Compiled list of most modifications (as well as corrected bugs) by chronological order (record starts 10 Sept. 2001) N.B: it is almost impossible to record here every single code lines affected eventually by these corrections of bugs or modifications of the code: we simply list a (hopefully) comprehensive list of (main) bugs found and corrected, or other changes/improvements, but without going into the detailed changes of code lines etc. 66. version 2.52 (Oct 20 2019): -Two bugs corrected, that were introduced in version 2.51 in the improved radiative corrections (RC) to chargino masses (routine SU_RCCINO): has rather minor effects for most of parameter choices, except in some specific parameter regions. -In chargino and neutralino RC, a new algorithm deviating from strict DRbar scheme (but by formally higher order terms): using (ino) pole masses, only within ino RC loops, to avoid inconsistencies in m_chi^+_1 ~ m_chi^0_1 regions. -miscellaneous algorithmic modifs with respect to v2.51, many changes due to better overall consistency between Suspect2 and Suspect 3 C++ versions, that have very tiny effects on results. 65. version 2.51 (Sept 10 2017): miscellaneous modifs with respect to v2.5, minor bugs corrected (mainly algorithmic, not affecting very much the final results). Mainly many changes due to overall consistencies with Suspect 3 C++ version. NB now agreement between Suspect 2.51 and Suspect 3 is below 10^(-3) or better for many tested scenarios. *** LATEST changes/corrections in SuSpect version 2.5 (Nov 16, 2016): 64. Major modification: more complete squarks, sleptons, sneutrinos, chargino, neutralino mass radiative corrections. (minor changes in the spectrum in e.g. mSUGRA model for not extreme parameter choices, but substantial impact on some pole masses in general MSSM in extreme/fine tuned regions of parameter space, like e.g. large sfermion mixing, large gluino mass, etc, regions.) 63. All sfermions and gaugino mixing angles as given in output are now DRbar 'tree level', evaluated at the unique input EWSB scale, and similarly for U,V,Z chargino+neutralino matrix elements: this is more 'universal' than previous version ones, that were calculated at a scale equal to geometric average of pole masses, including radiative corrections, and should avoid hidden double counting of R.C. effects. The impact on the sparticle mass spectrum is very moderate. *** Previous changes/corrections in SuSpect ver 2.43 (May 23, 2013): 62. Odd results for lightest h mass had been signaled (independently by Luc Pape and Yanou Cui) in the low scale input case with MA_pole, MU input and for (very) large scalar soft terms. After investigations, it was traced to the fact that lightest Higgs mass R.C. involve large cancellations between different contributions at accuracy of the order of the fine-tuning. Though the algorithm iterates until reasonable accuracy is reached, previous defaut maximal accuracy ~10^-3-10^-4 was not adapted to larger fine-tuning (as is more commonly explored these days). Moreover pole neutralino/chargino masses rather than DRbar ones were inconsistently used within loops at the last iteration (only for this low scale MA, MU input), which is normally a negligible higher order O(alpha^2) difference, but for large soft masses it caused large perturbations relative to the required accuracy, spoiling precise cancellations within loops. 61. Reading of SLHA input file issues with GNU fortran compiler corrected. *** Previous changes/corrections in SuSpect ver 2.41 (August 5, 2008): 60. Minor bugs fixed in SLHA (or other format) input/output reading/writing: -missing output scale writing for new bottom-up RGE option (signaled by Joseph Pradler) -missing arbitrary high scale input possibility (signaled by Csaba Csaki) -problems with some reading of strings characters in SLHA input (signaled by Sasha Pukhov) -mismatch for some input cases with old form of input file suspect2.in NB: none of these corrections affect the final spectrum results 59. bug corrected in faster RGE algorith introduced in previous ver 2.40 (was producing some instabilities for one-loop RGE choice in some corners of MSSM parameter space). NB for ``defaut" input choice (i.e. 2-loop RGE, full radiative corrections etc) this should affect the final spectrum by only minute changes. ** Previous changes in SuSpect ver 2.40 (May 