************************************************************************** RELEASE 1.0 ************************************************************************** - bug that affects Z production. In the file Z/Born_phsp.f, where ph_Wmass is used instead of the correct ph_Zmass in the subroutine set_fac_ren_scales - bug that affects gg_H production. Wrong scale in the file H/Born_phsp.f - from revision 437, corrected bug in the fermion-loop contribution to the virtual amplitude in the Wp_Wp_J_J process. It affects the distributions at the percent level. - from revision 459, corrected bug in programs that merge t and tbar samples (or W+ and W- samples): formula for the error on the total combined cross section was wrong. - from revision 474, corrected bug that affects reweighting of remnant contributions for finite top-mass corrections in gg_H production. Corrections increase if used in conjunction with hfact suppression. - from revision 487, corrected bug in lepton mass assignment. It may affect Zj, single-top production. - from revision 1040, corrected bug in the Les Houches event file. Event weights were not multiplied by the branching fraction, when the decay was implemented a posteriori, and when producing weighted events - from revision 1234, flg_withdamp is set to true in the pwhg_processes.f file - from revision 1267, default behavior for VBF Higgs boson production is with running width in Breit Wigner; old behaviour (fixed width) is restored with higgsfixedwidth 1 in powheg.input. It makes difference when the Higgs boson width is very large. - from revision 1380, added option to generate s- and t-channel single-top events without top decay products, i.e. without spin-correlation effects (requested by CMS). See manual for details. - From revision 1662, a bug affecting the remnants in Zj, WZ, WW, ZZ, HJ and HJJ was introduced, that caused zero remnant cross section. This bug was fixed in revision 1665. Thus, in all intermediate revisions, the above processes were in an inconsistent state. - From revision 2085, a bug affecting spin correlations in t-tbar decays in the antiquark-gluon and gluon-antiquark channel has been fixed - In revision 2029 (first version of ST_tch_4f) a wrong file was present, and results were clearly wrong. Bug fixed from revision 2084 (local copy of Bornzerodamp.f removed) - From revision 2117, an inconsistency in sigborn_rad has been fixed. The bug was such that the soft and collinear limits, used in radiation generation for determining the boundary of the real remnant, were computed using the btilde scales, rather than the radiation scales. The corresponding effect is of NNLO order, but the code was not supposed to behave this way. So, minor (NNLO) effects affecting radiation generations may be present as differences between version 2117 and all earlier ones, for all processes that are of non-vanishing alpha_S order at the Born level, and that use the bornzerodamp feature. - From revision 2159, RM48 BUG fixed. A bug in cernroutines.f, RM48 (Cernlib routine) was found and fixed. The call to RM48IN(ISEED,N1,N2), that should bring the generator in a state of initialization with seed ISEED after N1+N2*10^9 calls, if I2>0, fails. This yields problem if one needs to restart the program at a given random state (with N2>0). In particular, reweighting fails for events with N2>0. - From revision 2160, fixed problem in gen_radiation, that was assuming that the regular remnant routine gen_leshouches_reg was setting the weight of the event consistently. In practice, for several processes the weight was set to 1, which was adequate when the only generation mode was with weight 1, but became incorrect when weighted events were introduced. Now gen_radiation takes care of fixing the weight consistently. It affected the following processes: gg_H, gg_H_quark-mass-effects, ST_wtch_DR, ST_wtch_DS, Wj. However, the effect is essentially invisible, since regular remnants yield very small contributions. - From revision 2165 an MCFM bug was corrected in the implementation of the virtual corrections in the HJJ code. The bug was found by the GOSAM collaboration. The effect of the bug was found to be very small for typical distributions. - From revision 2176 a bug was corrected in the plotting of fixed-order NLO distributions for ggH, when a rescaling factor is present (default). The bug only affected the internal POWHEG-BOX plotting machinery for fixed order distributions and does not affect the final cross section or the LHEF output. Thanks to E. Bagnaschi for spotting this. - From revision 2218 on: a bug was corrected that affected the flavour assignment of the real radiation. In gen_radiation.f, at the end of the routine gen_radiation, a call to sigborn_rad is needed before