This notebook explores the possibility to use the Python package xray in order to ease the data processing from the TS/WEST database.
xray is an open source project and Python package that aims to bring the labeled data power of pandas to the physical sciences, by providing N-dimensional variants of the core pandas data structures.
In [1]:
%pylab inline
import xray
Below I'm using MDSplus to retrieve Tore Supra data, only because pywed.tsbase() didn't work from VPN. The idea is however the same with pywed.
In [2]:
from pywed import *
y,t = tsbase(34567, 'SIPMES')
In [5]:
import MDSplus as mds
conx = mds.Connection('mdsplus.partenaires.cea.fr:8000')
In [6]:
y = conx.get('_sig=gettsbase(34567,"SIPMES")')
t = conx.get('dim_of(_sig)')
In [7]:
plot(t,y)
Out[7]:
xray main data structure is DataArray. It is a labeled, N-dimensional array and a N-D generalization of a pandas.Series. Soem additional attributes, like the data units, can be set for information.
In [8]:
Ip=xray.DataArray(data=y, coords=[('time', t)])
Ip.attrs['units'] = 'MA'
print(Ip)
The values can be accessed with the values attribute:
In [9]:
Ip.values
Out[9]:
The name of the dimension:
In [10]:
Ip.dims
Out[10]:
The coordinates (here the time values):
In [11]:
Ip.coords
Out[11]:
Or its attributes (metadata)
In [12]:
Ip.attrs
Out[12]:
The coordinates values can be accessed through:
In [13]:
# Getting the coordinates values
Ip['time']
Out[13]:
In [14]:
# Or, equivalently:
Ip.coords['time']
Out[14]:
DataArray has a .plot() method similar to pandas's. (The plotting capabilities of xray are currently under active development...)
In [15]:
Ip.plot()
Out[15]:
In [16]:
Ip.plot(lw=2, linestyle='--', color='k')
Out[16]:
In [17]:
#the plot can be accessed to an existing axes
ax1 = subplot(311)
Ip.plot(ax=ax1)
Out[17]:
In [18]:
#Getting the 50th first element. Returns a DataArray
#Note that the attributes are persisted
Ip[:50]
Out[18]:
In [19]:
# Selecting data from their values and plot it
Ip.where(Ip > 0.3).plot()
Out[19]:
In [20]:
# Selecting data from its coordinates.
# Ex: positive-time values
Ip.where(Ip.time > 0).plot()
Out[20]:
In [21]:
# Selecting an 'interesting zone' with multiples conditions.
# Pay attention to the parenthesis around each conditions
interesting_zone = (Ip.time > -1) & (Ip.time < 20) & (Ip > 0.1)
Ip.where(interesting_zone).plot(lw=2, color='r')
Out[21]:
In [22]:
conx = mds.Connection('mdsplus.partenaires.cea.fr:8000')
gnl_values = conx.get('_sig=gettsbase(34567,"GNL")')
gnl_t = conx.get('dim_of(_sig,0)')
gnl_R = conx.get('dim_of(_sig,1)') # returns [0,1,2,3,4] ...
gnl_R = [1.97, 2.135, 2.3, 2.465, 2.63]
Creates an 2D DataArray object:
In [23]:
xray.DataArray(gnl_values)
Out[23]:
Same thing, but set the dimensions and coordinates. (time array shape is returned (1,N) from MDSplus, while xray requires (N,) --> squeeze)
In [32]:
gnl = xray.DataArray(gnl_values, coords=[('R', gnl_R), ('time', gnl_t.squeeze())])
In [33]:
print(gnl)
In [47]:
# Index by integer array indices
# Get Nl for R=2.135m (indice==1) and between the 10 000th and 20 000th time points
#
# gnl[dict(R=1, time=slice(10000,20000))].plot()
# or more simply:
gnl.isel(R=1, time=slice(10000,20000)).plot()
Out[47]:
In [ ]:
# Index by dimension coordinates labels
# getting the data for R=2.465 m and between t=[-0.2, 18] s
#
#gnl.loc[dict(R=2.465, time=slice(-0.2, 18))].plot()
# or more simply
In [45]:
gnl.sel(R=2.63, time=slice(-0.2, 18)).plot()
Out[45]:
The default method xray.DataArray.plot() sees that the data is 2 dimensional. If the coordinates are uniformly spaced then it calls xray.plot.imshow(). But It’s not necessary for the coordinates to be evenly spaced. If not, then xray.DataArray.plot() produces a filled contour plot by calling xray.plot.contourf().
In [41]:
gnl.where(gnl>1e10).plot()
Out[41]:
In [ ]: