accessing columns with JaggedArray elements without a for loop about awkward-0.x HOT 6 CLOSED

Thanks for posting— I intend to do this soon. I'm surprised that the second example doesn't already give you a jagged array, since it would be trivially returning self (a Numpy array would copy in that situation, but awkward arrays are immutable, so they return views in such situations).

from awkward-0.x.

Eek, apologies, I thought I'd already tried that call. The call

data["traces"][:]

does in fact return a <class 'awkward.array.jagged.JaggedArray'> that I can iterate over. However, I run into an error when I iterate. I think this is because of how I'm loading my data into awkward-array; the last record is different from all the others and doesn't have digitized data. I'll try to confirm that's what's going on.

Meanwhile, here's the behavior I see:

traces = data["traces"][:]

for trace in traces:
    # length of a trace is always 256
    if numpy.size(trace) != 256:
        print("different size")
        break

from awkward-0.x.

A quicker way:

(traces.counts == 256).all()

would return True if all traces happen to have size 256. traces.counts is just a Numpy array.

from awkward-0.x.

Yup, so

(traces.counts == 256).all()

returns False on my data.

Should I open another issue, asking how to sensibly store what's essentially a different kind of data in my awkward-array structure?

from awkward-0.x.

Sure. It sounds like all of this would be fixed by documentation, of course. :)

from awkward-0.x.

Fixed by #34. You should be able to do subslices now:

jaggedarray[:, start:end]

which always return JaggedArrays because start:end might not be within the second axis range for all elements.

There's also

jaggedarray.regular()

which attempts to turn the jagged array into a regular, non-jagged array, throwing an error if not all subentries have the same shape. In your data, you should be able to

jaggedarray[:-1].regular()

because all of your traces have the same size except for the last, which is empty. Actually, having the regular() function makes subslices irrelevant for your case (because you can do normal Numpy axis subslicing), but anyway, you have choices. :)

from awkward-0.x.