iolite Python for LA-ICP-MS Course - Week 8 Discussion

Joe

Hi all,

Week 8 of iolite's Python for LA-ICP-MS Course - User Interface Plugins - is now available!

You can watch the lesson on YouTube and find some additional info and links in the docs.

Please let us know if you have any questions.

Nathan

Hi Joe,
Just a bit of background: My idea was to re create a graph from O'Sullivan et al. 2020 to determine the potential origin of apatites based on trace elements data. Note that I do not want to do SVM (not for the moment at least) but I do have the background graph as .png that I could import into Iolite. I already do this kind of plot outside Iolite using a python script. So the idea was to mix my script with the LiveCloud example and to be able to see live how a change on the selection impacts the position of a point on the graph.
But I am stuck on likely something silly that I cannot identify about plotting the data

Any idea what it is? Do you want to see the code?

Joe

Nathan

Without seeing the code, my guess is that you're sending a normal list to the setData function, but it is expecting a numpy array. You can check by wrapping it in a np.array(x) and/or np.array(y). If the list is composed of integers you may also need to further specify that the data type should be float, e.g. np.array(x, dtype=np.float)

If that isn't it, post some code and I'll have a look.

All the best,

Nathan

wrapping it in np.array makes Iolite crashes... :-) However I do not think it is the issue here anyway
Here is the code

osplot.py

4kB

Joe

Nathan

Can you also send your osPlot.ui file?

In the code, you do x = np.nanmean(X), so x is just a single float (and same for y). Is that what you intended? Then when you wrap a float in np.array() the array has no size. If you want it to be a single value, you could leave it as is then do np.array([x]) and np.array([y]), I think.

Nathan

uploading a .ui file is not allowed. Zip neither
Yes I want a single value, one per selection. Then I will see what it shows if I do what you did on the LiveCloud and getting the point cloud for a selection

Joe

Nathan

Oh yeah - I suppose you would need to send it to the support email.

Did you try doing g.setData(np.array([x]), np.array([y]))?

Nathan

Ok, it does work this way. But I don't get why given that the x and y were already numpy.float64. Is it because it expects an array and not a single value?

Joe

Nathan

Yes, it expects an array, even if the array only has 1 value in it. However, I should fix this so it at least doesn't crash and provides a more useful error message 🙂

Nathan

ok, Thanks!

Nathan

Hi Joe,
I totally changed my idea for this week assignment. However I still think my first idea could be useful for me so I have two questions to move on this project as well.
The first one is how to set a .png at the background of the graph? I tried plenty of different stuff without success (error message or nothing happened). The png doesn't need to adapt to the scale, actually the scale is fixed.
The second point is how to turn the scale to log without the provided tool (that only turn the y axis to log)? I tried self.plot.bottom().setScaleType('stLogarithmic')
and different variation around, no success.
All the best
Nathan

Joe

Nathan

I don't think this was possible, so I've just added it now. In the next release you could do:
```
from iolite.QtGui import QPixmap
from iolite.QtCore import Qt
pm = QPixmap('/Path/To/Pixmap.png')
plot.axisRect().setBackground(pm, True, Qt.IgnoreAspectRatio)
```
where the second parameter is whether to scale the pixmap to the axis rect and the third is the aspect ratio mode.
You can change an axis to logarithmic by doing:
```
plot.left().setLogarithmic(True)
```
replacing left() as needed (e.g. with bottom()).

All the best,

Nathan

Thanks Joe!
Now I understand why I could not do it 😀