METS scale setter "No CSS history" on trunk_cpl

[Frank Siegert]

When using the attached run card on trunk_cpl r14935, the METS scale setter clustering has problems:

Process_Group::CalculateTotalXSec(): Calculate xs for '2_4__j__j__h0[tau-__tau+]__j__j' (Comix)
Starting the calculation. Lean back and enjoy ... .
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__d__cb__h0[tau-__tau+]__d__cb' in >32.89% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__u__db__h0[tau-__tau+]__c__sb' in >33.41% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__u__ub__h0[tau-__tau+]__s__sb' in >33.85% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__d__ub__h0[tau-__tau+]__s__cb' in >33.79% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__d__sb__h0[tau-__tau+]__d__sb' in >33.95% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__d__sb__h0[tau-__tau+]__u__cb' in >34.19% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__d__db__h0[tau-__tau+]__s__sb' in >32.34% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__d__db__h0[tau-__tau+]__c__cb' in >31.22% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__d__c__h0[tau-__tau+]__d__c' in >33.42% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__db__cb__h0[tau-__tau+]__sb__ub' in >33.98% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__u__cb__h0[tau-__tau+]__d__sb' in >34.18% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__u__cb__h0[tau-__tau+]__u__cb' in >31.78% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__d__c__h0[tau-__tau+]__s__u' in >34.53% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__c__db__h0[tau-__tau+]__c__db' in >31.93% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__db__cb__h0[tau-__tau+]__db__cb' in >33.21% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__u__ub__h0[tau-__tau+]__c__cb' in >32.77% of calls. Set \hat{s}.
0.284815 pb +- ( 0.0321343 pb = 11.2825 % ) 5000 ( 26938 -> 18.5 % )
full optimization:  ( 23s elapsed / 23m 48s left )
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__ub__cb__h0[tau-__tau+]__ub__cb' in >33.39% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__d__s__h0[tau-__tau+]__d__s' in >33.89% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__db__sb__h0[tau-__tau+]__db__sb' in >34.15% of calls. Set \hat{s}.
METS_Scale_Setter::CalculateScale(): No CSS history for '2_4__u__c__h0[tau-__tau+]__u__c' in >33.19% of calls. Set \hat{s}.
0.270043 pb +- ( 0.020303 pb = 7.51844 % ) 10000 ( 46753 -> 25.2 % )
full optimization:  ( 46s elapsed / 23m 17s left )
0.261761 pb +- ( 0.013231 pb = 5.05462 % ) 15000 ( 63071 -> 30.6 % )
full optimization:  ( 1m 9s elapsed / 22m 39s left )
...

On trunk this works without warnings.

[Stefan Hoeche]

The message is due to the flavour/colour structure of the process. In cases where the initial state partons have matching colour, but the t-channels don't, no clustering can be performed. However, the colour integrator is allowed to generate a corresponding point and this point is needed to compute the overall normalization. The reason why the problem does not show on trunk is that the corresponding ME is zero and scales are computed only after successful ME evaluation. This method cannot be applied on trunk_cpl, because scales are needed during the ME calculation already.

[Stefan Hoeche]

Milestone old deleted