8, 2008): 58. Faster RGE (Runge-Kutta) algorithm 57. Improved (simplified and faster) algorithms for some input situations (typically MA_pole,MU input etc) 56. Upgraded compatibility with latest SLHA (version 2) input/output files and conventions: (hopefully? fully compatible!) NB none of these modifications affect the 'basic' spectrum calculation, i.e no modifs affecting directly the spectrum calculation were made since previous version, so sparticle masses should be identical to ver. 2.34 (up to minute changes due to slight modifications in RGE algorithms, precision, etc). NB2 IMPORTANT: to users of previous versions 2.34 with SLHA input format: please note the following important changes in SLHA parameter conventions in the input file: suspect2_lha.in (all this is now conform with SLHA): *in block MODSEL: parameter value 1 is now for SUGRA (was 10 previously) 2 is now for GMSB (was 11 previously) 3 is now for AMSB (was 12 previously) *then, general MSSM at GUT scale is now driven by value 1 in MODSEL (SUGRA) and any arbitrary MSSM soft term values to be defined in block EXTPAR *in block EXTPAR: -parameter 26 is now MA_Pole input (previously was sign(mu)) -parameter 24 is MA_running(Q_EWSB) (previously was Ma_pole) ** New input/output choice options: 55. MA_pole, MU(EWSB scale) input choice with other input at GUT scale (SUGRA , non-universal, or general MSSM): this covers e.g. the case of Non Universal Higgs Models (NUHM) typically. 54. NEW: ``Bottom-up" RGE option: allows to start from arbitrary MSSM at low (arbitrarily chosen) scale, and run parameters consistently up to GUT (or any arbitrary) scale. Input file example in: suspect2_lha_bottup.in (sps1a benchmark point at GUT scale recovered from low energy input with MA_pole, MU(EWSB scale) input; corresponding output in suspect2_lha_bottup.out) -Bugs corrected in SuSpect 2.40 w.r.t. ver 2.34: 53. b-> s gamma calculation: one mixing angle convention input for interface with b-> s gamma (external) subroutine calculation (bsg.f) was inconsistent: can affect BR(b->s gamma) values in some general MSSM cases (not affecting mSUGRA cases). 52. CCB (charge and color breaking) constraints: some of these were incorrect (these constraints are printed as optional output, if required, through specific SuSpect options flags) ** Previous modifs in version 2.34 of May 13, 2005: 51. Addtion of two-loop RGE for all squark, slepton and trilinear soft-susy-breaking terms (affects subroutine "SU_DERIV2"). These higher order effects can have a non-negligible impact in some regions of the MSSM parameter space. -Previous modifs in version 2.33 of March 01, 2005: 50. Some additional useful informations (running A mass value, etc) + some typos corrected in SLHA format output file: suspect2_lha.out (pointed out by Sasha Pukhov) 49. Bug in Br(b->s gamma) calling (one input parameter uncorrectly defined) correction affects a bit Br(b->s gamma) values ** Previous modifs in version 2.32 of February 14, 2005: ** 48. A few cleaning up of writing in suspect2.f file 47. Reshuffling in output writing of first and second generation sfermion masses, to be consistent with SLHA conventions (so that e.g. LEFT = heaviest in mSUGRA) (Pointed out by Asesh Datta) 46. Additional input/output choice possibility: for INPUT=1 (i.e. NO reading of input file(s), all input defined in calling routine), now both standard SuSpect and SLHA -OUTPUT- format files are generated. Maybe convenient for scanning and/or for some interface with other codes. (NB this necessitated additional writing in the main routine in suspect2.f as well as in subroutine "SU_read_leshouches"). 45. Bug in g_mu -2 susy contribution: def. of smuon mixing angle was inconsistent for some input cases, in our recently included subroutine su_gminus2(..). Moreover, we now added the ~ -7% correction from leading-log 2-loop QED. (Pointed out by Manuel Drees). (Correction affects of course g_mu -2 output values). 44. Bug(s) in Br(b -> s gamma) contribution: 1. running charm quark mass was improperly used, instead of charm pole mass, as input of the relevant subroutine: bsg(..). 2. Another undefined b-> s gamma input (pointed out by Sasha Pukhov) (Corrections affects of course Br(b-> s gamma) output values). 