LHAPDF still not working!!!

[Juergen Reuter]

It is not possible to change PDFs :(((

[Juergen Reuter]

Local beam options are not working. Things like beams = p, p => lhapdf { $lhapdf_file = "foo" } are not working.

[kilian]

Yes and no. The question is rather whether we really want to keep this construct.

In the current implementation, the 'beams' command catches sqrts and possibly polarization etc. (not reactivated yet), but not the structure function set. The effect (of catching sqrts) has already confused users in the past.

The structure function set variables (and related) are used by the program when an 'integrate' or 'simulate'/'rescan' command is issued. So, the above syntax would be misleading, if reinstated.

A possible move would be to catch _no_ parameters with the 'beams' command, just the particle and structure function names. This makes it much more convenient to scan over beam parameters: no need to repeat 'beams' commands inside the scan. The opposite would be to catch all relevant parameters. Even then, I would only allow one pair of braces for the whole 'beams' command. Opinions?

[kilian]

Closed by mistake.

[Juergen Reuter]

Well, option 1 seems very clean, however users might be used to the fact that they can specify the names and sets of PDFs there. (though for scans over PDF error sets, that might be not a good idea). Hm, option 2 would mean to repeat the beams statement in all scan/rescan constructs, which is nasty, but at least consistent. But my gut feeling is for option 1. TO??

[ohl]

I prefer all-or-nothing. Either shallow binding for all parameters (including particles and PDFs) or deep binding for all. This slight convenience of treating some special is more than offset by the potential for confusion.

[Juergen Reuter]

So I guess then option 1. Or are there any objections?

[kilian]

Agreed. So I'll change the beam handling accordingly.

[kilian]

Done in r4750.

Note that this is an incompatible modification of the syntax. The new 'beams' command doesn't take any options. sqrts, PDF set, etc. may either be set globally (order is no longer relevant), or locally in the options of an 'integrate' command.

The new syntax and semantics should be more natural. The 'beams' command now only determines the structure of the beams, no parameters. We can easily scan (at least, once the 'scan' command has been resurrected) over beam parameters.

If the change is accepted, it is likely that examples have to be adapted. (And, of course, the Manual.)