Two dimensions, and no more loops
Stack 254 single trials into one table, average down the columns, and find out what this neuron is actually for — then find out why numpy is worth the trouble.
The second dimension
- ▸np.stack — a list of trials becomes a table
- ▸shape as (rows, columns) = (trials, frames)
- ▸trials[row, col] — and : for a whole column
- ▸axis=0 vs axis=1 — which dimension disappears
Vectorisation
- ▸the same job as a loop, in one line
- ▸time it yourself — tens of times faster
- ▸broadcasting, and the shape error you will hit
- ▸baseline-correct 254 trials without a loop