O'Mega fails models with symmetries among fields at vertices (applies only to FeynRules models)

[Christian Speckner]

See summary; in such models, diagrams are dropped. As an example, I have attached a FeynRules?-generated model which contains an anomalous triple vector coupling between a new charged vector X1/2 and a gluon. In order to implement this coupling, it is necessary to introduce several copies of the same vertex with different couplings and orderings of the fields. For the process X1 X2 -> X1 X2, there is a s-channel and a t-channel diagram, and as the aTVC vertex is split into 4 pieces, O'Mega should generate 32 diagrams. While the 1.9x version of O'Mega correctly generates those 32 diagrams, the current version only generates 3 three of them.

Among other things, this breaks the SM + anomalous gauge boson couplings in WHIZARD.

[Christian Speckner]

W2 version of the feynrules model

[Christian Speckner]

1.9x version of the same model.

[ohl]

This was introduced and incompletely resolved while making the changes for #4. Will be fixed ASAP.

[ohl]

Replying to ohl:

This was introduced and incompletely resolved while making the changes for #4.

That turned out to be wrong. The bug was introduced with the new colorizer in r2219.

[ohl]

Fixed in r2403

Turned out to be a combination of stupid mistake/oversight when implementing the new colorizer and CSE across flavors:

• Fusion().colorize_braket1: use matching ALL fusions, not only the first one
• DAG/Tree2: include the coupling in the disambiguation

[Christian Speckner]

Sorry, this is not fixed. I've attached another FR-generated testcase: the Standard Model with the triple gauge boson couplings emulated by Dim4_Vector_Vector_Vector_T with three different permutations of the fields. For e+ e- -> W+ W-, we have 8 diagrams this way (A in s-channel: 3, Z in s-channel: 3, H in s-channel: 1, nu in t-channel:1), which is what 1.9x correctly generates, but with the current O'Mega, I only get 4 diagrams. I guess that you also must include the ordering of the fields in the vertex disambiguation.

[Christian Speckner]

SM with emulated TGCs, WO2

[Christian Speckner]

SM with emulated TGCs, WO1.9x

[Juergen Reuter]

TO will try to do it tomorrow. But safely we move it to 2.0.2.

[ohl]

Replying to cnspeckn:

I've attached another FR-generated testcase: [...] For e+ e- -> W+ W-, we have 8 diagrams this way (A in s-channel: 3, Z in s-channel: 3, H in s-channel: 1, nu in t-channel:1)

That's what I get with r2413 too

      owf_e_plus_1 = vbar (me, - p1, s(1))
      owf_e_minus_2 = u (me, - p2, s(2))
      owf_w_minus_3 = conjg (eps (mw, p3, s(3)))
      owf_w_plus_4 = conjg (eps (mw, p4, s(4)))
      owf_h_1_12 = pr_phi(p12,mh,wd_tl(p12,wh), &
         + s_ff(G_e_plus_h_1_e_minus,owf_e_plus_1,owf_e_minus_2))
      owf_a_1_12 = pr_feynman(p12, &
         + v_ff(G_e_plus_a_1_e_minus,owf_e_plus_1,owf_e_minus_2))
      owf_z_1_12 = pr_unitarity(p12,mz,wd_tl(p12,wz), &
         + vlr_ff(G_e_plus_z_1_e_minus(1),G_e_plus_z_1_e_minus(2),owf_e_plus_1,owf_e_minus_2))
      owf_ve_bar_13 = pr_psibar(p13,fr_zero,wd_tl(p13,fr_zero), &
         + f_fvl(G_ve_bar_w_plus_e_minus,owf_e_plus_1,owf_w_minus_3))
      oks_e_pluse_minusw_plusw_minus = 0
      oks_e_pluse_minusw_plusw_minus = oks_e_pluse_minusw_plusw_minus + owf_z_1_12*( &
         + tkv_vv(G_z_1_w_plus_w_minus,owf_w_plus_4,p4,owf_w_minus_3,p3) &
         + (-1)*tv_kvv(G_z_1_w_plus_w_minus,owf_w_minus_3,p3,owf_w_plus_4,p4) &
         + tv_kvv(G_z_1_w_plus_w_minus,owf_w_plus_4,p4,owf_w_minus_3,p3))
      oks_e_pluse_minusw_plusw_minus = oks_e_pluse_minusw_plusw_minus + owf_a_1_12*( &
         + tkv_vv(G_a_1_w_minus_w_plus,owf_w_minus_3,p3,owf_w_plus_4,p4) &
         + (-1)*tv_kvv(G_a_1_w_minus_w_plus,owf_w_plus_4,p4,owf_w_minus_3,p3) &
         + tv_kvv(G_a_1_w_minus_w_plus,owf_w_minus_3,p3,owf_w_plus_4,p4))
      oks_e_pluse_minusw_plusw_minus = oks_e_pluse_minusw_plusw_minus + owf_ve_bar_13*( &
         + f_vlf(G_ve_bar_w_plus_e_minus,owf_w_plus_4,owf_e_minus_2))
      oks_e_pluse_minusw_plusw_minus = oks_e_pluse_minusw_plusw_minus + owf_h_1_12*( &
         + G_h_1_w_plus_w_minus*(owf_w_minus_3*owf_w_plus_4))
      oks_e_pluse_minusw_plusw_minus = - oks_e_pluse_minusw_plusw_minus ! 2 vertices, 1 propagators
      ! unit symmetry factor
      amp(1,h,1) = oks_e_pluse_minusw_plusw_minus
...
!    Fusion.Make():
!      Fusions for arbitrary topologies
!      Source: /trunk/src/omega/src/fusion.ml 
!      revision: 2403  checked in by ohl  at 2010-04-23 22:28:27 +0200 (Fri, 23 Apr 2010) 

I'm not setting this to worksforme immediately, but please comment.

[Christian Speckner]

I don't know what was wrong on my side, but it is working now (some recompiles are in between though). I swear I got the wrong number of diagrams before, and I'm pretty sure I was using an up-to-date SVN tree, but you never know... sorry. I'll close this as worksforme.