~madteam/mg5amcnlo/series2.0

Update notes for MadGraph5_aMC@NLO (in reverse time order)

2.9.10 (06/05/2022)

     OM: allow model with GC in their name to use helicity recycling

     OM: Forbid madspin to run with crazy value of BWcut

     OM: Fixing some IO issue with Madspin that was leading to a lock of the code

     OM: set a user interface for the set ewscheme option 

     OM: forbid helicity recycling for model with spin2 and spin3/2 (optimization is not implemented for such case)

2.9.9 (25/02/2022)

     OM: Fix a bug introduced in 2.9.0 for MLM generation in presence of mix EW/QCD process.

         The bug typically leads to a crash within the systematics.py due to wrong power of alpha_s

	 The bug might sometimes not lead to a crash but the impact is "limited" to a change in the (various)

	 scale choice associated to the events. Therefore this is likely to be within scale uncertainty. It can

	 impact the matching with the parton-shower (but this should be visible within the DJR validation plot). 

     OM: Fix some wrong zero-result that occur in presence of conflincting breit-wigner.

     OM: Fix an issue when using "$ X" when X~ can be onshell, the phase-space symmetry factor was wrongly set.

2.9.8 (21/02/2022)

     OM: Fix in madspin where onlyhelicity mode was not working anymore

         also allows to not specify any decay in that mode.

     OM: Fix in multi_run mode where some meta-data where incorrectly set within the merged lhef file.

         This occurs ONLY in presence of non positive definite cross-section. 

         Only normalization of the cross-section/weight/statistical error could be wrong. Shape are not impacted.

2.9.7 (29/11/21)

     OM: Fix the behavior of python seed for madevent/madwidth/mcatnlo/madspin

         Now the first of those to setup a python seed will forbid any future reset of the seed

	 Frequent reset of the seed when moving to one package to the next was creating

	 a bias in the effective branching ratio out of madspin (observed only of NLO sample)

	 see: https://bugs.launchpad.net/mg5amcnlo/+bug/1951120

	 Thanks Hannes for the information.

     RF+PT+SF: Fixed a 10-year-old bug that was there in the importance sampling over the FKS configurations.

	 - Fix to pick one shower scale for the S event at random among the FKS configurations,

	   instead of taking the weighted average.

	 - removed the n-body contributions from the random picking of the showerscale among FKS configurations.

        Those bugs were leading at an incorect pick of the shower scale. The observed impact is relatively small

	and occur in the matching region.

2.9.6 (02/11/21)

     OM: Forbid the possibity to ask for massless boson to be longitudinally polarised.

         Asking for those with large multi-particle label could have hide the fact that the code

	 was returning a non zero cross-section for such request.

     OM: for process like p p > w+{X} w+{Y}, w+ > l+ vl

         i.e. process with polarization of  identical particle where the decay involves multi-particles and 

         where the lepton does not all have the same mass, some of the process were incorrectly discarded.

     OM: fix an issue for photon initial state with lpp=+-4 where the lhef output was not setting the muon

         as the correct colliding particles (thanks to Yuunjia Bao) 	  

     OM: Fixing an issue that madspin was not always using the python executable that was used by the

         mg5_aMC executable

     OM+PT: Change the shower script running pythia8 in order to be working with pythia8.3.

         Running with Pythia8.2 (or lower) is then not possible anymore.

     OM: Fix issue that some loop-induced processes were not working anymore since 2.9.0

     OM: Fix the default set of cut present in the default run_card if gluon were present in the final state but no

         light (or b) quark. The absence of those cuts, in that situation, were leading to hardcoded cut that was

	 not easy to overwrite by non expert user.

2.9.5 (22/08/21)

      OM+LM: [LO only] Fix the factorization scale dependence for lpp=2/3/4.

             This was claimed to be using fixed scale computation while in

	     some case the scale was dynamical

             To have full flexibility we introduced two additional (hidden)

	     parameter

             fixed_fac_scale1 and fixed_fac_scale2 that allow to choose

	     fixed scale for only one beam.

      OM: fix auto-width that was not following	run_mode specification

      OM: fixing missing rwgt_info for reweighting with gridpack mode

      OM: fixing 'check' command with the skip_event mode

      OM: fixing auto-width computation for 3 body decay and identical particle which was sometimes leading to crash

      OM: Fix some potential infinite loop when running with python3

2.9.4(30/05/21)

      OM: Fix a python3 issue for madSpin when using in a gridpack mode (set ms_dir)

      OM: Fix an issue for non positive definite matrix-element when using the

          "set group_subprocesses False"  mode of MG5aMc

      OM: Fix a speed issue for gridpack in readonly mode where the grid were recomputed 

2.9.3(25/03/21)

      OM: Fix an issue with t-channel particles for massive initial state where sqrt(S)

          was close to the mass of the initial mass. boundary condition on t-channnel were

	  incorrectly setup leading to strange spectrum in various distribution (very old bug)

      OM: Fix an issue with a crash for loop computation

      OM: more python3 fixing bug

      OM: Fix a bug in the re-writting of the run_card when seed was specify leading to a wrong templating

      OM: various small fix of crash/better logging/debug information

2.9.2(14/02/21)

      MZ+RF+OM: Fix relevant to the case when PDG specific cuts are set in the NLO run_card.

        The generation of tau was not taking into account the information.

      OM: fix an issue when running with python3 where the normalization of pythia8 file were wrong

          when pythia8 was run in parralel.

      OM: Fix some wrong scale variation at NLO+PS (introduced in 2.9.0)

      OM: Fix loop-induced which was not working anymore

      OM: Fix an issue with the installation of pythia-pgs

2.9.1.2 (02/02/21)

      Kiran+OM: Fixing issue for the MSSM model

      OM: Fixing issue with mac support

      OM: fix a bug introduced within 2.8.2 related to a wrong phase-space mapping of conflicting resonances.

          Leading to zero cross-section and/or potential bias cross-section for more complex cases.

	  The issue occurs only if you have a conflicting resonances followed by a massless propopagator.

	  one example is generate p p > z h , z > e+ e-, h > b b~ a

2.9.0 (30/01/21)    **** Major speed-up update for LO computation ***

      Kiran+OM: Optimization of the matrix-element at run-time using recycling helicity method

                - This fasten LO computation by a factor around 2

		- This can be turned off via the command "output PATH --hel_recycling=False"

      OM: Various optimization fot T-channel integration

          - use different ordering for T-channel. Four different ordering have been implemented

	    at generation time, the code decides which ordering to use channel-per-channel.

	    This can be turned off via the command "output PATH --t_strategy=2"

	  - increase number of maximum iteration for double or more T-channel

	  - implement an alternative to the standard multi-channel.

	     instead of using the amplitude square it use the product of the denominator

	     (the default strategy use depends of the process)

	  - speed-up for VBF type of process  are often around 100 times faster

	  - This fixes a lot of issue for processes failing to generate the targetted number of events.

      OM: Better version of the color-computation (but this leads to only modest gain but for very complex processes.)

          This optimizations leads to a longer generation time of the code (only for complex processes).

	  This can be turned off  via "output PATH --jamp_optim=False"

      OM: New parameter in the run_card:

          - sde_strategy: allows to change the default multi-channel strategy

	  - a new phase-space optimization parameter are now easily availabe via the command "update ps_optim" 

      OM: New global parameter "auto_convert_model"

          if set on True (set auto_convert_model T or via input/mg5_configuration.txt)

	  all model crashing due to a python3 compatibility issue of the UFO model will be automatically converted

	  to a python3 compatible model. 

