I would like some feedback on my choice of data structure. I have a 2D X-Y grid of current values for a specific voltage value. I have several voltage steps and have organized the data into a cube of X-Y-Voltage. I illustrated the axes here: https://i.sstatic.net/iS7tW.jpg.
I currently use numpy arrays in python dictionaries for the different kind of transistors I am sweeping. I'm not sure if this is the best way to do this. I've looked at Pandas, but am also not sure if this is a good job for Pandas. Was hoping someone could help me out, so I could learn to be pythonic! The code to generate some test data and the end structure is below.
Thank you!
import numpy as np
#make test data
test__transistor_data0 = {"SNMOS":np.random.randn(3,256,256),"SPMOS":np.random.randn(4,256,256), "WPMOS":np.random.randn(6,256,256),"WNMOS":np.random.randn(6,256,256)}
test__transistor_data1 = {"SNMOS":np.random.randn(3,256,256), "SPMOS":np.random.randn(4,256,256), "WPMOS":np.random.randn(6,256,256), "WNMOS":np.random.randn(6,256,256)}
test__transistor_data2 = {"SNMOS":np.random.randn(3,256,256), "SPMOS":np.random.randn(4,256,256), "WPMOS":np.random.randn(6,256,256), "WNMOS":np.random.randn(6,256,256)}
test__transistor_data3 = {"SNMOS":np.random.randn(3,256,256), "SPMOS":np.random.randn(4,256,256), "WPMOS":np.random.randn(6,256,256), "WNMOS":np.random.randn(6,256,256)}
quadrant_data = {"ne":test__transistor_data0,"nw":test__transistor_data1,"sw":test__transistor_data2,"se":test__transistor_data3}
It may be worth checking out xarray, which is like (and partially based on) pandas
, but designed for N-dimensional data.
Its two fundamental containers are a DataArray
which is a labeled ND array, and a a Dataset
, which is a container of DataArray
s.
In [29]: s1 = xray.DataArray(np.random.randn(3,256,256), dims=['voltage', 'x', 'y'])
In [30]: s2 = xray.DataArray(np.random.randn(3,256,256), dims=['voltage', 'x', 'y'])
In [32]: ds = xray.Dataset({'SNMOS': s1, 'SPMOS': s2})
In [33]: ds
Out[33]:
<xray.Dataset>
Dimensions: (voltage: 3, x: 256, y: 256)
Coordinates:
* voltage (voltage) int64 0 1 2
* x (x) int64 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ...
* y (y) int64 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ...
Data variables:
SPMOS (voltage, x, y) float64 -1.363 2.446 0.3585 -0.8243 -0.814 ...
SNMOS (voltage, x, y) float64 1.07 2.327 -1.435 0.4011 0.2379 2.07 ...
Both containers have a lot of nice functionality (see the docs), for example, if you wanted to know max value of x
for each transitor, at the first voltage level, it'd be something like this:
In [39]: ds.sel(voltage=0).max(dim='x').max()
Out[39]:
<xray.Dataset>
Dimensions: ()
Coordinates:
*empty*
Data variables:
SPMOS float64 4.175
SNMOS float64 4.302