43. Bug, IF choosing 'one-loop only' Higgs mass calculation (i.e. only for input choice: ichoice(10)= 1), Higgs masses output values were not the consistent (DRbar scheme) ones. (Pointed out by Nada Sahoury) Note this bug did not affect in any way the more complete Higgs mass setup input choice: ichoice(10) =2 (i.e. full 1-loop + dominant 2-loop), presumably set by defaut by most users. 42. Bug (typo) corrected in one-loop corrections to stop masses: in contribution "crLRgau" (nb a generally very small contribution, so the corresponding changes in the final stop pole masses appear to be very negligible, for various input choices which have been checked) ** previous modifs in version 2.31 December 20, 2004: ** 41. Includes information in output files on experimentally constrained (g_mu -2, Br(b-> s gamma),..) or theoretically constrained (fine-tuning) quantities. Corresponding added subroutines (external file): bsg.f for Br(b-> s gamma) from Paolo Gambino; other subroutines (g_mu -2 etc) included in main routine suspect2.f. 40. to read input and write output files in SLHA (les Houches) format: two subroutines added: 1) to read input file in the SLHA format: subroutine SU_read_leshouches(ninlha,ichoice,imod) 2) to write output file in the SLHA format: subroutine su_lhaout(nout,ichoice,errmess,imod) -Plus one additional input file for SLHA format: suspect2_lha.in NB the parameters nl, nq (nb of lepton/ quark messengers in GMSB models) HAVE BEEN CHANGED from "integer" to (double precision) real (for compatibility with SLHA formats) -Modifs in version 2.3 of June 16, 2004: 39. A major modification is the consistent calculation in the DRbar renormalisation scheme of Higgs mass radiative corrections at one-loop + dominant 2-loop (courtesy of Pietro Slavich routines and input). Plus many additions/improvements in other radiative corrections, in particular more precise for top, bottom, tau masses, few bugs corrected (nb a more detailed description of the latest main changes/corrections will be given here soon!) -Previous modifs/corrected bugs in version 2.2 of Dec 19, 2003: (many thanks to Ben Allanach, Sabine Kraml, Sasha Pukhov, Pietro Slavich for many of these bugs/modifications signaled/initiated by them) 38. A new common/su_runmewsb/.. in suspect2_call.f to make available from outside suspect the running tau,b,t masses in DRbar scheme at EWSB scale: can be useful for many purpose. 37. Slight changes (suggested by Dirk Zerwas) in chargino diagonalization (redefs of diag. mat. elts) to be able to deal with BOTH M1,M2 <0 (nb previously was working only if EITHER M1<0 OR M2<0.) 36. Many improvements/refinements in the ichoice(1)=0 (direct spectrum calculation at arbitrary scale without RGE from GUT): Now there is RGE from mz to the (input chosen) arbitrary scale to be consistent with, e.g., Yukawa+gauge couplings being always defined at MZ scale. 35. Some small refinement in the rad. corr. to stop masses, now evaluated at 3 different momenta scales: mst1(p=mst1); mst2(p=mst2); theta_stop(p=sqrt[mst1*mst2]) 34. Upgraded interface with FeynHiggs(Fast) latest (1.5) version: many small changes in input transmitted to FeynHiggsFast; a major one is that the MA input to Feynhiggs is the (one-loop) pole A mass (rather than the running A mass, as it was uncorrectly made in previous version: can make some non-negligible corrections to Higgs masses for large tbeta). Also, whenever Feynhiggs calculation fails (as it may happen at VERY large tanbeta due to too large 2-loop corrections in On shell to DRbar scheme transitions) suspect give mh(1-loop) simple approximation result, with an appropriate warning. 33. Change of the meaning of the control parameter ichoice(9): gives now a more convenient and reliable check of final accuracy in spectrum: see suspect2.in for the present meaning. (Previously ichoice(9) was the nb of iteration to be guessed by the user for sufficiently good accuracy; while now it simply takes values: 1 or >= 2 and the nb of iterations are automatically calculated to give an accuracy of 1 % or 0.1 % respectively on the final spectrum result. Note that with pb 32. below now solved, this accuracy generally only needs 3 to 4 iterations). 32. Corrected unstability problems (resulting in lot of iterations needed) at large m0. Was essentially due to a lost of precision within our naive B0 def. due to cancellations occuring between different B0(q,m1,m2) loop functions: Now much more professional and precise B0 functions are implemented (from LoopTools http://www.feynarts.de/looptools/) which completelly solved such unstability problem. 33. Corrected bug: rmtop =y(6) -> rmtop =y(6)*vu : was only affecting the WRIting in output file of running top mass at GUT scale, but not its actual calculation. 32. Corrections in the grand unification scale determination algorithm: previously a bug implied that for ichoice(3)=1 the scale at which g1=g2 exactly was evaluated from the first iteration only: there are thus some small changes in this scale which affect a little the RGE for ichoice(3)=1. The changes on the final spectrum are however very mild for most of the parameter space. 31. GNU (Linux) Fortran Compiler initialization problems solved: the unconvenient compilation option: -finit-local-zero is no longer needed. So compile simply with: g77 -c suspect2.f suspect2_call.f HOWEVER!: our FeynHiggs version (feynhiggs.f given in our package) still needs this compilation option to work. 30. Corrected output writing error: quark masses and coupling presented in suspect2.out as evaluated at the EWSB scale were in fact the ones calculated at MZ scale. (NB changes in the presentation: now g1, g2, g3 are written instead of previously g^2_i). 29. Major modif of SUSY rad. corr.: suppression of threshold effects in RGE beta functions: instead now all (one-loop) SUSY +SM rad. corr. in gauge and Yukawa couplings are taken into account in the initial condition for the relevant couplings. This is theoretically equivalent, but numerically makes some small differences. (If interested, see for more details the new subroutines: su_runningcp and su_pixx and related calculations). 28. eq. to define running MA was uncorrect in the general MSSM (ichoice(1)=1) case, which made MA therefore inconsistent with the one in e.g. the mSUGRA, GMSB or AMSB case (the latter had correct MA eq. however). 27. bug in the rge of Ytau 9/4 -> 0..Instead should be 9.d0/4: should have small effect however. 26. Suppressing the previous freezing of Ytop running from mz to mtop: now R.C. to Ytop, Yb, Ytau are all defined at mz scale (and in the DRbar scheme). 25. -Coefficient of alphas**2 two-loop mt_pole/mt_running relation in DRbar scheme was uncorrect: small effect but rather important for Ytop precise value and thus affect MU value via RGE. -Last modifs/correct bugs in version 2.102 of Jan 30, 2003: 24. -in subroutine SU_GINOCR (rad.corr to gluino mass): relative sign of "B0 contributions" were wrong -> Corrected -> affects final gluino mass by a few 0.1 % 23. -Coefficient of alphas**2 two-loop mt_pole/mt_running relation in DRbar scheme was uncorrect (lower) by 10% -> corrected -> affects Y_t (top yukawa coupling) by about 0.1 % 22. -Quark masses and coupling written in output file suspect2.out as evaluated at the EWSB scale were in fact those at MZ scale. (signaled by A. Pukhov): corrected. (NB the EWSB scale values were consistently used however in calculations when relevant) 21. -in subroutine SU_SUSYCP: in the simple approximation choice for Higgs mass calculation (ichoice(10)=0): use of DREAL instead of DBLE to take the real part could result in compilation error on some machines (signaled by A. Pukhov): corrected. 20. -in subroutine SU_SUSYCP: in the simple approximation choice for Higgs mass calculation (ichoice(10)=0): several parameters undefined were taken zero (signaled by N. Ghodbane, E. Perez, A. Pukhov): --> However NO effects, the quantities calculated were not used in the final Higgs mass results, in this simple approximation. 19. -Unconstrained pMSSM case without RGE (ichoice(1)=0) produced some NaN results (unless a msugra or other model with RGE was called before a call with ichoice(1)=0): (signaled by A. Pukhov): corrected. (top,bottom, tau Yukawa couplings at EWSB scale were undefined) 18. -Unconstrained pMSSM case without RGE (ichoice(1)=0) heavy Higgs masses were slightly different from a msugra case with same input. (signaled by A. Pukhov): corrected. (MA defining eq. was slightly different and not as precise, in this particular pMSSM case) 17. -AMSB model (subroutine su_amsbsub): ct and cb (m0 coefficients) were undefined and taken zero should be cu, cd respectively (signaled by N. Ghodbane, E. Perez) --> affects AMSB sbottom, stop masses by about 10%! Things being added/corrected in Nov 20 (suspect 2.100) new version: 16. corrected 25/11/2002: MA value was previously not recoverable from calling code via COMMON/SU_MSSMHPAR/mhu2,mhd2,ma,mu (signaled by Nabil Ghodbane) 15. Corrected 25/11/2002: potentially dangerous mismatch in common length in COMMON/SU_MATINO/u,vv,z,xmn: xmn was previously missing in suspect2_call.f (signaled by Nabil Ghodbane) 14. Many convenient and cosmetic changes (e.g. all subroutines and commons names being changed as: name -> su_name to facilitate interface with other codes (to avoid double definitions clashes) 13. Correct bug: a few previously wrong U(1)_y assignements in some of the (non-dominant) one-loop effective potential tadpole contributions and one-loop corrections to the top mass: very minor numerical changes (about a percent at most change in those corrections). 12. Adding AMSB scenario: one subroutine (amsbsub(..)) and one common for interface with external calling of suspect (common/AMSB/..) added (plus all necessary input/output implementation: a few minimal changes in suspect2.in input file)) 11. Adding GMSB scenario: one subroutine (gmsbsub(..)) and one common for interface with external calling of suspect (common/GMSB/..) added (plus all necessary input/output implementation: a few minimal changes in suspect2.in input file) 10. (Signaled by Andreas Birkedal): M3 (gluino mass term) <0 , as required in some models, was not possible previously. Now is corrected. 9. Corrected sqrt(2)-WRONG Yukawa normalization made in some R.C. formulas with respect to expressions from PBMZ (most notable effects in mb SUSY RC, subroutine bmsusyrc) 8. Erroneous use of the running b mass, instead of the b pole mass at some places within the one-loop SUSY R.C to mb (subroutine bmsusyrc). Some other minor modifications (with minor consequences) have been made as well in the algorithm resumming those R.C. (e.g. like replacing the previously used tree-level sbottom,stop masses and mixing angle by R.C. masses within these one-loop mb R.C. expressions. 7. Tr(Y M^2) in RGE: contributions of 1st,2d generation missing irrelevant for mSUGRA, AMSB,.. (where Tr(Y M^2) =0 anyway) but can be relevant for non-universal GUT input parameter scenarios Previous modifs/corrections (25/04/2002): 6. Signaled by Naveen Gaur: the sbottom and/or stop sector masses and mixing are undefined when MU<0 ! -> We found it was a trivial bug in one of the basic function for rad.corr, the one-loop function A(M) defined as A = M*M *(1.D0-2*DLOG(M/scale)) (in order to increase calc. speed) so that it is not defined when M <0 ! correct expression for any M sign is of course: A = M*M * (1.D0-DLOG(M**2/scale**2)) 5. Signaled by Yeong Gyun Kim: The reinitialization of the value of 'msttr1' (l.934) for setting the default EWSB scale in the first iteration on RGE may be needed. It would not be a problem if we investigate just one SUSY input point. But in the mass run, the value of msttr1 of the previous run survives and affects (slightly) the results due to the small nb of iterations and different starting value used. -> Corrected l.934: if(msttr1.ne.0.d0) then -> if(msttr1.ne.0.d0.and.irge.gt.1) then 4. Bug corrected 25/04/02, which was consequence of BUG 3 (26/02/02) below: Since correction 3., DRbar scheme was wrongly used for mb(MZ) and mt(mt) in the Higgs mass calculation (which should be consitently defined in the MSbar scheme): this was wrongly reducing m_h (light Higgs) by ~ 1 GeV. Previous changes: 3. BUG corrected 26/02/2002 in Mb running from mb(mb) to mb(MZ) (thanks to comparisons made by Ben Allanach, Sabine Kraml and Werner Porod): flag for NNLO (two-loop) running added and correct DRbar scheme used for mb(MZ). 2. BUG Corrected 12/2/2002 !(thanks to a remark by Sabine Kraml): for ichoice(3)=1 (i.e. when EXACT gauge coupling unification requested) GUT scale E_GUT was not properly determined (but only severe for large tan beta) 1. CHANGES/ADDITIONS 10/9/2001 1) more flexible choice of Electroweak sym. breaking scale: either default value (Q_EWSB = Sqrt[m_t_L * m_t_R]) or arbitrary (user-supplied) value can be chosen. Controlled by ICHOICE(8) (see input file suspect2.in for more details and example) 2) Changes (simplifications) in the algorithm to calculate mb susy rad. corr. give more stable and reliable results for very large tan beta (tan beta > 50) 3) accordingly some minor additional information in input/output files for convenience.