2.8.3 (26/01/21):

      OM: Buch of bunch fixing related to python3 issue (mainly related to unicode encoding)

      OM: Various fix for reweighting with loop (mainly with python3 as well)

      OM: Fix a potential bug for polarized sample with at least three polarised particles with a least two

          identical particles and one different polarised particle.

      OM: Fix various bug for maddm interface (thanks Daniele)

      OM: Fix various issue with overall order usage

      OM: Fix compatibility with MacOS 11

      OM: fix additional GCC10 compatibility issue

      OM: fix issue for 1>n process where width were set to zero automatically (introduced in 2.8.0)

      OM: fix aloha output mode for python output

      OM: avoid a bug with helicity filtering that was kept for all benchmark when using

          multiple successive re-weighting

      OM: Edition of the reweighting card via "set" command is not starting from the original param_card

          used to generated the events file and not from the model default anymore.

      OM: Interference have now their default dynamical scale set to HT/2

2.8.2 (30/10/20):

      OM: Fix a bug when setting width to zero where they were actually set to 1e-6 times the width

          This can lead to bias in the cross-section if your process is very sensitive to the cross-section

	  due to gauge cancelation. Bug introduced in version 2.6.4.

      OM: Hide Block parameter loop from NLO model but for madloop run.

          The factorization scale is still determine by the setup of the run_card as before

      OM: Fix couple of python3 specific bug

2.8.1(24/09/20):

      OM: Change user interface related to FxFx mode

          - If you have multiple multiplicities at NLO,

	    - the default run_card is modified accordingly

	        - has icckw=3 and follow the official FxFx recomendation (i.e. for

		  scale and jet algo)

            - the shower card has two parameter that are dynamically set 

                - njmax (default: -1) is automatically set depending of the process definition

                - Qcut (default: -1) is now automatically set to twice ptj parameter

            - the default parton-shower is automatically set to Pythia8

          - The value of the cross-section after FxFx formalism (removing double counting) is

	    now printed on the log and available on the HTML page

      OM: Fix for the auto width for three body decay in presence of identical particles.

      OM: add support for __header__ in UFO model

      OM: allow restriction card to have auto-width

      OM: fixing some html link (removed ajax link forbidden by major web browser)

      OM: Various fix related to the python3 support

          - including more efficient model conversion method

2.8.0 (21/08/20):

      OM: pass to python3 by default

      OM: For LO process, you can now set lpp1 and lpp2 to "4" for process with initial photon in order to get the

          effective photon approximation. This mode behaves like the "3" one: The cut-off scale of the approximation

	  is taken from the fixed renormalization factorisation scale: "dsqrt_q2fact1/dsqrt_q2fact2"

      OM: The width ao T-channel propagator are now set to zero automatically.

          To return to the previous behaviour , you can use the options "set zerowidth_tchannel False"

	  (to set before the output command)

      OM: Change in madevent phase-space integrator for T-channel:

            - The integration of photon/Z/Higgs are now done together rather than separatly and follow importance

	       sampling of the photon channel

	    - A new option "set max_t_for_channel X" allows to veto some channel of integration with more than X

	       t-channel propagator. This options can speed-up significantly the computation in VBF process.

	       (We advise to set X to 2 in those cases)

	    - Fix a numerical issue occuring for low invariant mass in T-channel creating spurious configuration.   

      OM: In madspin_card you can now replace the line "launch"

          by "launch -n NAME", this will allow to specify the name of the

	  directory in EVENTS where that run is stored.

      OM: Change in the python interface of the standalone output. In top of the pdg of the particles,

          you can now use the process_id (the one specified with @X) to distinguish process with the same particle

	  content. This parameter can be set to -1 and the function then ignore that parameter.

      OM: Adding new option in reweighting to allow the user to use the process_id in presence of ambiguous

          initial/final state.

      OM: Update the makefile of standalone interface to python to be able to compile in multicore.

           (thanks Matthias Komm)

      OM: Update of the auto-width code to support UFO form-factors

      OM: Fixing numerical issue  with the boost in EPA mode.

                        ** PARRALEL VERSION FOR PYTHON 3 **

2.7.3.py3(28/06/20):

      ALL: Contains all feature of 2.7.3 (see below)

      OM: Fix a crash when running PY8 in matched/merged mode (bug not present in not .py3 version of the code)

      MZ: low_mem_multicore_nlo_generation is working again

          but not for LOonly mode and not in OLP mode

2.7.2.py3(25/03/20):

      ALL: Contains all feature of 2.7.2 (see below)

2.7.1.py3(09/03/20):

      ALL: Contains all feature of 2.7.1 (see below)

      OM: Fixed a lot of python3 compatibility issue raised by user

          Particular Thanks to Congqio Li (CMS), Richard Ruiz and Leif Gellersen

	  for their reports.

2.7.0.py3 (27/02/20):

      ALL: Contains all feature of 2.7.0			

      OM: Support for python3.7 in top of python 2.7

          - python2.6  is not supported anymore

	  - this requires the module "six" [pip install six --user]

      OM: dropping function set low_mem_multicore_nlo

      OM: dropping support for syscalc (c++ version)

      OM: introduction of new setup variable

          - f2py_compiler_py2 and f2py_compiler_py3

	    which will be used to overwrite f2py_compiler when using the associate python version

          - lhapdf_py2 and lhapdf_py3

	    same for lhapdf

      OM: introduction of a new command "convert model FULLPATH"

          - try to convert a UFO model compatible to python2 only to a new model compatible both with

	    Python2 and Python3 (no guarantee)

                             ** Main Branch  Update **

2.7.3(21/06/20)

      OM: Fixing some bug for read-only LO gridpacks (wrong cross-section and shape when generating events).

          Thanks to Congqiao Li for this

      OM: Allowing loop-induced process to run on LO gridpack with read-only mode.

          Thanks to Congqiao Li for this      

      OM: Fix a bug in the longitudinal polarization for off-shell effect, leading to deviation at large invariant mass.

      OM: Adding more option to the run_card for fine tuning phase-space integration steps

          All are hidden by default:

	    - hard_survey [default=1]: request for more points in survey (and subsequent refine)

	    - second_refine_treshold [default=1.5]: forbid second refine if cross section after first refine is

	       is smaller than cross-section of the survey times such treshold

      OM: new command for the editions of the cards:

           - set nodecay: remove all decay line from the madspin_card

	   - set BLOCKNAME all VALUE: set all entry of the param_card "BLOCKNAME" to VALUE

	   - edit CARDNAME --comment_line='<regular_expression>' : new syntax to comment all lines of a card

	       that are matching a given regular expression

      OM: For Mac only, when running in script mode, MG5aMC will now prevent the computer to go to idle sleep.

          You can prevent this by running with the '-s' option. like ./bin/mg5_aMC -s PATH_TO_CMD

2.7.2(17/03/20)

      OM: Fix a Bug in pythia8 running on Ubuntu 18.04.4 machine

      OM: Speed up standalone_cpp code by changing compilation flag

2.7.1(09/03/20)

      OM: Fixing issue (wrong cross-section and differential cross-section) for

          polarised sample when

	   1) you have identical polarised particles

	   2) those particles are decays

	  examples: p p > j j w+{0} w+{T}, w+ > e+ e-

      OM: In presence of identical particles, if you define the exact number of decays

          (either via the decay chain syntax or via MadSpin) then they are assigned in an ordered way:

          generate p p > z{0} z{T}, z > l+ l-, z > j j

	  means that the Longitudinal Z decays to lepton (transverse to jet)

	  Thanks to Jie Xiao for reporting such issues. 

