mapping for s-channel resonances

[kilian]

s-channel resonances (i.e., 2 -> 1 processes and decay) need a special mapping, which was available in W1. Reactivate this.

(See also #57.)

This might be postponed to 2.0.5.

[kilian]

Partly implemented in r2824.

The problem is that in a generic process, some phase-space channels may involve s-channel resonances, some don't. A mapping of the resonance is possible only be switching off all mappings of the (0,1) x range that are used by individual structure functions. The implementation therefore must (1) switch them off for the current channel (done), and (2) compute the (inverse) mappings for all channels without s-channel mapping with their Jacobians (not implemented yet). Essentially, I have to separate mapping from structure-function calculation in all cases.

Mapping the x range to adapt to a resonance also implied switching off the generator mode, as it is used for CIRCE.

Priority: P4 should stay since this may not be ready in time for 2.0.4.

Apparently, I did this somehow for W1, but completely forgot ...

[Juergen Reuter]

Wasn't that done by this "bypass" mechanism?

[Juergen Reuter]

Is there any update on this?

[kilian]

Temporarily postpone this after the user-interface issue and related (cut/analysis).

[Juergen Reuter]

Particularly striking example is the missing radiative return to the Z peak in CLIC beamstrahlung spectra.

[Juergen Reuter]

This feature has already been present. Deadline for the new release (which should contain this) is 19.3.12.

[Juergen Reuter]

Will presumably not done before 2.0.7... but gets highest priority for 2.0.8.

[Juergen Reuter]

Pressure from the CLIC people :)

[kilian]

Update: this is underway, but it essentially amounts to introducing a multichannel algorithm for the whole structure-function part of the program. Will take some more time.

It does work already for the LHC setup (PDFs). The switches for that are still off by default, however.

[Juergen Reuter]

Again the question: is there any relation to #55?

[kilian]

@JR: Yes, the new multichannel approach will revive #55. But first I'll have to make this working properly.

[Juergen Reuter]

Any updates on that!?

[Juergen Reuter]

Wow, this ticket is reaching an age of 2 years. Incredible.

[kilian]

As long as #468 is under construction, this has to be postponed.

[Juergen Reuter]

This is of the utmost importance as even our examples don't run well without this (cf. #595). Shall we maybe even move it to 2.2.0 ??)

[kilian]

Yes. I have to re-examine the issue, alongside with re-enabling the remaining structure functions #500. The workaround of previous versions doesn't go along with the current implementation, so no surprise that effective 2 -> 1 processes fail completely.

[kilian]

Let's see what can be done. In principle, the infrastructure for dealing with this problem is there, even some (unit) tests for multi-channel structure function parameterizations. I just have to learn how to make use of this.

[kilian]

r5279: smtest_8: It works. Hallelujah!