Error in OneLOop and Cross section is 'nan'.

[Enrico Bothmann]

Hi all,

using the attached Runcard for Pure Jets / MEPS@NLO I get the following output during integration of the lowest multi jj->jj (BVI) (Amegic and OpenLoops):

  ERROR in OneLOop dilog_r: |x|,iph =  .1000000000000000000E+01,          -2 , returning 0
  ...
  Process_Group::Differential(): Cross section is 'nan'.
  Phase_Space_Integrator::AddPoint(): value = -nan. Skip.

Of course it never converges.

Steffen said he has seen a similar thing for one of his setups, yet it worked fine when he switched to BlackHat, so it might be an OpenLoops related issue.

Cheers Enrico

[Enrico Bothmann]

Runcard

[Enrico Bothmann]

I've just confirmed that it starts integrating with BlackHat and it seems to converge, though there are still some intermingled

  Process_Group::Differential(): Cross section is 'nan'.
  Phase_Space_Integrator::AddPoint(): value = nan. Skip.

outputs.

[anonymous]

Hi, yes, this OpenLoops related (calls OneLOop for scalar integrals). For pure jets BlackHat is the faster choice as they have implemented the analytics. The nan's should not occur, they occur in the 2->2?

Cheers, Marek

[Enrico Bothmann]

Yep, they occur during 'Process_Group::CalculateTotalXSec(): Calculate xs for '2_2jjjjQCD(BVI)' (Amegic)', both with OpenLoops and BlackHat. But the latter integrates and converges (so far).

Cheers, Enrico

[anonymous]

With recent versions (SHERPA-trunk r29229 and OpenLoops rel-1-3-1), the "ERROR in OneLOop dilog_r" output has gone. The integration seems to converge (although I did not wait for it to finish completely).