      OM: Effective Photon approximation is now always done with a fix cutoff. This use the FIXED factorization scale

          of the associate beam as the cutoff of the Improved Weizsaecker-Williams.

      OM: Allow to install lhapdf6.2 by default

      OM: Fixing website used to download pdf since lhapdf removed their previous hepforge page for pdf set.

      OM: Fixed a bug (leading to a crash) introduced in 2.6.6 related to the ckkw/MLM check for BSM model

          with gluon with non QCD interaction

      OM: For fortran standalone with python binding, this is not necessary anymore to run the code from a specific directory.

          You need however need to use the standard path for the param_card or have the file ident_card within the same

	  directory as the param_card.dat.

2.7.0(20/01/20)

      OM: Allow for a new syntax in presence of multi-jet/lepton process:

         generate p p > 3j replaces p p > j j j 

      OM: Allow syntax for (fully) polarized particles at LO: [1912.01725]

           ex: p p > w+{0} j, e+{L} e-{R} > mu+ mu- Z{T}

               p p > w+{T} w-{0} j j, w+ > e+ ve, w- > j j

      OM: (Thanks to K. Mawatari, K. Hagiwara) implemention of the axial gauge for the photon/gluon propagator.

          via "set gauge axial"

      OM: Support for elastic photon from heavy ion implemented. For PA collision, you have to generate your diagram

          with "set group_subprocesses False"

      OM: The default run_card.dat is now by default even more specific to your process. 

          Nearly all the cuts are now hidden by default if they do not impact your current process.

          This allows to have less information by default in the run_card which should simplifies its 

          readibility. 

      OM: distinguish in the code if zero contribution are related to no point passing cuts or if

          they are related to vanishing matrix-element. In the later case, allow for a lower threshold.

          (This allow to fasten the computation of such zero contribution)

2.6.7(16/10/19)

      OM: Fix a bug introduced in 2.6.2, some processes with gluon like particles which can lead to the wrong sign for interference term. 

      OM: Fix a bug introduced in 2.6.6 related to the restriction of model which was leading to wrong result for re-weighitng with loop model (but impact can in principle be not limited to re-weighting).

      OM: systematics now supports the option --weight_format and --weight_info (see help command for details)

      OM: set the auto_ptj_mjj variable to True by default

      OM: the systematics_arguments default value is modified in presence of matching/merging.

      OM: reweight: add an option --rwgt_info to allow to customise the banner information associate to that weight

      RF: Fixed a bug in the warning when using FxFx in conjunction with Herwig++/Herwig7. Also, with latest version of

          Herwig7.1.x, the FxFx needed files are compiled by default with in the Herwig code. Thanks Andreas Papaefstathiou.

2.6.6(28/07/19)

      OM: Bug in the edition of the shower_card. The set command of logical parameter was never setting the 

          parameter to False. (Thanks to Richard Ruiz) 

      RF: Fixed a bug in the creation of the energy-stripped i_FKS momentum. Tested for several processes,

          and seems to have been completely harmless.

      OM: Forbidding the use of CKKW/default scale for some UFO model allowing gluon emission from quark with

          no dependence in aS

      OM: Add an option "keep_ordering" for reweighting feature to allow to sometimes use decay chain even if

          you have ambiguity between final state particles.

      OM: Fixed an issue with interference which sometimes happens when some polarization contribution were negative but not all of them.

      VH: Change in the pythia8 output mode (thanks to Peter Skands)

      OM: Any number in the cards (not only integer) can use multiplication, division and k/M suffix for times 1000 and 1 million respectively

      OM: Energy cut (at LO) are now hidden by default for LHC type of run but visible for lepton collider ones.

      OM: Same for beam polarization

2.6.5 (03/02/19)

      OM: Fix some speed issue with the generated gridpack --speed issue introduced in 2.6.1--

      OM: Fix a bug in the computation of systematics when running with Python 2.6.

      OM: import model PATH, where PATH does not exists yet, will now connect to the online db

          if the model_name is present in the online db, then the model will be installed in the specified path.

      MZ: Applgrid+aMCFast was broken for some processes (since 2.6.0), due to wrong 

          information written into initial_states_map.dat. This has been fixed now

      OM: change in the gridpack. It automatically runs the systematics.py (if configure in the run_card)

      OM: Fix  a MLM crash occuring for p p > go go (0,1,2 j)

      OM: Fix issue for BSM model with additional colored particle where the default dynamical scale choice 

          was crashing

2.6.4 (09/11/18)

      OM: add specific treatement for small width (at LO only and not for loop-induced)

          if the width is smaller than 1e-6 times the mass, a fake width (at that value) is used for the

          numerical evaluation of the matrix-element. S-channel resonances are re-scaled according to 

          narrow-width approximation to return the correct total cross-section (the distribution of events 

          will on the other hand follow the new width). 

          The parameter '1e-6' can be changed by adding to (LO) run_card the parameter: "small_width_treatment"

      OM: add a new command "install looptools" to trigger the question that is automatically trigger 

          the first time a loop computation is needed.

      RF: Fixed a bug when using TopDrawer plots for f(N)LO runs, where the combination of the plots could lead

          to completely wrong histograms/distributions in case of high-precision runs.

      OM: Fix some MLM crash for some processes (in particular BSM processes with W'). 

      OM: Fix a bug in the reweighting due to the new lhe format (the one avoiding some issue with py8)

      OM: Fix a behavior for negative mass, the width was set to negative in the param_card automatically

          making the Parton-shower (and other code) to crash since this does not follow the convention.

      OM: Change compiler flag to support Mojave.

2.6.3.2 (22/06/18)

      OM: Fix a bug in auto-width when mass are below QCD scale.

      OM: Fix a bug for g b initial state where the mass in the lhe file was not always correctly assigned

          Note that the momentum was fine (i.e. in the file P^2 was not equal to the mention M but to the correct one)

      OM: Improvment for madspin in the mode spinmode=none

      OM: Fix a bug in MadSpin which was making MadSpin to work only in debug mode

2.6.3 (15/06/18)

      OM: When importing model, we now run one additional layer of optimisation:

           - if a vertex as two identical coupling for the same color structure then the associated lorentz 

             structure are merged in a single one and the vertex is modified accordingly

      OM: When restricting a model, we also run one additional layer of optimisation

           - Opposite sign coupling are now identified and merged into a single one

           - if a vertex as two identical coupling (up to the sign) for the same color structure

             then the associated lorentz structure are merged in a single one and the 

             vertex is modified accordingly 

      VH+OM: changing the ALOHA naming scheme for combine routine when the function name starts to be too long. 

      OM: adding a hidden parameter to the run_card (python_seed) to allow to control the randon number 

          generated within python and be able to have full reproducibility of the events

      OM: Fixing some issue with the default dynamical scale choice for 

            - non minimal QED sample

            - heft model when multiple radiation coming from the higgs decay/scattering

          This can also impact MLM since it use the same definition for the dynamical scale

      OM: Fix some issue for DIS scattering where the shat was wrongly defined for low energy scattering.

