nagative points in vegas

Hello Durmus,

you are getting negative numbers because your set of diagrams is not gauge invariant.
For example, you are removing top-quark in the intermediate state, but at the same time
you have top quark in the final state.

You can cross check the consistency of your results after removing intermediate particles (which is dangerous in general),
with simple result
for
u, U->G,~G2 process
first.

Another advice is to Check the model first,
 there is an option for this in the graphical interface --> Edit Model -> Check Model

I have downloaded model from http://home.fnal.gov/~kckong/mued/
and the "Check Model" gave me an error -- some variables in the dependences were not defined.

Did you check the model first?

Regards,
Alexander

Hi Alexander,

I did the exercise you mentioned and verified the breaking of GI. I believe the problem is mainly from breaking of GI though there could be some other problems with the model file itself. I actually corrected the model file and then checked it in Calchep. It was saying it to be OK although there could be a problem that Calchep can not check.

 I will do some extensive checking over the weekend since it still gives some negative points about 10^-8% .

Negative points about 10^-8% more likely are caused by numerical instability. Just ignore it.

You can check the following way to confirm it:
make clean
open file CALCHEP/include/nType.h
Uncomment first line

#define _LONG
and complile CalcHEP once more.

Now CalCHEP will work with numerical type "long double" which has 4
digits more. Speed of calculation will be about the same.

Best
    Alexander Pukhov

On 08/18/2016 03:07 PM, Durmus Karabacak wrote:
> Question #349730 on CalcHEP changed:
> https://answers.launchpad.net/calchep/+question/349730
>
> Durmus Karabacak posted a new comment:
> Hi Alexander,
>
> I did the exercise you mentioned and verified the breaking of GI. I
> believe the problem is mainly from breaking of GI though there could be
> some other problems with the model file itself. I actually corrected the
> model file and then checked it in Calchep. It was saying it to be OK
> although there could be a problem that Calchep can not check.
>
> I will do some extensive checking over the weekend since it still gives
> some negative points about 10^-8% .
>

Hi Durmus,

the level of 10^-8 negative numbers, I believe,
is related to the fact that the widths of the KK particles
are manual inputs, rather than evaluated automatically by CalcHEP,
there could be slight inconsistency there, which negligibly breaks GI,
so at this point I would not put much attention to 10^-8% negative numbers.

If you have corrected the model,
you can upload it to HEPMDB -- High Energy Physics Model Database,
https://hepmdb.soton.ac.uk
where we store models for CalcHEP, Madgraph and other packages.

Regards,
Alexander

Exclusion of diagrams can lead to breaking of gauge invariance.
 From the other side CalcHEP makes summation other all physical and
unphysical ( temporal and longitudial polarizations as well as
Faddeev-Popov ghosts) gluon states. Some of unphysical states have
negative norm. Because of gauge invariance sum over all unphysical
states should be zero.
If gauge invariance is broken, then result can be negative. Don't
exclude diagrams with color charge if you are not sure that you can do
it gauge invariant manner.

Best
    Alexander Pukhov

On 08/17/2016 03:27 PM, Durmus Karabacak wrote:
> Question #349730 on CalcHEP changed:
> https://answers.launchpad.net/calchep/+question/349730
>
> Description changed to:
> Hi, I ve been trying to simulate pp -> G, t, T process in mUED (For the
> model files I have used the one implemented by Datta et.al. shared at
> http://home.fnal.gov/~kckong/mued/ (arXiv:1002.4624)). To see the
> mistake I have done I looked at the following simplest case:
>
> Enter process: u,U -> G, t, T
> Exclude diagrams: G, A, Z, Z2,B2, t
>
> where I am interested in the process mediated by only G2 (2. Level KK
> gluon)
>
> In the numerical session I have done the following changes:
>
> In model parameters: Rinv = 1200
> Cuts:
> M(t, T): 50
> N(t): -2.5 | 2.5
> N(T): -2.5 | 2.5
> J(t,T): 0.2
> T(G): 50
> N(G): -2.5| 2.5
>
> In phase space mapping:
> Kinematics: 12 -> 3, 45
> 45 -> 4, 5
> Regularization: 45 | MKG2 | wKG2 | 2
>
> For the MC simulation:
> NSess: 10
> NCalls: 1000k
> The crossection results in each nSess generates negative points at the order of 10^-2 % without PDF sets. When I do the same calculation with Cteq6l the negative points becomes around 60%. I am really not sure where my mistake is.
> Thank you very much in advance.
>