calling sigborn_real and gen_real_idx. The call has an effect if bornzerodamp is on, since the sigreal depends upon the soft-collinear cross section, that in turn depends upon the Born cross section. The dependence is in the bornzerodamp function. The bug caused a (rare) appearance of a message: "POWHEGBOX:pick_random: could not pick a value" - From revision 2264: setting ih1 2 and/or ih2 2 allows doing neutron beams. Implemented according to a suggestion of Jan Kretzschmar. Antineutrons (-2) are also possible, although not useful. - From revision 2330: fixed a bug in the hvq package, such that the fifth component of pup (i.e. the nominal mass) of the W- second decay product (the last entry in the Les Houches record) was set up equal to the W+ second decay product (the third to last entry). The four momenta were not affected by the bug. Found by Christian Schwanenberge and reported by Gia Khoriauli for ATLAS. - From revision 2361: fixed bug in reshufflemoms. Occasionally the program tried to reshuffle a single particle, leading to a small fraction of discarded events. - From revision 2443: fixed bug due to an incompatibility between the way MadGraphStuff assigns the underlying Born colour and the smartsig flag. The born_colour routine in the MadGraphStuff relies upon the fact that when the Born amplitudes are computed, the value of the colour subamplitudes are kept in storage. However, when smartsig is on (which is the default) not all subamplitudes are computed, since the BOX detects is some amplitudes are equal, and only computes a minimal set, possibly leaving the value of some subamplitudes in an inconsistent state. This bug affects: HJ HJJ HWJ HZJ that are the only subprocesses that use the MadGraphStuff. Colour assignment affects the radiation carried out by the shower Monte Carlo, i.e. beyond the hardest jet. It is unlikely that this bug may cause observable effects. - From revision 2538: bug fixed in revision 2443 was not fixed in Version-pre2-1/sigborn.f, causing the relevant programs to segfault (reported by Florian Bernlochner). - From revision 2557: bug fixed in HZJ and HWJ concerning the color assigment in the event Les Houches file. Bug fixed in the virtual part of the HZJ due to zero width in the Chi propagator. It is unlikely that these bugs may cause observable effects. - From rev. 2679: revised setting the light quark masses in the hvq program. In rev. 2528, following requests from an experimental group, the options bmass_lhe and cmass_lhe were introduced. Inserting the lines cmass_lhe 1.5 bmass_lhe 4.8 in the powheg.input file affects the mass assignment of the c and b quarks whenever they are treated as light quarks (i.e. both in top production, only charm in b production, and none in c production). The generation of the LH event is completed by reshuffling the final state momenta in such a way that the c and b quarks, if treated as light are assign the given masses. This modification caused a problem in t tbar production with decays. In this case, the mass of quarks arising from the top decay products are set according to the variables tdec/cmass 1.5 tdec/bmass 4.8 but the call to the reshuffling routine (called lhefinitemasses) washes out these assignment. In order to recover a reasonably consisten behaviour, since rev. 2679 the behaviour is as follows: 1) Light quarks are reshuffled in all cases to the default values 0.33d0,0.33d0,0.5d0,1.5d0,4.8d0, and charged leptons are reshuffled to the values 0.51099891d-3,0.1056583668d0,1.77684d0 2) In case the program is doing top production with decays, these values are overridden by the variables tdec/emass, tdec/mumass, tdec/taumass, tdec/dmass, tdec/umass, tdec/smass, tdec/cmass, tdec/bmass. 3) The cmass_lhe and bmass_lhe parameters, if present, override all previous assignment. - From rev. 2819: workaround to an lhapdf6 bug in the setq2min routine, that causes the member of a pdf set to change when setq2min is called. - From rev. 2993: fixed bug in MadGraphStuff, that prevented the creation of the Born cross section in certain cases (as in single top production) - From rev. 3073: fixed major bug affecting the W2jet generator, such that the kinematics of mu- and anti-nu-mu was interchanged in the W- -> mu- and anti-nu-mu. The bug did not affect the electron channel and the mu+ channel. Reported by Chris Hays from ATLAS (early June 2015) - From rev. 3076: fixed a major bug affecting the VBF_W-Z generator. In W+ production, the charged lepton and neutrino momenta were exchanged. Reported by Chris Hays from ATLAS (early June 2015) - Rev 3443: Fixed bug in the HJ virtual amplitudes. The amplitudes computed by MCFM were already in CDR. So no need to perform the CDR -> DR conversion Problem reported by Gionata Luisoni