Changes between Version 18 and Version 19 of Reweight


Ignore:
Timestamp:
Feb 1, 2016, 12:11:17 PM (9 years ago)
Author:
Olivier Mattelaer
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • Reweight

    v18 v19  
    33= Description of the method
    44 The method consist to use a sample of events (weighted or unweighted events) and associate to those events a new weight corresponding to a new theoretical hypothesis.
    5  This weight can be propagate through the all simulation chain in order to avoid to have to perform the full-simulation on a huge number of sample.
     5 It corresponds to a multidimensional version of the unidimensional re-weighing method commonly used by experiments.
     6 Once computed, this weight can be propagate through the all simulation chain in order to avoid to have to perform the full-simulation on a huge number of sample.
    67 This methods works only if the original hypothesis and the new one are both significant in the same part of the phase-space.
    7  the new weight is given by
    8  $$W_{new} = |M_{new}|^2 /|M_{old}|^2 * W_{old} $$
     8 
     9 == Leading Order
     10 At Leading Order, the new weight is given by
     11 $$W_{new} = |M^{new}_h|^2 /|M^{old}_h|^2 * W_{old} $$
     12 where h is the helicity associated to the events and $|M^{new/old}_h|^2$ is the matrix element for that helicity.
     13 If the events is not associated to a specific helicity, then the sum over the helicity is used instead.
     14
     15 This method is fully LO accurate and do not present any bias. Note that the statistical fluctuation of the original sample can be enhanced by the reweighting.
     16 To get an idea of such propagation, one can use the naive formula of propagation of error:
     17  $$\Delta\mathcal{O}_{new} = \bar R \Delta\mathcal{O}_{old} + \Delta R \mathcal{O} $$
     18 where $\bar R$ is the avarage of the ration of the matrix-element and $\Delta R$ the associated variance.
     19
     20
     21 == Next-to-Leading Order
     22 For Next to Leading order samples, two type of reweighting are available:
     23 1. '''kamikaze reweighting:''' This correspond to a Leading order type of reweighting where the reweighting is done only with tree-level matrix element:
     24     a. For S-event (N particles in the final state) the weight is given by
     25 $$W^S_{new} = |M^{new}_{born}|^2 /|M^{old}_{born}|^2 * W^S_{old} $$
     26     b. For H-event (N+1 particles in the final state) the weight is given by
     27 $$W^H_{new} = |M^{new}_{real}|^2 /|M^{old}_{real}|^2 * W^H_{old} $$
     28
     29 2.  '''NLO reweighting:'''
     30
     31
     32
    933
    1034= Reweighting of NLO sample
    … …  
    110134The '''-f''' options prevent the question to be asked.
    111135
    112  = Options
    113   Note all the following options will be available in Madgraph5_aMC@NLO since version 2.3.2
     136 == Options of the code:
     137  Note all the following options will be available in Madgraph5_aMC@NLO since version 2.3.2.
     138  They have to be included in the reweighting_card before the first launch command.
    114139   1. "change model <XXX>" performed the reweighting within a new model (you then need to profide a full param_card and not a difference)
    115140   2. "change process <XXX>" change the process definition of the process.
    … …  
    118143   5. "change helicity <True|False>": perform reweighting for the given helicity (True --default--) or do the sum over helicity.
    119144   6. "change rwgt_dir <PATH>": change directory where the computation is performed. This can be use to avoid to recreate/recompile the fortran executable if pointing to a previously existing directory.
    120    7. change mode LO:
     145   7. change mode LO: For NLO sample, this flag force to use the kamikaze reweighting (Leading Order type rew
    121146
    122147