          Low energy scattering are not adviced since they break the factorization theorem.

          In particular the z-boost of the events are quite ill defined in that scenario.

      OM: changing the format of the param_card for NLO model to match expectation from the latest PY8

      OM: Update of MadSpin to allow special input file for the case of spinmode=none. 

          With that very simple mode of decay, you can now decay hepmc file or wrongly formatted leshouches event

          (in that mode we do not have spin correlation and width effect)

      PT: in montecarlocounter.f: improved colour-flow treatment in the case gluons are twice colour-connected to each other

          new gfunction(w) to get smoothly to 0 as w -> 1. (for NLO+PS run)

      OM: Fix some issue for the new QED model (including one in the handling of complex mass scheme of such model)

      OM: Fixing an issue of the param_card out of sync when running compute-widths

      OM: Adding Qnumbers block for ghost (the latest version of py8 was crashing due to their absence)

2.6.2 (29/04/18)

      Heavy ion pdf / pdf in general:

      -------------------------------

      OM: Support for rescaling PDF to ion PDF (assuming independent hadron), this is well suited for Lead-Lead collision, p-Lead collision and fix-target

      OM: Support in systematics.py for ion pdf. Possiblity to rescale only one beam (usefull to change only on PDF for fix target experiment)

      OM: Removing internal support for old type of PDF (only supported internal pdf are now cteq6 and nnpdf23)

      User Interface

      --------------

      OM: introduce "update to_full" command to display all the hidden parameter.

      OM: introduce "update ion_pdf" and "update beam_pol" to add related section in the run_card.

          the polarization of the beam is set as hidden parameter instead as default parameter

      OM: improve handling of (some) run_card parameter:

            - add comment that can be displayed via "help NAME"

            - add autocompletion for some parameter

            - add direct rejection of parameter edition if not in some allowed list/range

      Bug fixes:

      ----------

      OM: Fix issue with madspin introduced in 2.6.1 (failing to generate correct diagram for NLO process)

      OM: fix crash in 1>N reweighting

      RF+MZ: Fixed a problem with (f)NLO(+PS) runs in case the Born has identical QCD-charged

      	     particles. Cross sections were typically correct, but some distributions might

	     have shown an asymmetry. 

      OM: Change in LO maching for HEFT (or any model with hgg vertex) in the way to flag jet that should

          not take part in the matching/merging procedure.

      OM: Fixed a bug for loop induced in gridpack mode

      RF: Fixed a bug for ApplGrid: in rare cases the ApplGrid tables were filled twice for the same event

      OM: Fixed a bug for fixed target experiment when the energy of the beam was set to 0. 

      RF: Fixed an issue where too many files were opened for fNLO runs in rare cases

      OM: Fix issue on Madevent html output where some link where broken

      OM: Fix issue for the display lorentz function (was also presenting security issue for online use)

      OM: Fix issue for spin 3/2 (one in presence of fermion flow violation and one for custom propagator)

      Enhancement:

      -----------

      OM: add the value of all the widths in Auto-width in the scan summary file

      OM: For 1>N, if the user set fixed_run_scale to True, then the scale is choose accordingly

          and not following the mass of the inital state anymore

      RF: For the HwU histograms, if no gnuplot installation is found, write the gnuplot scripts in v5

          format (instead of the very old v4 format).

      OM: Change the default LO output directory structure. Now by default the lepton and neutrino are split 

          in two different directory. This avoids to face problem with the assymetric cut on lepton/neutrino

      OM: loop-filter commands are now working for loop-induced processes

      OM: New  method avoiding that two process are running inside the same output directory.

          This is implemented only for Gridpack and LO run so far. 

          The new method should be more robust in case of crash (i.e. not wrongly trigger as before)

      OM: For LO scan, if a crash (or ctrl-c) occurs during the scan, the original param_card is now

          restored.

2.6.1 (12/12/17)

      RF+MZ: It is now possible to add LO matrix elements (with [LOonly]) 

             to Fx-Fx merged samples. Thanks to Josh Bendavid for testing.

      OM: Re-factoring the code asking which program to use (both at LO and NLO)

          - design modular, designed for PLUGIN interactions

          - the length of the question auto adapts to the size of the shell

      OM: Allowing to have the gridpack stored on a readonly filesystem

      OM: Fix a bug in matching/merging forbiding the pdf reweighting for some processes (since 2.4.0)

      OM: Creation of online database with the name of known UFO model. If a use try to import a model

          which does not exits locally, the code will automatically check that database and download the 

          associate model if it exists. You can contact us if you are the author of one model which is not

          on our database. 

	  The list of all available model is available by typing "display model_list"

      OM: Model with __arxiv__ attribute will display "please cite XXXX.XXXXX when using this model" when

          loaded for the first time.

      OM: A fail of importing a UFO model does not try anymore to import v4 model

      OM: Many model present in models directory have been removed, however they can still be imported

          since they are available via automatic-download

      OM: Refactoring of the gridpack functionality with an infinite loop to reach the requested number of events

      OM+RF: Adding new class of cut at LO/NLO defined via the pdg of the particle

      VH: Support for the latest version of MA5

      MZ: Adding support for lhapdf v6.2

      OM: Fixing various bug in the spinmode=onshell mode of MadSpin

      OM: Fix a bug for model with 4 fermion in presence of restrict_card

      OM: Fix aloha bug in presence of complex form-factor.

      OM: improve auto-detection and handling of slha1/slha2 input file when expecting slha2. 

2.6.0 (16/08/17)

      New Functionalities:

      --------------------

      RF+OM: Added the possibility to also have a bias-function for event generation at (f)NLO(+PS)

      OM: Improve Re-WEIGHTING module

          1) creation of a single library by hyppothesis. 

          2) library for new hyppothesis can be specify via the new

             options: change tree_path and change virt_path

          3) allows to re-weight with different mass in the final states (LO only)

             This forces to rewrite a new lhe file, not adding weight inside the file

             (via the command: change output 2.0)

          4) allows to run the systematics on the newly generated file (for new output file)

             (via the command change systematics True)

          5) Fix some Nan issue for NLO reweighting in presence of colinear emission

	  6) various bug fixing, speed improvement,...

      OM: Add new option to SYSTEMATICS program:

           --remove_weights, --keep_weights, --start_id

           See "help systematics for more details."

      OM: Allow to specify param_card, run_card,... directly via an html link.

      Bug Fixing:

      -----------

      OM: Update condor class to support CERN type of cluster (thanks Daria Satco)

      OM: Fixing a bug leading to a crash in pythia8 due to one event wrongly written when splitting the

          events for the parralelization of pythia8

      OM: Fixing an issue, leading to NAN for some of the channel of integration for complicated processes.

      OM: Fix a bug in gripack@NLO which forbids to run it when the gridpack was generated with 0 events.

      RF: Fixed bug #1694548 (problem with NLO for QCD-charged heavy vector bosons).

      RF: Another fix (adding on a fix in 2.5.5) related to FxFx merging in case there are

      RF: Fixed bug #1706072 related to wrong path with NLO gridpack mode

          diagrams with 1->3 decays.

      RF: Fixed a bug (found by SF) that gave a seriously biased resonance mass when using MadSpin for decaying a 2->1 process.

      PT: Fix in montecarlocounter.f. Previously, for NLO+PS it was reading some subleading-colour information, now all

          information passed to the MC counterterms is correctly leading colour. 

