Add Differential Cross Section interface, Compton and PerturbativeQED Model #18

AntonReinhard · 2023-10-05T16:53:20Z

CI_UNIT_PKG_URL_QEDbase: https://github.com/AntonReinhard/QEDbase.jl#1735e72

AntonReinhard · 2023-10-08T10:50:12Z

Currently fails because it depends on QEDjl-project/QEDbase.jl#25

szabo137

First comments on the draft.

src/compton/differential_cross_section.jl

szabo137 · 2023-10-09T06:54:41Z

src/compton/differential_cross_section.jl

+    in_phase_space::AbstractVector{NumericType},
+    out_phase_space::AbstractVector{NumericType},
+)::Float64 where {NumericType<:QEDbase.AbstractFourMomentum}
+    if (!isapprox(sum(in_phase_space), sum(out_phase_space); rtol = sqrt(eps())))


Additionally, one must check if the passed momenta are on-shell, i.e. p*p==mass(<respective particle with momentum p>)^2 for all incoming and outgoing particles.

However, we should have two versions of this function: i) an unsafe version, which does not check energy-momentum conservation and on-shell momenta, and ii) a safe version, which checks both.

I thought the idea was that _differential_cross_section is inherently unsafe and any assertions happen in _assert_valid_input, which is where the on-shell-ness is tested currently.

This is correct, but the check for on-shell momenta is not input validation but a physics, i.e. the differential cross section is zero, if any of the external momenta does not obey p*p=mass(...)^2, similar to the energy-momentum conservation. The word unsafe means just an implementation of this function, which assumes, that the input momenta are all on-shell and conserve momentum and energy. The reason for this is, that it should be possible, to check this somewhere else, and therefore not again in the differential cross-section. For example, RAMBO produces momenta, which are automatically on-shell and energy-momentum conserving, which means these properties must not be checked here again.

src/differential_cross_section.jl

test/compton.jl

SimeonEhrig · 2023-10-09T08:36:11Z

Currently fails because it depends on QEDjl-project/QEDbase.jl#25

You can run the unit tests with a development branch as dependency: https://github.com/QEDjl-project/QED.jl/blob/dev/docs/src/ci.md#unit-tests-for-ci-users

This PR adds a simple implementation of fermion and boson propagators. This should be merged *before* #18 because the latter will use the functionality. - replace dev-version of `QEDbase` with release `v0.1.5` (cf. QEDjl-project/QEDbase.jl#39) --------- Co-authored-by: Uwe Hernandez Acosta <[email protected]> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: CompatHelper Julia <[email protected]> Co-authored-by: Anton Reinhard <[email protected]>

AntonReinhard · 2023-11-13T14:15:06Z

The docs building step currently fails because referenced types/functions have moved to QEDbase now, I'm not sure how to proceed with that. I don't want to delete the references or put the links to the QEDbase docs in manually.

Maybe MultiDocumenter would handle this?

SimeonEhrig · 2023-11-13T14:38:48Z

I don't think it solves the problem. MultiDocumenter compiles only the rendered HTML documentation of different packages.

I found this issue: JuliaDocs/Documenter.jl#688
So it is not official documented.

Maybe we can use as workaround markdown links [foo()](doc.com/api#foo), if we have a API reference page.

AntonReinhard · 2023-11-13T14:53:44Z

src/compton/differential_cross_section.jl

+    # TODO: replace this with some less computationally intensive way to figure out how many states there are
+    photon_in_bstate = Vector{SLorentzVector{ComplexF64}}(
+        base_state(
+            Photon(), Incoming(), photon_in, _spin_or_pol(process, Photon(), Incoming())
+        ),
+    )
+    electron_in_bstate = Vector{BiSpinor}(
+        base_state(
+            Electron(),
+            Incoming(),
+            electron_in,
+            _spin_or_pol(process, Electron(), Incoming()),
+        ),
+    )
+
+    # average over incoming polarizations/spins, but sum over outgoing pols/spins
+    normalization = 1.0 / (length(photon_in_bstate) * length(electron_in_bstate))


@szabo137 How should we do this?
Some anonymous or _ function that just maps from AbstractPol/AbstractSpin to 1 or 2?

I think this should be another interface function for scattering processes, e.g. averaging_norm(proc::AbstractScatteringProcess). For the specific implementation, I think a map of the AbstractSpin/AbstractPol to 1 or 2 seems convenient.

Nevertheless, I am currently working on a refactoring of the differential_cross_section function, which includes all internal functions like incident_flux or matrix_element as interface functions on AbstractScatteirngProcess. Maybe we should discuss the internals of the current implementation of differential_cross_section in this context.

