Sadler EM19373-2

Loading raw data
Removing spikes and aligning to ZLP
Principle Component Analysis

In [1]:
%matplotlib notebook 
import matplotlib.pyplot as plt
import hyperspy.api as hs
hs.set_log_level('ERROR')
import numpy as np
import os

from IPython.display import HTML, display

import warnings
warnings.simplefilter(action='ignore', category=FutureWarning)
plt.rcParams.update({'figure.max_open_warning': 0})
warnings.filterwarnings("ignore")
#def hide_code():
display(HTML('''<script>
code_show=true; 
function code_toggle() {
 if (code_show){
 $('div.input').hide();
 } else {
 $('div.input').show();
 }
 code_show = !code_show
} 
$( document ).ready(code_toggle);
</script>
<form action="javascript:code_toggle()"><input type="submit" value="Click here to toggle on/off the raw code."></form>'''))

#hide_code()

Loading raw data

In [2]:
os.chdir(r'/dls/science/groups/e02/Mohsen/Sadler_data/em19373-2/GATAN/A6/Y/SI data (18)/')
In [6]:
# #load from saved hspy versions
# eels_core = hs.load('eels_core.hspy')
# eels_low = hs.load('eels_low.hspy')
In [3]:
#######To load the raw data
eels_core = hs.load('EELS Spectrum Image (high-loss).dm4')
eels_low = hs.load('EELS Spectrum Image (low-loss).dm4')
In [4]:
eels_core.sum().plot()
eels_low.sum().plot()
In [5]:
eels_core.axes_manager
Out[5]:

< Axes manager, axes: (50, 150|2048) >

Navigation axis name size index offset scale units
x 50 0 -0.0 0.20383207499980927 nm
y 150 0 -0.0 0.20383207499980927 nm
Signal axis name size offset scale units
Energy loss 2048 130.00000193715096 0.10000000149011612 eV

Removing spikes and aligning to ZLP

In [9]:
#Spike removal 
eels_low.isig[2.7:].spikes_removal_tool()
In [11]:
eels_core.spikes_removal_tool()
In [12]:
#Aligning the data wrt the zero-loss peak
eels_low.align_zero_loss_peak(calibrate = True,print_stats = True, also_align=[eels_core])

Initial ZLP position statistics
-------------------------------
Summary statistics
------------------
mean:	-0.0149
std:	0.0381

min:	-0.1
Q1:	0
median:	0
Q3:	0
max:	0.1
In [13]:
#After aligning  / spike removal
eels_core.save('eels_core')
eels_low.save('eels_low')
In [ ]:
###### Skip to here if loading pre-processed data
In [14]:
eels_core.axes_manager
Out[14]:

< Axes manager, axes: (50, 150|2043) >

Navigation axis name size index offset scale units
x 50 0 -0.0 0.20383207499980927 nm
y 150 0 -0.0 0.20383207499980927 nm
Signal axis name size offset scale units
Energy loss 2043 130.2000019401312 0.10000000149011612 eV
In [17]:
#Interactive sum spectrum over image roi
eels_core.plot()
rec_roi = hs.roi.RectangularROI(left = 1., right = 1.5, top = 1.9, bottom= 2.0)
roi_reg = rec_roi.interactive(eels_core)
roi_reg_sum = hs.interactive(roi_reg.sum, event= roi_reg.axes_manager.events.any_axis_changed)
roi_reg_sum.plot()

Principal component analysis

In [23]:
eels_core.plot()
In [24]:
eels_core.decomposition()

Decomposition info:
  normalize_poissonian_noise=False
  algorithm=svd
  output_dimension=None
  centre=None
In [25]:
eels_core.learning_results.summary()
Out[25]:
'Decomposition parameters\n------------------------\nnormalize_poissonian_noise=False\nalgorithm=svd\noutput_dimension=None\ncentre=None'
In [26]:
eels_core.plot_explained_variance_ratio()
Out[26]:
<matplotlib.axes._subplots.AxesSubplot at 0x7f7f123f1f50>
In [28]:
eels_core.decomposition(algorithm= 'RPCA_GoDec', output_dimension= 4)

Decomposition info:
  normalize_poissonian_noise=False
  algorithm=rpca
  output_dimension=4
  centre=None
In [29]:
eels_core.plot_decomposition_results()
In [30]:
#To avoid nmf error due to negative values
eels_core_pos = hs.signals.EELSSpectrum(eels_core.data.clip(min=0))
In [39]:
eels_core_pos.decomposition(True, algorithm= 'nmf', output_dimension= 7)

Decomposition info:
  normalize_poissonian_noise=True
  algorithm=nmf
  output_dimension=7
  centre=None
scikit-learn estimator:
NMF(alpha=0.0, beta_loss='frobenius', init=None, l1_ratio=0.0, max_iter=200,
    n_components=7, random_state=None, shuffle=False, solver='cd', tol=0.0001,
    verbose=0)
In [43]:
eels_core_pos.axes_manager = eels_core.axes_manager
In [44]:
eels_core_pos.axes_manager
Out[44]:

< Axes manager, axes: (50, 150|2043) >

Navigation axis name size index offset scale units
x 50 0 -0.0 0.20383207499980927 nm
y 150 0 -0.0 0.20383207499980927 nm
Signal axis name size offset scale units
Energy loss 2043 130.2000019401312 0.10000000149011612 eV
In [42]:
eels_core.axes_manager
Out[42]:

< Axes manager, axes: (50, 150|2043) >

Navigation axis name size index offset scale units
x 50 0 -0.0 0.20383207499980927 nm
y 150 0 -0.0 0.20383207499980927 nm
Signal axis name size offset scale units
Energy loss 2043 130.2000019401312 0.10000000149011612 eV
In [45]:
eels_core_pos.plot_decomposition_results()
#Issue with the edge of a detector quadrant
In [46]:
results_l = eels_core_pos.get_decomposition_loadings()
results_f = eels_core_pos.get_decomposition_factors()
In [47]:
from matplotlib import pyplot as plt
In [84]:
hs.plot.plot_images(results_l, cmap = 'jet', per_row=4, suptitle='NMF Decompostion Loadings', axes_decor='off', 
                   label=['Component {}'.format(i) for i in range(results_l.data.shape[0])])
plt.savefig('/dls/science/groups/e02/Mohsen/Sadler_data/em19373-2/GATAN/A6/Y/SI data (18)/eels_nmf_loadings.png', dpi = 300)
In [87]:
hs.plot.plot_spectra(results_f, style='cascade')

plt.axvline(284, c='C1', ls=':', lw=0.5)
plt.text(x=284, y=0.50, s='C-K$_\\alpha$', color='C1')
plt.axvline(188, c='C3', ls=':', lw=0.5)
plt.text(x=180, y=3.37, s='B-K$_\\alpha$', color='C3')
plt.savefig('/dls/science/groups/e02/Mohsen/Sadler_data/em19373-2/GATAN/A6/Y/SI data (18)/eels_nmf_factors.png', dpi = 300)
In [38]:
eels_core_pos.save('eels_core_pos')
In [54]:
eels_core_pos.sum().plot()
In [ ]: