PDF fitting to the HERA average dataVarious comparisonsAMCS- January 29th 2008 Compare ZEUS-JETS/ inbetween/ H1 parametrizations Compare starting Q2: Q02 = 4/ Q02 = 2 GeV2 Compare massless/ massive variable flavour schemes for heavy quark treatment Compare methods of error treatment in 2: Quadratic/ Hessian/ Offset
ZEUS-Jets Parametrization in between H1 parametrization
First compare different parametrizations:ZEUS-JETS, inbetween and H1 (hopefully Joel has defined these!) in terms of uv, dv, Sea, GlueThis page Q2=4 GeV2 other Q2 values in EXTRASCentral values are really very similar- quite remarkable since ZEUS and H1 parametrizations are not- however the size of errors differs, with in between being the most conservative
ZEUS-Jets Parametrization in between H1 parametrization
Now in terms of U, D, Ubar, DbarThis page Q2=4 GeV2 other Q2 values in EXTRAS
ZEUS-Jets Parametrization in between H1 parametrization
xubarxdbarxcbarxsbarNow in terms of ubar, dbar, sbar, cbarThis page Q2=4 GeV2 other Q2 values in EXTRASThe similarity of these is perhaps even more remarkable given the different treatment of charm- clearly the fixed fraction fc=0.15 is about right compared to dynamical turn on at Q2=mc2
ZEUS-Jets Parametrization in between H1 parametrization
Finally in terms of d/u and dbar-ubarHere we do see a difference in central values. I like the fact that in-between reflects the fact that we dont know anything!This page Q2=4 GeV2 other Q2 values in EXTRAS
Inbetween with Q02=2
Inbetween with Q02=4 Now comparing Q02=4 (standard) with Q02=2=mc2 for the inbetween parametrizationIn the latter case the fixed fraction fc=0.0 rather than fc=0.15. One can thus make it fully consistent with dynamical generation.Starting at a different Q0 is equivalent to a different parametrization. It will be necessary if we are ever to move to NNLO- it makes charm treatment more rational.Central values fairly similar (d-valence?) error estimates smaller for Q02=2
uv,dv,Sea, gluonThis page Q2=4 GeV2
Now U,D,Ubar,Dbar This page Q2=4
Inbetween with Q02=4Inbetween with Q02=2
.Central values similar error estimates smaller for Q02=2
ubar, dbar, sbar, cbar This page Q2=4
Inbetween with Q02=4Inbetween with Q02=2
.Central values similar (high-x dbar?) error estimates smaller for Q02=2
Now d/u and dbar-ubar This page Q2=4.Error estimates smaller for Q02=2
Inbetween with Q02=4Inbetween with Q02=2
Return to the standard Q02=4. Stick with inbetween. Compare dynamical generation of charm zero-mass (ZMVFN) with massive dynamical generation a la Thorne (RTVFN) uv,dv,Sea glueZMVFNRTFVNuv,dv,Sea, gluonThis page Q2=4 GeV2Other Q2 in EXTRASCentral values similar. Errors fairly similar- larger at low-x for RTVFN
ZMVFNRTFVN U,D,Ubar,DbarThis page Q2=4 GeV2
Central values reflect the fact that charm turns on more slowly- hence less U and Ubar at low-x. Errors slightly larger for RTVFN
RTFVN has a slow turn on of charm the other sea quarks try to compensate to make a similar total SeaZMVFNubar,dbar,sbar,cbarubar,dbar,sbar,cbarThis page Q2=4 GeV2 Q2 =10 next pageRTFVN
ubar,dbar,sbar,cbarThis page Q2=10 GeV2ZMVFNRTFVN RTFVN has a slow turn on of charm still hasnt quite caught up at Q2=10.
