NNLO HERAPDF1.0? H1/ZEUS fitters meeting Jan 26 th 2010 AMC-S
description
Transcript of NNLO HERAPDF1.0? H1/ZEUS fitters meeting Jan 26 th 2010 AMC-S
NNLO HERAPDF1.0?H1/ZEUS fitters meeting Jan 26th 2010
AMC-S
HERAPDF1.0 at NNLOWe have already agreed results for the central values at two αS(MZ) values: 0.1145 which is preferred by the fit and the standard 0.1176
NOTE that NNLO PDFs are supposed to look different from NLO: gluon evolution is slower whereas sea evolution is faster.
This is illustrated by showing Q2=2 and Q2=10 so one can see that the larger low-x gluon at Q2=2 (for alphas=0.1145) DOES NOT evolve into a much larger gluon at Q2=10 – But the Sea PDF does evolve more.
1. In the PDF4LHC community there is now an underlying drive to move to NNLO
2. there is a large uncertainty in the heavy quark treatment at NLO
3. -These discrepancies are smaller at NNLO and it is agreed that the community should move in this direction
4. Theoreticians are making predictions at NNLO and will only use NNLO PDFs (e.g. Catani et al arxiv: 1002.3115) this means that they are using MSTW08 and ABKM and GJR rather than CTEQ66/NNPDF- it is clear that having an NNLO set can give greater visibility
So I have run the NNLO fit for αS(MZ) =0.1145 for all the model and parametrisation variations
NOTE:
Model variations much as before BUT Q2min > 5 a bit more significant
Parametrisation variations: i)Q20=1.5 + negative gluon term is more significant
ii) Bdv≠ Buv more significant
iii) DUbar ≠ 0, DDbar ≠ 0 become insignificant
iv) Duv ≠ 0 is less significant
NNLO HERAPDF1.0
This is due to a stronger negative gluon term
The shape variation of d_v is mostly down to Bdv ≠ Buv
NNLO NLO
The greater width of the yellow bands is due to Q2min=5 cut
The shape variation of d_v is mostly down to Bdv≠Buv
This is due to a stronger negative gluon term
Sensitivity to the parametrisation
Follow the usual procedure – return to 9 PDF parameters
10th param chisq a
+none 645.9
+Euv 623.7
+Duv 637.6
+Dg 644.8
+Bdv 640.7
Our usual 10th parameter Euv is the best choice
I have retrieved a few of the 9 parameters + one at a time jobs
These are the ones that have ever made any difference—
the DUbar,DDbar,DDv have never been important at this atge
variation a_s=0.1145 a_s=0.1176
Standard Q2>3.5
623.7/582=1.07 638.3/582=1.10
Q2>2.5 655.4/598=1.10
Q2> 5
Q2> 7
535.7/556=0.96
491.1/540=0.91
No big shape change 5 to 7
10+Bdv 619.8
10+Duv 620.4
10+DUbar 623.7
10+DDbar 623.7
10+Dg 623.7
NLO has 574/582=0.98
NLO has 511/556=0.92
and no big shape change 3.5 to 5
And starting with our usual parametrisation I have retrieved some of the important variations
NNLO cross-checks to Voica
Alpha_s=0.1145 chisq=623.7 Voica 621.9
Buv 0.803493854 0.028409602 Cuv 4.66702624 0.158989171 Euv 6.8648385 1.52593798Cdv 3.78576767 0.246948783ADbar 0.23571 0.009246213 BDbar -0.132146295 0.00498902822 CUbar 7.20525693 0.708651104 CDbar 4.64765555 0.970524693 Bg 0.102284448 0.0222984419 Cg 8.33672271 0.601400488
Buv 0.80434 0.29045E-01 Cuv 4.6748 0.15963 Euv 6.8983 1.5455 Cdv 3.7873 0.24843 ADbar 0.23742 0.94999E-02 BDbar -0.13109 0.50767E-02 CUbar 7.2478 0.72054 CDbar 4.6805 0.99849 Bg 0.98402E-01 0.22316E- Cg 8.2939 0.60679
Alphas=0.1176 chisq=638.3 Voica 636.3
Buv 0.860567384 0.0234952116 Cuv 4.59396084 0.168363544 Euv 5.778106 1.36545351Cdv 3.8125908 0.230886994ADbar 0.25686 0.009623 BDbar -0.121103638 0.00482499241 CUbar 9.41256469 0.718575901 CDbar 4.89010829 1.02883157 Bg 0.142760628 0.0252330946Cg 7.01748255 0.594072802
Buv 0.86171 0.23412E-01 Cuv 4.5953 0.16458 Euv 5.7522 1.3747 Cdv 3.8235 0.21896 ADbar 0.26120 0.88901E-02 BDbar -0.11967 0.44277E-02 CUbar 9.4858 0.70083 CDbar 4.8932 0.83099 Bg 0.13663 0.24780E-01 Cg 6.9707 0.52585
Variations for alpha_s=0.1145
Q2>5 Chisq=535.7 Voica 534
Buv 0.737846811 0.0376851683Cuv 4.70101564 0.144694628Euv 8.98782954 1.84885886Cdv 3.71919318 0.304912115ADbar 0.2302353 0.0144872BDbar -0.127821207 0.00894921063CUbar 5.77891039 0.88565066CDbar 3.80712724 0.883081754Bg 0.0694983119 0.028270994Cg 8.4448627 0.671635174
Buv 0.73810 0.38608E-01 Cuv 4.7085 0.14372 Euv 9.0454 1.8595 Cdv 3.7195 0.31609 ADbar 0.23135 0.14651E-01 BDbar -0.12717 0.89702E-02 CUbar 5.8010 0.90184 CDbar 3.8222 0.93516 Bg 0.66989E-01 0.28187E-01 Cg 8.4219 0.68706
Bdv.ne.Buv Chisq=619.8 Voica 617.9
0.803624474 0.02621938754.62203842 0.1555526586.81104917 1.51427551
1.0639416 0.07196704114.7427327 0.3755726080.2431588 0.007255562
-0.128492536 0.003931686477.22572511 0.6835760397.72553066 1.05739964
0.0883410173 0.02019807787.50122243 0.411532943
Buv 0.80488 0.26924E-01 Cuv 4.6255 0.15540 Euv 6.8059 1.5238 Bdv 1.0749 0.13217 Cdv 4.7848 0.56859 Adbar 0.24552 0.96894E-02 Bdbar -0.12712 0.50777E-02 CUbar 7.2628 0.69045 CDbar 8.0111 2.7948
Bg 0.83155E-01 0.23503E-01 Cg 7.4040 0.68866
Voica agrees on the effect of Bdv.ne.Buv
Voica agrees on the effect of the Q2cut > 5
Voica agrees on the effect of the
negative gluon term
Voica agrees on the effect of change in alphas
Voica has found a relatively strong difference
with ACOT
So do we release it?
With the same settings as HERAPDf1.0? IF SO…
There is an argument for going back to alphas=0.1176
(The variations I have shown are for alphas=0.1145)
extras