      OM: Fix systematics computation for lepton collider

      OM+VH: Remove the proposition to install pjfry by default, due to many installation problem. The user can still force to 

          install it, if he wants to.

      OM: Fix a problem of madspin when recomputing width for model loaded with --modelname option

      OM: Fix events writing for DIS (thanks to Sho Iwamoto)

      OM: Fix a problem of output files written .lhe.gz, even if not zipped (python2.6 only)

      OM: Fixing some issue related to the customised propagator options of UFO model

      Code Re-factorisation:

      ----------------------

      RF: Refactor of the HwU.f fortran code. Gives more flexibility and potentially lower

          memory requirements.

      RF: Refactor of the (NLO) code related to extra weights (for PDF/scale uncertainties).

      OM: Increase modularity of the code for the support of plugin (maddm and MPI@LO)     

          - Now we support the HPC plugin allowing to generate LO-gridpack on MPI machine

          - Plugin can now use the "launch" keyword

2.5.5(26/05/17)

      OM: Fixing bug in the creation of the LO gridpack introduced in 2.4.3. Since 2.4.3 the generated 

          gridpack were lacking to include the generated grid for each channel. This does not lead to 

          bias but to a significant slow down of the associated gridpack.

      OM: Supporting user function calling other non default function.

      OM: adding the command "update to_slha1" and "update to_slha2" (still beta)

      RF: some cleanup in the NLO/Template files. Many unused subroutines deleted.

      OM: fixing some bug related to complex_mass_scheme

      OM: fixing bug in ALOHA for C++ output (in presence of form-factor)

      OM: fixing lhe event for 1 to N process such that the <init> block is consistently set for the shower

      OM: ExRootAnalysis interface is modified (need to be requested as an analysis)

      RF: Fix for FxFx merging in case there are diagrams with 1->3 decays.

2.5.4(28/03/17)

      OM: Add a warning in presense of small width

      OM: Fix a bug related to a missing library (introduced in 2.5.3)

      OM: Improve stability of the onshell mode of MadSpin

      VH: Fix some problem related to LHAPDF

2.5.3(09/03/17)

      PT: Modified the default shower starting scale in montecarlocounter.f.

          The new reference scale from which the dampening profile is computed is sum_i mt_i/2, i

          being Born level final-state momenta.

      OM: New "special mode" for madspin accessible via "set spinmode onshell".

          This mode allow for full spin-correlation with 3 (or more) body decay but the decaying particle

          remains exactly onshell in this mode. Loop-induced production/decay are not allowed in this mode.

      OM+RF: Allowing for creation of LHE like output of the fixed order run at NLO.

          This LHEF file is unvalid for parton-shower (all PS should crash on such file). It will be 

          unphysical to shower such sample anyway.

          Two hidden parameters of the  FO_analyse_card.dat allow some control on the LHEF creation

       	  "fo_lhe_weight_ratio" allows to control the strength of a partial unweighting [default:1e-3]

             increasing this number reduce the LHEF size.

          "fo_lhe_postprocessing" can take value like nogrouping, norandom, noidentification.

             nogrouping forbids the appearance of the LHEF(version2) tag <eventgroup>

             norandom   does not apply the randomization of the events.

             noidentification does not merge born event with other born like counter-event

      RF: Better job handling for fNLO runs.

      VH: Fixing various problem with the pythia8 interface (especially for MLM merging)

      Team: Fixing a series of small crash

2.5.2(10/12/16)

      OM: improve systematics (thanks to Philipp Pigard)

      OM: new syntax to modify the run_card: set no_parton_cut

          This removes all the cut present in the card.

      OM: change the default configuration parameter cluster_local_path to None

      OM: change the syscalc syntax for the pdf to avoid using & since this is not xml compliant

      OM: avoid to bias module to include trivial weight in gridpack mode

      OM: Fix a bug making 2.5.1 not compatible with python2.6

      OM: Improve "add missing" command if a full block is missing

      OM: Fixing a bug reporting wrong cross-section in the lhef <init> flag (only in presence of 

          more than 80 channel of integration)

2.5.1 (04/11/16)

       PT+MZ: New interface for Herwig7.

              Fixed a bug in the Herwig7/Herwig++ counterterm (relevant to 2 -> n, n>2:

              in particular, the bug affected the dead zone for final-final colour connection

              in processes with more than two particles in the Born final state)

       VH: Parallelization of PY8 at LO

       OM: add the possibility to automatically missing parameter in a param_card

           with command "update missing" at the time of the card edition. Usefull for 

           some SUSY card where some block entry are sometimes missing.

       OM: Possibility to automatically run systematics program at NLO or Turn it off at LO

           (hidden entry of the run_card systematics_program = systematics|syscalc|none)

       RF: Some refactoring of the NLO phase-space generation,

           including some small improvements in efficiency.

       OM: Plugin can be include in a directory MG5aMC_PLUGIN the above directory need to be in

           the $PYTHONPATH

       OM: Fix systematics for e+ e- initial state.

       VH: Fix various bug in the HepMc handling related to PY8 (LO generation)

       OM: allow install maddm functionality (install ./bin/maddm executable)

2.5.0 (08/09/16)

     FUNCTIONALITY

       VH+OM: Adding an official interface to Pythia8 (parton shower) at Leading-Order

	        More information at https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/LOPY8Merging	   

       VH+OM+MA5: Adding an official interface to MadAnalysis5 for plotting/analysis/recasting

	        More information at https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/UseMA5withinMG5aMC	   

       OM: Introduces a new function for LO/NLO interface "systematics"

            This function allows to compute systematics uncertainty from the event sample

            It requires the event sample to have been generated with 

                - use_syst = T (for LO sample)

                - store_reweight_info = T (for NLO sample)

            At LO the code is run automatically if use_syst=T (but if SysCalc is installed)

       VH+OM: Adding the possibility to bias the event weight for LO generation  via plugin.

            More informtion: https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/LOEventGenerationBias

       VH+SP: extend support for CKKWL

     CODE IMPROVMENT / small feature

     -------------------------------

       OM: Modify the structure of the output format such that all the internal format have the same structure

       OM: Adding the Plugin directory. Three kind of plugin are currently supported

           - plugin defining a new type of output format

           - plugin defining a new type of cluster handling

           - plugin modifying the main interface of MG5aMCnlo 

           More informations/examples are available here:

           https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/Plugin

       OM: Adding the possiblity of having detailled help at the time of the edition of the cards.

            help mass / help mt / help nevents provided some information on the parameters.

       OM: NLO/LO Re-weighting works in multi-core

       OM: add an automatic update of the param_card to write the correct value for all dependent parameter.

       OM: add the check that the param_card is compatible with the model restriction.

       OM: Adding the run_card options "event_norm" for the LO run_card (same meaning as NLO one)

       VH: extend install command to install: lhapdf/pythia8

       VH: Interfaced MadLoop to COLLIER for tensor integral reduction.	   

       OM+VH: At the first loop/NLO computation, a new question will now be asked to choose which program

           to install to compute the loop. You can still install additional method later via the "install" command

       OM: Replace the mssm model by the MSSM_SLHA2 following the SLHA2 instead of the SLHA1 convention

     BUG FIXING

     ----------

       OM: Fix a bug in the helicity by helicity reweighting method. (introduced in 2.4.3)

       OM: Fix a bug in the reweight_card where relative path was not understood from the local directory 

            where the program was runned by the user.

