These are chat archives for thunder-project/thunder

Apr 2016
Apr 28 2016 06:48
@jwittenbach Great! Thank you.
AJ Keller
Apr 28 2016 16:01
does anyone know where i can find some DSP thunder examples?
specifically Series
I read all the docs and followed all the tutorials on
Jeremy Freeman
Apr 28 2016 16:06
@pushtheworldllc saw your pipeline description above, sounds cool!
what kinds of DSP are you thinking of?
getting the data from cassandra to thunder should be reasonably straightforward, though i don't know any examples, most direct way would probably be to get an RDD out and then do thunder.series.fromrdd
AJ Keller
Apr 28 2016 16:12
hey @freeman-lab right of the bat a 60Hz notch. for this one case of detecting eye blinks (to validate the system works) a simple voltage thresh hold would suffice
i also could tagged hte data in a way to isolate two classes, blinking and baseline or (not blinking)
AJ Keller
Apr 28 2016 16:36
i’m new to the idea of map functions
so it’s stumbling to see how those correlate into filters
Davis Bennett
Apr 28 2016 18:05
@pushtheworldllc here's an example for applying a digital filter to elements of a series object:
import thunder as td
from scipy.signal import bessel, filtfilt

# define the filter
b, a = bessel(4, .5)
ser = td.series.fromrandom(shape=(10, 1000))
filtered_series = v: filtfilt(b, a, v))
Jeremy Freeman
Apr 28 2016 18:08
@d-v-b great example!
yeah in general there's a lot you can do pretty easily by combining map with other tools
and we can definitely make methods for the most common and general-purpose operations
that's the idea behind things like convolve and detrend
Boaz Mohar
Apr 28 2016 18:46

@freeman-lab @jwittenbach Hi, is there a way in thunder 1.0 / bolt to transpose an images object so that the distributed axis will change. For example:

import thunder as td
data = td.images.fromrand(shape=(100, 3, 4, 5))

In this example I am changing the distributed axis from time points to planes.

AJ Keller
Apr 28 2016 18:52
@d-v-b thanks!