Note that all available option (and default value) are defined in Source/MadWeight_File/Python/MW_param_default.inc Please read that files in order to know all the possible options. The main options are explained here. But not all of them.
Block Run
Define the cluster and some constraint on the run.
Options for block MW_run | option for job schedulling (step=2,5,6) | |
tag | Effect | |
1 | cluster choice (0: single machine 1: condor 2: SGE) | |
11 | condor requirements (only for condor) | |
2 | number of data to analyse | |
21 | avoid to consider the X first event of the file input.lhco | |
22 | if True (T) add new data directory to those already existing -use only for create new dir then launch job with relaunch/refine- (since 2.1.12) | |
3 | control efficiency block by block | |
4 | normalizes Weight by cross-section | |
5 | number of points in MadGraph integral | |
6 | number of points in MadWeight integral | |
7 | [0.01] requested final precision for each weight | |
710 | [0.5] minimal requested precision for the first refinment criteria (see below) | |
711 | [0.02] first refinment criteria (see below) | |
720 | [0.1] requested precision for the second refinment criteria (see below) | |
721 | [2d-3] second refinment criteria | |
731 | [2] number of iteration in permutation selection -first loop- | |
732 | [2] number of iteration in configuration search (not used anymore -since MW2.4-) | |
733 | [5] number of iteration in permutation selection -second loop- | |
734 | [1] number of iteration in plotting diagram mode (not used for the moment) | |
735 | [10] maximal number of iteration in final loop | |
801 | [0] stop the control step when the number of idle job is strictly lower than ... | |
802 | [0] stop the control step when the total number of idle and running job is lower or equal to ... | |
9 | [F] use the cut defined in run_card.dat for weight computation | |
91 | [F] use the cut defined in run_card.dat for cross-section computation | |
weight_format | %(card)s.%(event)\t%(value)s\t%(error)s? string formatting the weighted output (python string formatting--see below--) | |
cross_format | %(card)s\t%(value)s\t%(error)s? string formatting the cross-section output (python string formating--see below--) |
In these Block you can define some (trully) advanced user option
How are computed the weights If you want to change the defaults for the integration treatment, this explains how the computation is performed. The integral to compute is in a fact cut in different pieces computed one after the other (permutation and different channel of integration) The computation is done in the following way
- Launching a survey of 2 iterations with MW_RUN_6 /2[=5000] points in each integrals.
- Check the survey, if the max of the survey has a precision less than MW_RUN_710 [50%] then going back to step one but with 5 times more points launch for each integrals.
- We organize the contribution in the decreasing contribution order.
- We refine the contribution in this order, up to the points where the contribution are pointless: when the contribution is smaller than the error on the most important contribution time MW_RUN_711 [0.01].
- This refinement is stopped when the error is smaller than the error of the main contribution time MW_RUN_721 [0.1]
- This refinement is done in MW_RUN_735 iterations, on the MW_RUN_733 first iteration the grid is adapted and we launch MW_RUN_6 points by iteration. Reducing MW_RUN_733 can solve some problems of grid instabilities
- For the following iteration, the grid is not refined anymore and the number of points is computed in order to achieve the requested precision. In consequence changing MW_RUN_735 doesn't influences the program except if you reach the maximal number of authorized points by iteration
Concerning the output formatting: The syntax is available here: http://docs.python.org/release/2.5.2/lib/typesseq-strings.html
The available information are
- card: The card number
- number: The number associated to the MW counting (number of events selected) --not available for cross-section--
- lhco_number: number of the event in the LHCO file --not available for cross-section--
- trigger: number of the trigger in the LHCO file --not available for cross-section--
- value: result
- error: integration error on the result
Block Perm
Define how to treat the permutation in MadWeight.
Option for block MW_perm | option for treating permutation (step=6) | |
tag | Effect | |
1 | makes permutation | |
2 | [T] permutates jet with bjet for weight computation | |
21 | [follows rule 2] don't consider b-tagging for the selection of events (change since 2.2.1) | |
3 | authorize one ISR jet (not implemented yet) |
Block Parameter
MadWeight can run on different parameter value on the same run. For that you have three different choice (differentiate by the value of the entry with tag 1)
- value 0: use the param_card defined by the user. Those one must be enter in the Card directory with name: param_card_1.dat, param_card_2.dat,...
- value 1: Apply from the param_card.dat the modification define in MadWeight_card.dat (see below)
- value 2: Same than one but all parameter modification are correlated (see below)
Option for block MW_parameter | |
tag | Effect |
1 | how create all the param card (see above for value description) |
2 | [F] cumulative mode add those card to the previous one (and create new event directory for those new card if step 5 already runs) |
n*10+1 | name of the block to modify for variable n (branching ratio is special see below) |
n*10+2 | tag of the value to modify (if they are a double tag use two times this line (in the correct order) |
n*10+3 | different value to insert (so you can use this tag more than once) |
Block Generation
This block parametrize how MadWeightAnalyzer will find the approriate phase-space generator. More information on those option in MadWeightAnalyzer page.
Option for block MW_Gen | |
tag | Effect |
1 | chooses which propagator to allign in ECS: 0-> thinest/ 1-> chooses more local possibility |
2 | chooses which propagator to allign in blob: 0-> thinest/ 1-> chooses more local possibility |
3 | maximal multiplicity for ECS sector |
4 | uses blob solutions which optimizes Breit-Wigner integration |
5 | uses blob solutions which optimizes TransferFunction integration |
6 | use s balanced solutions (more than one true is possible) |
10 | allows ECS A (0 neut) for change of variables |
11 | allows ECS B (1 neut) for change of variables |
12 | allows ECS C (1 neut) for change of variables |
13 | allows ECS D (2 neut ttbar) for change of variables |
14 | allows ECS E (2 neut HWW) for change of variables |
15 | allows ECS F (2 neut gamma-gamma) for change of variables |
Block EventSelection
This Block define is defined in version 2.1.13 for the first time. This block authorize some cuts selection on the particle present in the lhco file. With that blok you can easily constraint the the events used in MadWeight whitout using a complex selection. This blok has a special syntax like the following
Block EventSelection # TAG VALUE UTILITY PartName_variable min # first entry imposes a minimal value PartName_variable max # second entries imposes a maximum value
PartName must belong to
- begin (constraint on the first line of the lhco file)
- jet
- bjet (if btag used -MW_perm tag 2 and 21-)
- electron
- positron
- photon
- muon
- amuon
- tau (lhcotype=5)
- atau
- miss
- init (lhcotype=7)
!variable correspond to the different column of the lhco file and those name are the following one:
- card
- lhcoid
- eta
- phi
- pt
- jmass
- ntrk
- btag
- hadem
- dum1
- dum2
special cut on the energy is also availble
- E
Be carefull that you can change with that the definition between a jet/bjet, electron/positron,....
-- Main.OlivierMattelaer - 02 Dec 2008
- generate_min_events.py.txt: Generate for a set of param_card
Attachments (1)
- generate_min_events.py.txt (4.4 KB) - added by trac 7 years ago.
Download all attachments as: .zip