Merge lp:~maddevelopers/mg5amcnlo/NLO into lp:~maddevelopers/mg5amcnlo/FKS5

[Valentin Hirschi]

=============================
|| MadLoop 5 version -1.0 ||
=============================

Major upgrade and modifications with respect to the existing trunk.
This branch basically represents MadLoop5 in its current stage of development and it will take part of the 2.0 release of MadGraph 5 along with MadFKS5 which will ultimately allow to have aMC@NLO incorporated in MadGraph 5 2.0.

This version of MadLoop 5 is ***stable*** and implements:

I) Generation and computation of virtual amplitudes for any process in the full QCD sector of the SM.
  > The model file to be imported for this is 'loop_SM_QCD-full'. (default restriction is not working for loop models yet.)
  > The syntax for process generation is now initial_state > final_state 'born_orders' [pert_orders] 'squared_orders' 'process_options' (Notice that the 'add' functionality still work at NLO)
  > There is only one output mode which is fortran standalone, that you can access as usual with the command 'output standalone'
  > Here is an example of test you could perform with this version:
      a) Compile CutTools once and for all by doing:
          'cd loop_material/CutTools/ && make && cd ../..'
      b) Then run the MG5 shell:
          './bin/mg5'
      c) Run some process, like:
          'import model loop_SM_QCD-full'
          'generate g g > t t~ [QCD]'
          'output standalone Trial'
      d) Then compile the test program 'check_sa.f' in your generated process:
          'cd Trial/SubProcesses/P0_gg_ttx/'
          'make'
          './check'
         You should see some trial PS points being evaluated.
  > Some comments are in order:
      a) The interface needs a lot of polishing and acceptance tests must be written.
      b) Decay chains are supported at NLO in principle (so the structure is thought to be compliant with it), but never tested yet.
      c) Multiple particle labels cannot be used for NLO. (MadFKS will take care of that)
      d) Most massive QCD processes have been crosschecked against MadLoop4, NGLuon and soon GoSam. But the proper systematic parallel tests are still missing.
      e) Mixed order perturbation and loop-induced processes (i.e. no born) are supported for generation only right now, and not for output yet.
      f) Fermion-flows are in principle dealt with, but never tested in practice (never tried majorana yet)
      g) Once CutTools is compiled, all tests should be passing except those related to the loop drawing.

II) What is missing?
  > Loop drawings are there in principle but bugged. Olivier will take care of that as soon as he has some time.
  > Interface upgrade with all NLO process features and the corresponding acceptance tests.
  > The parallel tests for the process I have crosschecked already, like gg>ng, n=2,3,4 and q q~ > q q~, g g > q q~, g g > t t~, etc...
  > The UFO model for the full SM (but notice that it could in principle already be handled by this version, if existing)
  > [...]

===================
|| FINAL COMMENT ||
===================

This is not at all a good-looking finished and polished product but rather a raw sketch of a first version of MadLoop 5 which is stable and complete enough to be brought to light.
I could have waited more for the above mentioned missing items to be implemented but I thought that since the portion of new code in MadLoop5 with respect to the MG5 trunk is already quite big, it is crucial that you guys give me some feedback and remarks on the actual implementation.
This will ease the merging of MadFKS5 and the future upgrade of the existing MG5 trunk.

[Valentin Hirschi (valentin-hirschi)]

This NLO branch was merged to the FKS5 one with Marco's approval.
Once FKS5 becomes mature enough, we will propose a new merge of FKS5 into madgraph5 2.0 with the same reviewers.

[marco zaro (marco-zaro)]

review: Approve