and the dimensions is one-to-one, suggested by the identical encoding matrix. The self-loop connection of the ensemble was set to be the directed connection weights probed out of the previous network.
conn = nengo.Connection(neurons, neurons, transform=direct_weight,synapse=0.001,seed=r_seed)
In the new network, I also projected the output of the ensemble to a 2D space by using the encoding matrix (which is 10 by 2) in the previous network.
I would expect the two networks had the same output of dynamics, however, the second network did not output harmonic oscillation as the first network did. I was wondering what might be the problem here.