Debian Codegraph @

hugo/content/ref/fit/ex/galaxi.md

raw
1 +++
2 title = "Experiment at GALAXI"
3 weight = 30
4 +++
5
6 ## Experiment at GALAXI
7
8 This is an example of a real data fit. We use our own measurements performed at the laboratory diffractometer [GALAXI](https://www.fz-juelich.de/en/jcns/jcns-2/expertise/in-house-x-ray/galaxi) in Forschungszentrum Jülich.
9
10 {{< galleryscg >}}
11 {{< figscg src="/img/draw/FitGALAXIData_setup.jpg" width="350px" caption="Real-space model">}}
12 {{< figscg src="/img/draw/FitGALAXIData.png" width="350px" caption="Fit window">}}
13 {{< /galleryscg >}}
14
15 * The sample represents a 4 layer system (substrate, teflon, hmdso and air) with Ag nanoparticles placed inside the hmdso layer on top of the teflon layer.
16 * The sample is generated with the help of a `SampleBuilder`, which is able to create samples depending on parameters defined in the constructor and passed through to the `create_sample` method.
17 * The nanoparticles have a broad log-normal size distribution.
18 * The detector is created to represent the PILATUS detector from the experiment (line 19).
19 * In the simulation settings the beam is initialized and the detector is assigned to the simulation. A region of interest is assigned at line 39 to simulate only a small rectangular window. Additionally, a rectangular mask is added to exclude the reflected beam from the analysis (line 40).
20 * The real data is loaded from a tiff file into a histogram representing the detector's channels.
21 * The `run_fitting()` function contains the initialization of the fitting kernel: loading experimental data, assignment of fit pair, fit parameters selection (line 62).
22
23 ## Experiment description
24
25 To successfully simulate and fit results of some real experiment it is important to have
26
27 + A good guess about the sample structure and the initial values of the sample parameters.
28 + Full information about the instrument geometry: size and angular range of the detector.
29 + A 2D numpy array containing the intensities measured in the detector channels.
30
31 ### Experiment
32
33 As an example we will use our own measurements performed at the laboratory diffractometer [GALAXI](https://www.fz-juelich.de/en/jcns/jcns-2/expertise/in-house-x-ray/galaxi) in Forschungszentrum Jülich.
34
35 Our sample represents a 3-layer system (substrate, teflon and air)
36 with Ag nanoparticles sitting on top of the teflon layer.
37 The PILATUS 1M detector was placed at a distance of 1730 mm from the sample.
38
39 {{< figscg src="/img/draw/setup_galaxi_experiment.png" class="center" >}}
40
41 The results of the measurement are represented by the intensity image taken in certain conditions
42 (beam wavelength, grazing angle, detector angular range) and stored in a 32-bit tiff file. To be able to fit these data we have to
43
44 * prepare a description of the simulation
45 * load the experimental data in BornAgain's fitting engine
46
47 ### Preparing the simulation description
48
49 From the experimental setup we know the number pixels and detector angular range.
50 In BornAgain, we will represent this setup using the `SphericalDetector` object.
51 A detector corresponding to a PILATUS detector:
52
53 ```python
54 npx, npy = 981, 1043
55 phi_min, phi_max = (-1.7*deg, 1.7*deg)
56 alpha_min, alpha_max = (-0.6*deg, 1.24*deg)
57
58 detector = ba.SphericalDetector(npx, phi_min, phi_max, npy, alpha_min, alpha_max)
59 ```
60
61 ### Setting the region of interest
62
63 To speed-up the simulation and to avoid an influence from uninteresting areas on the fit flow it is often convenient to define a certain region of interest `roi`. In our example we set the `roi` to the rectangle with lower left corner coordinates and upper right corner coordinates, where coordinates are expressed in detector units ('deg')
64
65 <!--
66 ```python
67 simulation.setRegionOfInterest(85.0, 70.0, 120.0, 92.)
68 ```
69 -->
70
71 {{< galleryscg >}}
72 {{< figscg src="/img/draw/galaxi_imported_data.png" width="350px" class="center">}}
73 {{< figscg src="/img/draw/galaxi_cropped_data.png" width="350px" class="center">}}
74 {{< /galleryscg >}}
75
76 The final simulation setup looks as follows:
77
78 ```python
79 simulation = ScatteringSimulation()
80 simulation.setDetector(detector) # this is our rectangular detector
81 simulation.setSample(sample) # sample creation is not covered by this tutorial
82 simulation.setBeamParameters(1.34*angstrom, 0.463*degree, 0.0)
83 simulation.setBeamIntensity(1.2e7)
84 ```
85 <!--
86 simulation.setRegionOfInterest(85.0, 70.0, 120.0, 92.)
87 -->
88
89 During the fit, only the part of the detector corresponding to the `roi` will be simulated and used for $\chi^2$ calculations.
90
91 ### Importing the real data using Fabio library
92
93 The [Fabio library](https://github.com/silx-kit/fabio) provides a convenient way to import experimental data in the form of a `numpy` array.
94
95 ```python
96 import fabio
97
98 img = fabio.open("galaxi_data.tif.gz")
99 data = img.data.astype("float64")
100 ```
101
102 The main requirement is that the shape of the numpy array coincides with the number of detector channels (i.e. `npx, npy = 981, 1043` for given example).
103
104 {{< show-ex file="fit/scatter2d/expfit_galaxi.py" >}}
105
106 Data to be fitted: {{% ref-data-scat2d galaxi_data.tif.gz %}}