Problem with beam polarisation in equivalent photon approximation (EPA)

Asked by Aleksander Filip Zarnecki on 2019-12-21

Dear authors,

I tried to generate a process:

alias quark = u:d:s:c:b
alias QUARK = U:D:S:C:B
process aeqqn = A, E1 => ( quark, QUARK, N1 )

using EPA in e+e- collisions at 1.5 TeV CLIC:

beams = e1, E1 => circe2 => epa, isr

epa_x_min = 0.013333333 # 10 GeV from 750 GeV beam
epa_mass = me
epa_q_max = 4 GeV

Without beam polarisation the calculated cross section is about 4790 fb.
When I set electron beam polarisation the cross section is exactly twice as large (9580 fb) for both -80% and +80% polarisation. Something has to be wrong in my opinion, as the unpolarised cross section should be the average of -80% and +80% cross sections (the positron beam is assumed not to be polarised). The cross section is the same for any polarisation setting. Even if I set it to 0%:

beams_pol_density = @(-1), @(+1)
beams_pol_fraction = 0%, 0%

it is 9580 fb. But when I remove these lines (comment them out) it is back to 4790 fb...

 I use same circe2 file in all cases (standard CLIC file). I tested it with version 2.7.0 and 2.8.2 (same results).

I will be grateful for your reply. Which cross section number is the correct one?


Question information

English Edit question
WHIZARD Edit question
Wolfgang Kilian Edit question
Last query:
Last reply:
Launchpad Janitor (janitor) said : #1

This question was expired because it remained in the 'Open' state without activity for the last 15 days.

Juergen Reuter (j.r.reuter) said : #2

Was opened on Christmas. Strange. Nobody reacted^^

Additional hint:
It looks like this behaviour is only observed when circe2 is switched on. If there is no circe2 beam
spectrum, the cross section between the case of beams_pol_fraction = 0%, 0% and the one without
any polarisation setting are identical.

Launchpad Janitor (janitor) said : #4

This question was expired because it remained in the 'Open' state without activity for the last 15 days.

Juergen Reuter (j.r.reuter) said : #5

We still need some more time.

Juergen Reuter (j.r.reuter) said : #6

Filip, I can confirm your findings, using the SINDARIN below for the simple process
gamma e+ -> u dbar nuebar, the unpolarized cross section using 1.4 TeV is 2.5 pb, the
cross section for the polarization combinations are
e-, e+ -+/ ++ 10 pb, while -- and +- are 10^-12 pb. So the average would be the sum of the
four divided by 4, which 20/4 = 5 pb. This is what I get when I do
beams_pol_density = +1,+1
beams_pol_fraction = 0,0
then this cross section corresponds to 5 pb, so the one expected from the average of the
polarized components. The same is true for starting with a forbidden combination of
beams_pol_density = -1,-1
beams_pol_fraction = 0,0

Dear Juergen,

Thank you for checking. All the numbers you give agree with what I get from my simulation.
BUT! The point is that, if you remove the polarisation definition from your script, i.e. remove the lines:
  beams_pol_density = -1,-1
  beams_pol_fraction = 0,0
you will get only 2.5 pb (and not 5pb, as one would expect). Please check!

I would assume that without any polarisation definition the unpolarised cross section should be calculated. But then it should be equal to the cross section for 0%x0% polarisation...

When not using CIRCE2 spectra (monochromatic beam) the cross section for "unpolarised" beams (no polarisation settings) and 0%polarisation are both around 3pb. So it looks like CIRCE2 is doubling the polarised EPA cross section...


Juergen Reuter (j.r.reuter) said : #8

Yes, I also see that not using a beam polariztaion setting, the cross section gets diminuished by a factor of 2. I haven't switched of CIRCE2 yet, but I checked that switching off ISR on the positron hard process leg does not change the picture.

Wolfgang Kilian (whkilian) said : #9

After a more careful inspection (US & WK):

Without CIRCE, one should set
   beams_pol_fraction = 0.0001, 0.0001
to inspect the zero-polarization-fraction case, because otherwise the polarization setup will be recognized as trivial and ignored.

With the "tiny polarization" setting, the problem occurs both with and without CIRCE.

There is no problem for processes that do not involve EPA.

My conclusion: The chaining of the polarization density matrix from the beam with the polarization transfer matrix for EPA is wrong. Most likely this is an actual bug. I'll create a gitlab issue for that and will try to find a solution.

Juergen Reuter (j.r.reuter) said : #10

This is fixed in svn, public git, and nightly builds, and will be released in v2.8.3 and v3.0.0beta, cf. the corresponding linked bug.

Juergen Reuter (j.r.reuter) said : #11

Fix committed.