I’m looking at the nengo multiplication.ipynb example (also here). Why does the `combined`

ensemble have 220 neurons?

The incoming ensembles each have 100 neurons, so I would expect the combined to have 200 neurons instead.

I’m looking at the nengo multiplication.ipynb example (also here). Why does the `combined`

ensemble have 220 neurons?

The incoming ensembles each have 100 neurons, so I would expect the combined to have 200 neurons instead.

The number of neurons in the `combined`

ensemble is independent of the number of neurons in the input ensembles. Rather, the number of neurons in the `combined`

ensemble determines how well it is able to represent the 2-dimensional value required to perform the multiplication. Increasing the number of neurons in the `combined`

ensemble will increase the accuracy of the multiplication, and decreasing the number of neurons will decrease the accuracy of the multiplication.

As a side note, the multiplicative ensembles in Spaun use 150 neurons (with 50 neurons for the input ensembles).