2.4.3 (01/08/16)

        OM: Reduce the amount of log file/output generated for LO run (output can use up to three times less output).

        OM: For the LO combination of events (unweighting) pass to the method previously used for loop-induced.

            This method is faster and requires less I/O operation.

            This fully remove the need of the file events.lhe.gz which is not created anymore (further reduce the ouput size)

        OM: Optimise the code in order to be able to run scan with more than 2k steps.

        OM: Optimise the lhe_parser module (use for the unweighting/re-weighing/...) around 20% faster than before.

        OM: Fix a bug in MadSpin where the cross-section reported in the <init> block of the LHEF

            was wrongly assigned when multiple process were present in the LHEF and that different Brancing ratio

            were associated to each of those processes.

        RF: For NLO process generation, fix a problem with turning on PDF reweighting with sets that have only a

            single member. Also, allow for reweighting with up to 25 PDF sets (and their error members) for a single run.

        OM: Fixing bug allowing to specify a UFO model by his full path for NLO computation (thanks Zachary Marschal).

        OM: Fixing bug in LO re-weighting in case of helicity by helicity re-weighting. Now the events is boost back in 

            the center of mass frame to ensure consistency with the helicity definition.

2.4.2 (10/06/16)

        OM: fix a compilation problem for non standard gfortran system

        OM: reduce the need of lhapdf for standard LO run. (was making some run to test due to missing dependencies)

2.4.1 (10/06/16)

        OM: Fix a bug in fix target experiment with PDF on the particle at rest.

            The cross-section was correct but the z-boost was not performed correctly.

        OM: Fix various bug in MadSpin

        OM: Fix some bug in MLM merging, where chcluster was forced to True (introduced in 2.2.0)

        OM: Allow to specify a path for a custom directory where to look for model via the environment 

            variable PYTHONPATH. Note this used AFTER the standard ./models directory

2.4.0 (12/05/16)

        OM: Allowing the proper NLO reweighting for NLO sample

        RF: For NLO processes allow for multiple PDF and scales reweighting, directy by inputting lists

            in the run_card.dat.

        VH: Interfaced MadLoop to Samurai and Ninja (the latter is now the default)

        HS: Turn IREGI to off by default

        MZ: new NLO generation mode. It is more efficient from the memory and CPU point of 

            view, in particular for high-multiplicity processes. 

            Many thanks to Josh Bendavid for his fundamental contribution for this.

            The mode can be enabled with

            > set low_mem_multicore_nlo_generation True

            before generating the process.

        OM: Adding the possibility to use new syntax for tree-level processes:

            QED==2 and QCD>2: The first allows to select exactly a power of the coupling (at amplitude level

            While the second ask for a minimum value.   

        RF: In the PDF uncertainty for fixed-order NLO runs, variations of alphaS were not included.

        OM: In MLM matching, fix a bug where the alpha_s reweighting was not fully applied on some events. 

            (This was leading to effects smaller than the theoretical uncertainty)

        OM: Fixing the problem of using lhapdf6 on Mac

        MZ: Faster interface for LHAPDF6

        OM: Add support of epsilon_ijk in MadSpin

        OM: Fix multiple problem with multiparticles in MadSpin

        OM: Improve spinmode=None in MadSpin

        OM: Update the TopEffTh model

        MZ: Fix problem with slurm cluster

        OM: Improve scan functionalities

        PT: New way of handling Pythia8 decays

        RF: Fixed a bug that resulted in wrong event weights for NLO processes when requiring

            a very small number of events (introduced in 2.3.3)

        OM: Allow to keep the reweight information in the final lhe file for future computation 

        MZ: updated FJcore to version 3.1.3 (was 3.0.5)

2.3.3 (15/10/15)

        OM: Allow new syntax for the param_card: instead of an entry you can enter scan:[val1, val2,...]

            To perform a scan on this parameter.

        OM: Having two mode for "output pythia8" one (default) for pythia8.2 and one for pythia8.1 (with --version=8.1)

        RF: Rewriting of job-control for NLO processes. Better accuracy estimates for FO processes

        RF: Fix for factorisation scale setting in FxFx merging when very large difference in scale in the

            non-QCD part of a process. 

        RF: Better discarding of numerical instabilities in the real-emission matrix elements. Only of interested for

	    processes which have jets at Born level, but do not require generation cut (like t-channel single-top). 

        RF: Added an option to the run_card to allow for easier variation of the shower starting scale (NLO only).

        RF: Fixed a problem in the setting of the flavour map used for runs with iAPPL >= 1. 

        RF: Allow for decay processes to compute (partial) decay widths at NLO accuracy (fixed order only).

        OM: (SysCalc interface) Allow to bypass the pdf reweighting/alpsfact reweighting

        MZ: fixed bug related to slurm clusters

	OM: remove the addmasses.py script of running by default on gridpack mode. 

            if you want to have it running, you just have to rename the file madevent/bin/internal/addmasses_optional.py to

            madevent/bin/internal/addmasses_optional.py and it will work as before. (Do not work with SysCalc tag)

        OM: make the code compatible with "python -tt" option

2.3.2.2 (06/09/15)

        VH: Finalized the MG5aMC-GoSam interface

2.3.2(20/08/15)   

        OM: Improve reweighting module. (https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/Reweight)

            New Capabilities: 

             - allow to reweight by loop-induced processes

             - allow to change model 

             - allow to change process definition

             - allow to unweight the sample to have an idea of the statistical power.

             - allow to perform some crude reweighting on NLO sample (reweighting by LO matrix element). NLO

               accuracy is not preserved (in general) for such computation.

            New dependencies:

	     - require the f2py module (part of numpy)

	OM: change the kt-durham cut (at LO) such that particle comming from decay are not impacted if cut_decays

            is on False.

        VH: Fixed the check in helas wavefunction appearance order in an helas diagrams. It failed in cases

	    where additional wf were created during the fix of fermion flow in presence of majorana fermions.         

	RF: Fixed a bug in the aMCFast/ApplGrid interfaced introduced in the previous version.

        OM: Fix a crash when using mssm-no_b_mass model (due to the SLHA1-SLHA2 conversion)

        OM: Fix a bug in the add_time_of_flight function (not called by default) where the displaced vertex information

            was written in second and not in mm as it should. Note that this function can now be run on the flight

	    by adding the following line in the run_card: "  1e-2 = time_of_flight #threshold for the displaced vertex" 

	RF: Small fix that leads to an improvement in the phase-space generation for NLO processes

        OM: Fix a crash introduce in 2.3.0 when running sequentially in the same directory (thanks Gauthier)

        OM: Improve aloha in the case of some expression reduces to pure float.

        OM: In MadSpin, allow to specify cut for the 1>N decay in spinmode=none.

	RF: Fixed a bug that gave bogus results for NLO runs when using an internal PDF which is not

            NNPDF (like for the old cteq_6m, etc).

	RF: Fixed a bug in the PDF combination in the HwU histograms: there was no consistent use if Hessian

            and Gaussian approaches for MSTW/CTEQ and NNPDF, respectively.

        OM: Fixed a small bug in EWdim6 which was removing a coupling in AZHH interaction.

        OM: improve customize_model function to avoid problem with unity coupling.

	RF: Improved the treatment of the bottom Yukawa. Thanks Marius Wiesemann. 

2.3.1  

     OM+VH: Automation of event generation for loop-induced processes.