ZMVFNdbar-ubar and d/u at Q2=4RTFVN No differences worth remarking on
Now compare methods of treating errors: Quadratic/Hessian/Offset (back to standard fit ZMVFN and ZEUS-JETS parametrization)QuadraticHessianOffset procedural errors- rest quadraticuv,dv,Sea,gluonThis page Q2=4 GeV2Other Q2 in EXTRASCentral values very similar (not obvious for Hessian) Errors generally largest for OFFSET procedural (high-x dv?), but not much difference compared to using ZEUS data alone- systematic errors not so big now
Offset just procedural errorsOffset 27 of the 47 total errors (MINUIT limitations!) procedural are not included hereuv,dv,Sea, gluonThis page Q2=4 GeV2Other Q2 in EXTRASProcedural seem to be the most significant
HessianQuadraticOffset proceduralU,D,Ubar,DbarThis page Q2=4 GeV2Other Q2 in EXTRASCentral values very similar (not obvious for Hessian) Errors generally largest for OFFSET procedural (high-x D?), but not much difference compared to using ZEUS data alone- systematic errors not so big now
QuadraticHessianOffset proceduralubar,dbar,sbar,cbarThis page Q2=4 GeV2Other Q2 in EXTRASCentral values very similar (not obvious for Hessian) Errors generally largest for OFFSET procedural, but not much difference compared to using ZEUS data alone- systematic errors not so big now
Conclusions/ to doGo with some sort of inbetween parametrizationMove to Q02=2 and RTVFNUse latest data set!Use either Hessian or OFFSET not quadratic I favour OFFSET but offsetting all 47 is a pain (even when minuit cooperates). Offsetting the 4 procedural seems the most important. Hessian method does seem to reproduce similar size errors and is more straightforward - ideas?Agree with Joel on these alternative fits
EXTRAS
ZEUS-Jets Parametrization in between H1 parametrization
Q2=10 GeV2in terms of uv, dv, Sea, Glue
ZEUS-Jets Parametrization in between H1 parametrization
Q2=1 GeV2in terms of uv, dv, Sea, Glue
ZEUS-Jets Parametrization in between H1 parametrization
Q2=10 GeV2Now in terms of U, D, Ubar, Dbar
ZEUS-Jets Parametrization in between H1 parametrization
xubarxdbarNow in terms of ubar, dbar, sbar, cbarQ2= 10 GeV2xcbarxsbar
ZEUS-Jets Parametrization in between H1 parametrization
xdbarxubarQ2= 1 GeV2xsbarxcbarNow in terms of ubar, dbar, sbar, cbar
ZEUS-Jets Parametrization in between H1 parametrization
xubarxdbarThis page is all for Q2=1.8225= mc2, not for Q2=1xsbarxcbarNow in terms of ubar, dbar, sbar, cbar
Q2= 10 GeV2Finally in terms of d/u and dbar-ubar
uv,dv,Sea, gluonThis page Q2=10GeV2Inbetween with Q02=4Inbetween with Q02=2
Inbetween with Q02=4Inbetween with Q02=2uv,dv,Sea, gluonThis page Q2=1GeV2
ZMVFNRTFVN thompsonuv,dv,Sea, gluonThis page Q2=10 GeV2
ZMVFNRTFVN uv,dv,Sea, gluonThis page Q2=1 GeV2
Offset just procedural errorsOffset 27 of the 47 total errors (MINUIT limitations!) procedural are not included hereU,D,Ubar,DbarThis page Q2=4 GeV2Other Q2 in EXTRASProcedural seem to be the most significant
Offset 27 of the 47 total errors (MINUIT limitations!) procedural are not included hereOffset just procedural errorsProcedural seem to be the most significantUbar,dbar,sbar,cbarThis page Q2=4 GeV2Other Q2 in EXTRAS
QuadraticHessianOffset proceduralOffset 27uv,dv,Sea, gluonThis page Q2=10GeV2
QuadraticHessianOffset 27uv,dv,Sea, gluonThis page Q2=1GeV2
HessianQuadraticOffset proceduralOffset 27U,D,Ubar,DbarThis page Q2=10GeV2
QuadraticHessianOffset proceduralOffset 27ubar,dbar,sbar.cbarThis page Q2=10GeV2
HessianQuadraticOffset 27d/u and dbar-ubarThis page Q2=4 GeV2Other Q2 in EXTRASCentral values very similar (not obvious for Hessian) Errors generally largest for OFFSET procedural, but not much difference compared to using ZEUS data alone- systematic errors not so big now
QuadraticHessianOffset 27d/u and dbar-ubarThis page Q2=4 GeV2Other Q2 in EXTRAS
Top Related