[Eg1_run] Models update

[Sebastian Kuhn]

Dear (all-)inclusive EG1ers,

Nevzat and I have arrived at what we believe is our "final" (for now ;-) version of all models to be used for EG1 inclusive analyses and publications. Since things have changed quite a bit over the intervening time, here is a short primer:

1) ALL code and ALL input files (a lot!) are in my CUE account ~kuhn/Models , as always
2) If you already HAVE a copy of models, here are the files you most likely will have to add/replace:
-    Yoni.f, newSFs.f, strucfunc.f and DSFs.f for FORTRAN code
-   yelas.dat, f##.ful (## = 00, 90, 99, 11, 12, 21 and 22), resxxx_fit_final.dat as input files
    (in particular xxx = a1n, but it can't hurt to copy the others just to be sure)
-  Makefile (at least as example to tell you what you need to compile)
3) If you DON'T have them already, there are a bunch of additional files you might need which are also in the same location. Your Fortran compile errors will probably tell you which ones.

If you want to get proton or neutron structure functions, you can just run "strucfunc.f" with smearing set to 0 and the appropriate target code. Note that any of your code that still refers to "models.f" and subroutines therein needs to make sure that all structure functions and form factors are calculated with the new replacement code in "newSFs.f" instead - looking at strucfunc.f will tell you which ones to use and how. (You might still need models.f for other bits, e.g. radiative correction code).
If you want to get the DEUTERON structure functions, you should use "strucfunc", select "D" as target and then set "smearing preference" to 2. This will use the new method developed by Wally Melnitchouk and Yoni Kahn to implement smearing at the level of structure functions, which works rather nicely. Using "0" instead of "2" will get you Peter's implementation instead; however, asymmetries aren't smeared at all in this case and you get inconsistencies with the model for A1n and A2n. It might be a good alternative for systematic error checks!
Note that "smearing preference = 2" automatically includes the QUASIELASTIC peak in the structure function, so you don't need to handle this separately (only the ELASTIC peak).
One additional thing that has changed with this new implementation is that now you can choose the variation of models separately for proton and neutron for the deuteron. This gives you double as many variations (you should only vary one at a time), but hopefully more realistic systematics. I reproduce below my advice as to WHICH variations do what and which ones to use for systematic errors.

As always, I'm eager to get feedback - have fun! - Sebastian

> here are the recommended parameters (to be set in your calling program):
>
> IPOL = 1
> IA1 = 4
> SFChoice = 11
> AsymChoice = 11
>
> Here are the variations that I would try:
>
> IA1 = 5: Vary A1 model in DIS by the size of its uncertainty (increase)
> IA1 = 6: Vary A1 model in DIS by the size of its uncertainty (decrease)
> (run each, but take the average of the two deviations from IA1=4 as systematic error)
>
> IPOL = 2: Add twist-3 term to g2 in DIS
> IPOL = 5: A2 = 0 in DIS
> (again, might want to run both and use the average deviation as systematic error)
>
> AsymChoice = 12: Small variation of A1 model in the Res. Reg.
> AsymChoice = 13: Old version of A1 model in Res. Reg.
> Run both and add the deviations in quadrature, meaning DON'T average
> AsymChoice = 14: Replace A2 in Res. Reg. with MAID
> AsymChoice = 15: Old version of A2 model in Res. Reg.
> Run either only 15 or average 14 and 15; 14 should probably be used only below W = 1.6
> SFChoice = 12: Increase F1 and F2 simultaneously by their error
> SFChoice = 13: Change R by its error and propagate to F1 (leave F2 unchanged).
> (do both of these and add the deviations - don't average)