Judging from Physics Reference manual of Geant4 and it reference paper as gEnergy loss in thin layers in GEANT (NIM A362 (1995) 416)h, it looks like current implementation of G4UniversalFluctuation has wrong conditional clauses. Line 105th if(meanLoss < minLoss) return meanLoss; Does not required or not fit an approach of Urban model in the paper. Line 175th if(suma < sumalim) // very small Step is changed to if ( exp(-suma) > sumalim) // very small Step Please read Physics Reference Manual http://wwwasd.web.cern.ch/wwwasd/geant4/G4UsersDocuments/UsersGuides/PhysicsRef erenceManual/html/node31.html equation No (7.23) and following statement and also equation (12) of the paper. Here is the point judging very small step or not. I did test changed version and compare to the original and you can get the result form http://www.slac.stanford.edu/~tkoi/energy_loss_fluctuation.ppt . Please compare also the result to the figure 5 and 6 of the paper. Original result (G4v7.0) is similar to the figure 5 which obtained with Geant3.1.5 and fixed result is similar to the figure 6 which obtained with Geant3.1.6. (v7.0 should be similar to the later.) Please take care, in the paper gamma (lorentz factor) and its corresponding energy of electrons are wrong. Ex, in the figure 4, gamma = 1000 is corresponding 5.11 GeV e-(511MeV is correct valued). I believe gamma values are correct and energy values are wrong form judging plot shapes and average energy losses in the table 1. So that I did calculate electron energy of 235 keV (gamma = 1.5) for lower plots of the figure 5 and 6.
Thanks for your study. With Vladimir and Laszlo we are studying this point, and we have already commited a corrected version which come back to values closes to geant4.5.2 Will be included in the next public corrections patch. It you have access to the repository, please try stand-V07-00-00
