wiki:DevelopmentIssues

Version 39 (modified by Olivier Mattelaer, 12 years ago) ( diff )

--

Public release of MadGraph5 v2.0 including aMC@NLO

List below the different points which necessarily need to be adressed before the public release of MadGraph5 v2.0.

When editing the above, please try to stick to the self-explanatory conventions (Syntax, fixed numbering and placement in the right section when fixed).

To fix before v2.0 beta release

| J.A. | Have fixed-order and showered td plots automatically put in HTML/<run_name>/plots.
This would be used by the experts only as it necessitates modification to the [...]_plot.f codes, but it is still desirable before the release.
aMC@NLO general | O.M | Having all acceptance/parralel test working !!
aMCatNLO 75| O.M. | launch LO is crashing
aMCatNLO 76| O.M. | any syntax error in the run_interface makes this one to crash, and the file RunWeb is not corrcetly remove (forbidding to rerun in this directory)

To fix before v2.0 release

no mail number|OM| Contextual help in aMCatNLO_run interfaces not valid (indicates multi_run for example)
4 | O.M. | Generate some html output for checking the generation of the fortran code and of the output.
22FKS | All | We need to discuss what functionality we want to release,
in particular the complex mass scheme. It is not correct as such right now but people could easily correct it and scoop us. Same for QED in principle. What feature do we want to lock and how? Please make your proposal below. R.F. | I wouldn't worry about this. The code is complex and before somebody figures out how to do this, we should have release it.
9 | V.H. | Fix RAMBO phase-space point generation for 3-body interactions
simply always generate the single configuration with the third particle at rest. It is good enough for the checks.

To be adressed (the fatest the better)

15FKS | O.M. | Generalize the function get_interaction_q_power
so that it uses aloha to retrieve this information (comment copied from FKS5 branch review) R.F. | Is this necessary before the release? We support only virtual corrections within the Standard Model.
28FKS | O.M. | Refactor helas_call_writers.py
by creating a loop version of this file to be put in the 'loop' directory and creating a LoopHelasCallWriter daughter class to build more elegant structures for the loop helas output. (comment copied from FKS5 branch review)
25FKS | V.H. | We should improve our handling of gauges in MG5
so to make it part of the UFO model which would now also specify the propagators of the particles included. This would allow for many gauges and the specification of any value for the chsi parameter. It could also possibly help for the exploration of the spin 3/2 landscape of propagators we have now.
32FKS | O.M. | Refactor process_checks.py by creating a loop_process_check.py
should be quick and very superficial anyway.
no mail number|OM| Having the shower to run in multi-core/cluster (This is very long and is easy to split)
aMC@NLO 44 | O.M. | Having the compilation/test on multi-core, and possibly also the generation/output
| V.H. | Adding the option of having double-double for multiple-precision instead of true quad-prec.
This would be much faster (slow-down of factor 8-10 for dd vs 100 for qp.) and also allow compilation with gfortran <4.6.

Fixed issues

[ Fixed ] | They now pass. But we should not put this on the wiki again as it is obvious that all test have to pass for the release :).

aMC@NLO 20 | O.M. | The tutorial aMCatNLO and the help of launch should explain the details of launch especially for running the shower

[ Fixed ] M.Z. | Minimal info now given in the tutorial about the '-i' option of launch. Then the user can sort out by himself what to do once in the run interface.

aMC@NLO 71| O.M. | aMCatNLO-Utilities should auto-detect the compiler/ the fact that we are in 64 bit/…
and also having a file make_opts used by all sub makefile present in this utilities.

[ Fixed ] M.Z. | in r370. (and 16 for MCatNLO-utilities)

aMC@NLO 5 | J.A. | When launching aMC@NLO some 'set' commands are not recognized and output their help.
This is because aMCatNLOLauncher uses exec_cmd from ExtendedCmd which in turn will call the do_set() of common_run_interface.py for which 'check_set()' does not recognize some options like complex_mass_scheme and loop_optimized_output.

[ Fixed ] M.Z. | Apparently got fixed when moving to trunk version 1.5.2.

aMC@NLO ?? && ?? |J.A && O.M | Change the tag before the > for the aMC@NLO run interface.
So to make it explicit where the user stand and avoid confusion.

[ Fixed ] V.H. | I changed it to aMC@NLO_run>.

2 | O.M | Commands help and auto-completion
Update the auto-completion and help for the loop functionalities.

[ Fixed ] V.H. | The auto-completion should now be ok, especially when entering loop processes definitions. The help has gained a color scheme and is now updated with the loop features except for the aMC@NLO launch section which Marco will fill in very soon. So let us know if anything suspicious is left in this department.

10 | M.Z. | Fix the STOP: momentum conservation errror from madloop
[ Fixed ] M.Z. | (r355), it was a stupid bug inside check_poles
1 | R.F | Default coupling orders at NLO
Change the default chosen coupling order configuration for processes which can have mixed order contributions at LO already. Typically, one wants p p > t t~ [QCD] to return only contributions with QED = 0. [ Fixed ] V.H. | Now in the case proposed above, it will inform the user that the configuration (QED=0,QCD=*) is chosen and warns that additional contributions of configuration (QED=2,QCD=*) are detected but discarded.
3 | V.H | Tracking of unstable points in MadFKS
For now simply from the IR cancellation test. At the end of a run, it should give their number versus the total number of calls to the virtual as well as writing out the first ten (with the renormalization scale used) in a log file like 'UPS.log' in the G* folder. [ Fixed ] M.Z. |
5 | O.M. | Have a nice error/treatment if fast jet is missing
[ Fixed ] M.Z. | (r341)
6 | O.M. | Explaining how to launch the shower (and the point for the counter-term in the tutorial
[ Fixed ] M.Z. | (r343)
7 | O.M. | add process crash in MC@NLO (if no generate are done before)
[ Fixed ] M.Z. | (r341)
8 | M.Z. | param_card updates do not affect results
[ Fixed ] M.Z. | (r352), was only related to tests
S1 | S.F. | Stefano Frixione pointed out that the code does not run on the lxplus machine at CERN because the readline pythone module is not recognized.
Olivier, I know that this module is very OS-sensitive, and that you have been hardcoding checks on its versions to handle. Do you know what goes wrong, can you fix it?

[ Fixed ] O.M. | was not related to readline but to md5

Attachments (3)

Note: See TracWiki for help on using the wiki.