Ah that makes sense. Do you want to make a PR into this branch or how do you plan to do it?

I think differential_cross_section can be disentangled from this one here. Therefore I suggest making a separate PR from dev, and after merging, we can update this one here.

I started the refactoring of differential_cross_section here: #38

This changes the scattering process interface from differential_cross_section to more fine granular functions, i.e. incident_flux, averaging_norm, matrix_element, and phase_space_factor. The actual differential_cross_section is downgraded from an interface function to a default implementation using the new interface. Currently, it's a draft-PR because it is not finished yet and the tests might be broken.

AntonReinhard · 2023-11-13T15:00:35Z

I don't think it solves the problem. MultiDocumenter compiles only the rendered HTML documentation of different packages.

I found this issue: JuliaDocs/Documenter.jl#688 So it is not official documented.

Maybe we can use as workaround markdown links [foo()](doc.com/api#foo), if we have a API reference page.

This would work, but for example AbstractParticle alone is referenced 8 times already, so I'd have to put that link into lots of places... I guess there's not really a better way right now though. Other than not referencing at all.

szabo137 · 2023-11-13T22:44:37Z

I have another crossref for the docstring issue: JuliaDocs/Documenter.jl#1343

Another workaround would be, to include the docstring of the at_ref function/type to the documentation of QEDprocesses, e.g. in the api.md page. However, this is a bit hacky, because it means to include QEDbase in the makedocs of QEDprocesses (I can provide a showcase or PR if needed). In this case, at_ref works again without specifying the URL, but it seems to be a very ugly solution.

AntonReinhard · 2024-03-27T17:05:25Z

I'm sorry, I somehow managed to kill this PR by accident. I saved the commits but I might have to open a new PR.

AntonReinhard force-pushed the compton branch from 7bb89cc to db41346 Compare October 8, 2023 02:17

AntonReinhard requested a review from szabo137 October 8, 2023 10:56

AntonReinhard force-pushed the compton branch from ce58af2 to b9447ef Compare October 9, 2023 07:44

szabo137 reviewed Oct 9, 2023

View reviewed changes

AntonReinhard force-pushed the compton branch 3 times, most recently from 86b319a to cc48356 Compare October 13, 2023 13:46

AntonReinhard force-pushed the compton branch 3 times, most recently from 163cdb5 to deb0a04 Compare October 17, 2023 15:31

szabo137 mentioned this pull request Oct 18, 2023

Add Propagators #28

Merged

1 task

szabo137 added this to the Release-next milestone Oct 24, 2023

AntonReinhard force-pushed the compton branch from deb0a04 to 91f1864 Compare November 13, 2023 11:48

AntonReinhard commented Nov 13, 2023

View reviewed changes

AntonReinhard force-pushed the compton branch from 3671e4e to be7636c Compare November 23, 2023 14:36

AntonReinhard closed this Mar 27, 2024

AntonReinhard force-pushed the compton branch from be7636c to f703634 Compare March 27, 2024 16:14

AntonReinhard mentioned this pull request Mar 27, 2024

RE: #18 Compton #47

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add Differential Cross Section interface, Compton and PerturbativeQED Model #18

Add Differential Cross Section interface, Compton and PerturbativeQED Model #18

AntonReinhard commented Oct 5, 2023 •

edited

Loading

AntonReinhard commented Oct 8, 2023

szabo137 left a comment

szabo137 Oct 9, 2023

AntonReinhard Oct 9, 2023

szabo137 Oct 9, 2023

SimeonEhrig commented Oct 9, 2023

AntonReinhard commented Nov 13, 2023

SimeonEhrig commented Nov 13, 2023

AntonReinhard Nov 13, 2023

szabo137 Nov 13, 2023

AntonReinhard Nov 14, 2023

szabo137 Nov 14, 2023

szabo137 Nov 14, 2023

AntonReinhard commented Nov 13, 2023

szabo137 commented Nov 13, 2023

AntonReinhard commented Mar 27, 2024

Add Differential Cross Section interface, Compton and PerturbativeQED Model #18

Add Differential Cross Section interface, Compton and PerturbativeQED Model #18

Conversation

AntonReinhard commented Oct 5, 2023 • edited Loading

AntonReinhard commented Oct 8, 2023

szabo137 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

SimeonEhrig commented Oct 9, 2023

AntonReinhard commented Nov 13, 2023

SimeonEhrig commented Nov 13, 2023

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

AntonReinhard commented Nov 13, 2023

szabo137 commented Nov 13, 2023

AntonReinhard commented Mar 27, 2024

AntonReinhard commented Oct 5, 2023 •

edited

Loading