Slicing tensors in pyobs¶

[1]:

import pyobs
import numpy

[2]:

# generate autocorrelated data
N=500 # number of configs

# central values
mu=[0.5, 1.2, -3.4, 60.2, -10.1, 5.01]

# diagonal cov matrix
cov=[(mu[i]*0.05)**2 for i in range(len(mu))]

# autocorrelation time
tau=4.0

rng = pyobs.random.generator('tutorial')
data = rng.markov_chain(mu,cov,tau,N)
data = numpy.reshape(data,numpy.size(data),)

yobs = pyobs.observable()
yobs.create('ensA',data,shape=(3,2))

Random generator initialized with seed = 649309182 [tutorial]

In jupyter notebooks we can also print an observable in a pretty format by typing it at the end of a cell.

[3]:

yobs

[3]:

0.484(23)       1.118(95)
-4.06(13)       61.3(2.8)
-10.40(30)      5.25(26)

We slice our test observable in various ways. We select rows, columns and specific sub-matrices

[4]:

print('first row: ', yobs[0,:])
print('second column: ', yobs[:,1])

first row:  0.484(23)   1.118(95)

second column:  1.118(95)
61.3(2.8)
5.25(26)

[5]:

idx = numpy.array([0,1])
print('upper 2x2 matrix \n',yobs[idx,:])

idx = numpy.array([1,2])
print('lower 2x2 matrix \n',yobs[idx,:])

idx = numpy.array([0,2])
print('third 2x2 matrix \n',yobs[idx,:])

upper 2x2 matrix
 0.484(23)      1.118(95)
-4.06(13)       61.3(2.8)

lower 2x2 matrix
 -4.06(13)      61.3(2.8)
-10.40(30)      5.25(26)

third 2x2 matrix
 0.484(23)      1.118(95)
-10.40(30)      5.25(26)

Observables can also be reshaped and transposed, like normal numpy arrays

[6]:

print('transpose \n',pyobs.transpose(yobs))
print('reshape \n',pyobs.reshape(yobs,(6,)))

transpose
 0.484(23)      -4.06(13)       -10.40(30)
1.118(95)       61.3(2.8)       5.25(26)

reshape
 0.484(23)      1.118(95)       -4.06(13)       61.3(2.8)       -10.40(30)      5.25(26)

[7]:

print('diag of upper 2x2 matrix \n',pyobs.diag(yobs[numpy.array([0,1]),:]))

diag of upper 2x2 matrix
 0.484(23)      61.3(2.8)