		OM: Automatic change of the p/j definition to include the b particle if the model has a massless b.

	RF: Reduce the collision energy for the soft and collinear tests: for 100TeV collider many were failing

	    due to numerical instabilities. 

        OM: Fixing bug associate to the epsilon_ijk structure

        OM+VH: Various bug fixing for the loop-induced processes

        OM: Fix a crash in MadWidth which occurs for some 4 body decay

        PT: Fixed a bug concerning the use of Herwig++ with LHAPDF. Bug was introduced in 2.3.0.beta

	OM: Fix a crash in ALOHA for form-factor in presence of fermion flow violation

2.3.0.beta(10/04/15) OM+VH: Adding the possibility to compute cross-section/generate events for loop-induced process

		JB+OM: Addign matchbox output for matching in the Matchbox framework

                OM+VH: Change the handling of the run_card.

                      - The default value depends now of your running process

                      - cut_decays is now on False by default

                      - nhel can only take 0/1 value. 1 is a real MC over helicity (with importance sampling)

                      - use_syst is set on by default (but for matching where it is keep off)                    

                      - New options added: dynamical_scale_choice, it can take the following value

		            -1 : MadGraph5_aMC@NLO default (different for LO/NLO/ ickkw mode) same as previous version. 

                             0 : Tag reserved for user define dynamical scale (need to be added in setscales.f).

                             1 : Total transverse energy of the event.

                             2 : sum of the transverse mass 

                             3 : sum of the transverse mass divide by 2 

			     4 : \sqrt(s), partonic energy 

                OM: Cuts are also applied for 1>N processes (but the default run_card doesn't have any cut).         

                PT: Set command available for shower_card parameters

                OM: New MultiCore class with better thread support

                RF: Fixed a bug in the aMCfast/APPLGrid interface introduced in version 2.2.3

		RF: Fixed a bug in the setting of the integration grids (LO process generation) for the minimum

		    energy needed for photons. The bug did not lead to any bias in event generation.

		RF: Re-factoring of the structure of the code for NLO+PS computations.

		RF+VH: Replaced the default topdrawer histograms with HwU histograms for f(N)LO runs

                    and allow it also for aMC@NLO runs.

		RF+VB: Allow for variable bin-sizes in MG5_aMC+aMCfast+ApplGrid runs.

                MZ+RF: Added 'LOonly' asNLO mode to export processes without any real and virtuals 

                       (useful e.g. for higher multiplicities when merging)

		RF: Added support for the computation of NLO+NNLL jet veto cross sections

		RF: Fixed a bug in the Pythia8 interface: FxFx was not correctly initialized and all

                    events were incorrectly kept (introduced in v.2.2.3)

                OM: Improve the function "print_result" (in the running interface)

                    add an option --format=short allowing to print the result in a multi-column format

                OM: Possibility to not transfer pdf file to the node for each job. 

                       This is done via a new option (cluster_local_path) which should contain the pdf set.

                       This path is intented to point to a node specific filesystem.

                    New way to submit job on cluster without writting the command file on the disk.

                OM: Allowing MadSpin to have a mode without full spin-correlation but handling three (and more) 

                    body decay. (set spinmode=none).

                OM+PA: Fixing various bug in MadSpin.

2.2.3(10/02/15) RF: Re-factoring of the structure of the code for fNLO computations.

                OM: Fix a bug in MadWeight (correlated param_card was not creating the correct input file)

		RF: When requiring more than 1M events for (N)LO+PS runs, do not go to higher precision than 0.001

                    for the grids and cross section (can be overwritten with the req_acc run_card parameter).

		RF: Make sure that reweight info (for PDF and scale uncertainties) also works for UNLOPS events.

		RF: When setting the B's stable in the shower_card, also set the eta_b (PDG=551) stable.

		OM: Change the Breit-Wigner splitting for the multi-channel integration, use the bwcutoff instead of

                    the hardcoded value 5.

                MZ: Fix to bug 1406000 (segfault appearing when doing FxFx merging). Thanks to Josh Bendavid for

                    having reported it

                MZ: Fix to a bug occurring when generating event in the "split" mode: the required output was 

                    not correctly specified

                OM: The built-in pdf "nn23lo" and "nn23lo1" where associate to the wrong lhapdfid in the lhef file

                    This was creating bias in using SysCalc. (Thanks Alexis)

                OM: Fix a bug in the LO re-weighing  module which was removing the 

                    SysCalc weight from the lhe file (thanks Shin-Shan)

                Team: Fixes to different small bugs / improvement in the error and warning messages

		RF: For aMC runs, If a NAN is found, the code now skips that PS point and continues instead of

		    leading to NAN.

                RF: For fNLO runs the virtuals were included twice in the setting of the integration grids. 

                    This was not leading to any bias in previous version of the code.

2.2.2(06/11/14) OM: Correct a bug in the integration grid (introduces in 2.1.2). This was biasing the cross-section of 

                    processes like a a > mu+ mu- in the Effective Photon Approximation by three order of magnitude.

                    For LHC processes no sizeable effect have been observe so far.

                MZ: some informations for aMC@NLO runs which were before passed via include files are

                    now read at runtime. The size of executables as well as compilation time / memory usage

                    is reduced for complicated processes

                RF: Fix crash #1377187 (check that cuts were consistent with the grouping was too restrictive) 

		RF: For NLO running: added 'strip' to the makefiles to reduce executable sizes (removes symbol info)

		Stefano Carrazza (by RF): fix for the photon PDF for the internal NNPDF sets

		RF: Improved the check on the consistency of the cuts and the grouping of subprocesse (LO running)

                PT: enabled PYTHIA8.2

                OM: restore the usage of external gzip library for file larger than 4Gb which were crashing with

                    the python gzip library

                OM: Fixing the default card for Delphes

                OM: Improve support of lsf cluster (thanks Josh) 

                OM: Adding support for the UFO file functions.py (which was ignored before)

                OM: Reduce the amount of RAM used by MadSpin in gridpack mode.

                OM: discard in MadWidth partial width lower than \Lambda_QCD for colored particle.

2.2.1(25/09/14) OM: Fix a bug preventing the generation of events at LO due to a wrong treatment of 

                      the color-flow.

2.2.0(24/09/14) VH: General mixed order corrections in MadLoop (only), including QCD/EW/QED and 

                    the UFO@NLO model 'loop_qcd_qed_sm'.

                VH: Re-design of both the tree and MadLoop matrix elements output to compute

                    contributions of different coupling order combinations independently from one another.

                VH+HS: Tensor integral reduction as implemented in PJFry and IREGI readily available

                    and interfaced to MadLoop's output.

                VH: Re-structuring of MadLoop's standalone output so as to easily create a single dynamic 

                    library including many processes at once. Useful for interfacing MadLoop to other MC's

                    and already working with Sherpa.

                VH+HS: This branch contains all the fixes for proper treatment of the latest BSM@NLO models 

                    produced by FeynRules@NLO. In particular, the fixed related to the presence of majorana 

                    particles in loop ME's.

                RF: Corrected the behaviour of the pdfcode parameter in the shower_card for NLO+PS runs.

                PT: Redesigned shower_card.dat and eliminated modbos options for Herwig6          

                RF: Change the SPINUP information in the NLO LHEF from 0 to 9 (i.e. sum over helicities)

                RF: Fixed a bug in the check on the determination of the conflicting BWs.

