Other background than polynomial

[ivo.a...@psi.ch]

Dear all

I seems that the only way I can define a background in the models for a fit is by

 model.add_polynomial_background(order=bg_order)

Is it possible to define a custom background or to add an additional component to the polynomial (I would need a narrow gaussian to take care of a large spike at E=0.0 in my EDX spectra)?

Thanks for any advice

[Thomas Aarholt]

Gruss gott Ivo,

It's absolutely possible to do :) it might be easiest if we try for your exact use case - what sort of background do you want to add?

For the gaussian you can use hs.components1D.Gaussian (or some variation of that, possibly with hs.model first. I'm on mobile). Press Tab in Jupyter Notebook to see the suggested Methods. 

You can feed the gaussian arguments for centre, A (height), and sigma. 

I suggest you join the gitter chat (bottom right on the hyperspy webpage) for easier discussion!

Vennlig hilsen Thomas 

--
You received this message because you are subscribed to the Google Groups "hyperspy-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hyperspy-user...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Thomas Aarholt]

I had the wrong syntax for the gaussian:

`gaussian = hs.model.components1D.Gaussian(centre=0.0)`

You then append it to the model by:

`m.append(gaussian)`

[ivo.a...@psi.ch]

Ok,

so does this mean it is up to me to keep track which components belong to the background and which describe my signal. In this case what does ``model.fit_background()`` do? I.e. how are background components distinguished from signal componens?

Thanks

[Tomas Ostasevicius]

Hi,

Each component has `comp.isbackground` boolean value that I hope is used to distinguish this.

Tomas

--

[Thomas Aarholt]

Tomas is correct. Reading the code (this should *really* be documented better) in edsmodel.py, it fits all components with isbackground == True.

On 4 May 2017 at 15:17, Tomas Ostasevicius <t.ostas...@gmail.com> wrote:

Hi,

Each component has `comp.isbackground` boolean value that I hope is used to distinguish this.

Tomas

On Thu, 4 May 2017, 14:12 , <ivo.a...@psi.ch> wrote:

Ok,

so does this mean it is up to me to keep track which components belong to the background and which describe my signal. In this case what does ``model.fit_background()`` do? I.e. how are background components distinguished from signal componens?

Thanks

--
You received this message because you are subscribed to the Google Groups "hyperspy-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to hyperspy-users+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

You received this message because you are subscribed to the Google Groups "hyperspy-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to hyperspy-users+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Thomas M. Aarholt

Reading for DPhil in Nuclear Materials

University of Oxford

 

Mobile: 07794820885

Email: thomas...@gmail.com