		RF: Added the aMCfast+APPLgrid interface (arXiv:1406.7693 [hep-ph])

                PT: Redesigned shower_card.dat and eliminated modbos options for Herwig6          

                RF: Change the SPINUP information in the NLO LHEF from 0 to 9 (i.e. sum over helicities)

                RF: Fixed a bug in the check on the determination of the conflicting BWs.

                MZ: enabled LHAPDF6 interface 

                OM: Fixed a crash in some HEFT merging case.

                OM: Fix various compatibility problem created by the LHEFv3 version (Thanks to S. Brochet)

                OM: Fix a bug for MadSpin in gridpack mode

                OM: Add a routine to check the validity of LHE file (check_event command)

                OM: Fix bug for UFO model with custom propagators

                OM: Fix Bug in the computation of cross-section in presence of negative contribution 

                OM: Change colorflow information of LHE file in presence of two epsilon_ijk

                    since PY8 was not able to handle such flow in that format.

                OM: Add the function print_result for aMC@(n)LO run.

                OM: Add some shortcut in the card edition 

                    set ebeam 500 # edit both beams

                    set lpp 0     # edit both lpp1 and lpp2

                    set lhc 14    # configure for LHC 14TeV

                    set ilc 1000  # configure for ilc 1TeV

                    set fixed_scale 100 # set all scale to fixed and at 100GeV

		    set showerkt T # set showerkt on T in the shower card

 		    set qcut 20    # set the qctu to 20 in the shower card 

                OM: Fix a bug in the card edition mode which was sometimes returning to default value

                    which were edited by hand and not via the set command.

                Seoyoung Kim (by OM): Implementation of the htcaas (super-)cluster support.

		Juan Rojo (by RF): extended the 3 internal NNPDF sets for scales relevant for a 100TeV collider.

                OM: Fix a problem with the creation of DJR plot with root 6

                OM: allow the set the width to Auto in NLO computation (width computated at LO accuracy)

                OM: Adding the possibility to have automatic plot after the parton shower for Herwig6/Pythia6.

                    This require MadAnalysis and the pythia-pgs package. 

2.1.2(03/07/14) OM: Fix a bug in ALOHA in presence of customized propagator (Thanks Saurabh)

                OM: Fixing some compilation issue with MadWeight (Thanks A. Pin)

                OM: Fixing a bug preventing MadWidth to run due to the model prefixing (depending

                    on the way it was called)

                OM: Fixing a bug in MadSpin in the mssm model

		RF: Added the invariant mass and DeltaR cuts for *same flavour* opposite sign lepton

                    pairs to the run_card for NLO-type generations.

		RF: Added FxFx and UNLOPS merging with Pythia8

		RF: Prevent an infinite loop in MadSpin by forcing the correct sign to the invariants

		RF: Catch a possible error related to grouping subprocesses and setcuts

		OM: Fix an error when using the "customize_model" command

                S. Mrenna (by OM): Fix the include file in pythia8 output to be compliant with the latest

                    PY8 version

		RF: Added a string with functional form for the scales to the event file banner (NLO only)

                S. Brochet (by OM): Fix a bug in MadSpin with the writting of the mother ID in the LHE file.

                    Force the tag in the banner to always have the same case

                    increase momenta precision for the LHE file written by MadSpin 

                    (thanks a lot to S. Brochet for all those patch)

                PT: Integrated Jimmy's underlying event for Herwig6

                OM: improve "add model" functionality allow to force particle identification.

                PT: Bug fix in the normalisation of topdrawer plots for option 'sum' (as opposed to 'average')

		RF: Fixed a bug related to the random seed when the code was not recompiled for a new run.

                OM: Fixed a bug in MadEvent(LO) run, the generated sample were bias in presence of 

                    negative cross-section. A negative cross-section is possible only if you use a NLO PDF 

                    and/or if you edit the matrix.f by hand to have a non-definite positive matrix-element.

		OM: When importing a model, check that there is not more than 1 parameter with the same name.

                PT: Subsantial recoding of montecarlocounter.f and of a subroutine in fks_singular.f. Will help 

                    future extensions like EW NLO+PS matching and numerical derivatives      

                OM: Fixing a wrong assignement in the color flow in presence of epsilon_ijk color structure.

                    Those events were rejected by Pythia8 due to this wrong color-flow.

                MZ: Added the possibility to run the shower on a cluster, possibly splitting the lhe file 

                MZ: The c++ compiler can be specified as an option in the interface. On MACOSX, clang should

                    work now

                OM: MadEvent output is now LHEFv3 fully compliant. A parameter in the run_tag (lhe_version) 

                    allows to return LHEF version 2 format for retro-compatibility.

2.1.1(31/03/14) OM: Change the way the UFO model is handle by adding a prefix (mdl_) to all model variable.

                    This avoid any potential name conflict with other part of the code. This feature can be

                    bypassed by using the option --noprefix when importing the model.

                OM: New command "add model XXX" supported. This command creates a new UFO model from two UFO model.

                    The main interest stand in the command "add model hgg_plugin", which add the effective operator

                    h g g to the original UFO model. The model is written on disk for edition/future reference.

		RF: Reduced the calls to fastjet and skipped the computation of the reweight coeffients when

 		    they are not needed.

                OM: Fixed a bug for LO processes where the MMLL cut was not applied to the event sample.

                PA: Fix a bug in MadSpin to avoid numerical instabitities when extracting t-channel invariants

                    from the production event file (see modification in driver.f, search for 'MODIF March 5, 2014') 

                OM: Better determination of which particles are in representation 3/3bar since FR is ambiguous on that point.

                    Now the determination also looks for 3 -3 1 interactions to check if that help.

                OM: Fix a bug(crash) in MW linked to the permutation pre-selection module.

		RF: Better comments in the code for user-defined cuts in the ./SubProcesses/cuts.f function.

                    Also the maxjetflavor parameter in the run_card is now actually working.

                OM: Update SysCalc to:

                      - Fix a bug that some file where sometimes truncated.

                      - Allow for independant scale variation for the factorization/renormalization scale.

                RF+OM: Improve the handling of conflicting Breit-Wigners at NLO

		RF: Print the scale and PDF uncertainties for fNLO runs in the summary at the end of the run

2.1.0(21/02/14) MADWEIGHT RELEASE:

                ------------------

                OM+PA: First Official release of MadWeight inside MG5_aMC

                      Main update:

                        - ISR corrections

                        - possibility to use narrow-width approximation

                        - introducing a module for the pre-selection of the parton-jet assignment.

                        - extended formalism for the transfer function (more generic)

                        - possibility to evaluate the weights for multiple choices of transfer function 

      			  on the same phase-space point. The phase-space is optimized for the first set of 

                          parameters.

		      Speed update:

                        - More efficient way to group the computation for identical process with different final state.

                        - Possibility to Monte-Carlo over the permutation.

                        - More efficient way to choose between the various change of variable.

                        - Possibility to use mint (not compatible with all of the options)

			- Possibility to use sobol for the generation of PS point (sometimes faster than pure 

                          random point generator.

                MadEvent/aMC@NLO UPDATE/BUG FIXING:

 		-----------------------------------

                OM: Fix critical bug (returns wrong cross-section/width) for processes where the center of mass 

                    energy of the beam is lower than 1 GeV. So this has no impact for LHC-collider phenomenology.

                    This can also impact computation of decay-width if the mass of that particle is below 1 GeV.