Available from http://projects.hepforge.org/mstwpdf/
Updated NNLO parton set described in arXiv:0706.0459 [17] obtained using the new variable flavour number scheme described in hep-ph/0601245. As well as the best-fit set of partons, we also provide 30 additional sets representing the parton uncertainties obtained using the Hessian approach described in hep-ph/0211080. The interface code to the grids is provided in both Fortran and C++ versions. The interpolation code provided in previous MRST releases has been modified to account for the discontinuities of the NNLO partons at the heavy flavour thresholds. The grids have been extended to lower x = 10-6 and higher Q2 = 109 GeV2 than previous MRST releases, with an extrapolation returned outside these limits.
The complete package mrst2006nnlo.tar.gz contains the following files: a README file, the Fortran interface code (pdf.f), the C++ interface code (pdf.h, pdf.C), a Fortran example program (example.f), a C++ example program (example.C), a Makefile, the best-fit grid file (Grids/mrst2006nnlo.00.dat), and 30 additional grid files for the eigenvector sets (Grids/mrst2006nnlo.01.dat, Grids/mrst2006nnlo.02.dat, etc.).
New three- and four-light-flavour sets of partons generated
using the standard (variable flavour) MRST2004 partons as input, as described
in hep-ph/0603143
[16]. NLO and LO (in pQCD) versions are provided, the latter being based on the
earlier set of MRST2001 LO partons (see below). In practical applications of
these parton sets, care must be taken to use the correct (fixed-flavour)
definition of the coupling alpha_s.
Requires the grid files mrst2004f3lo.dat
, mrst2004f3nlo.dat
, mrst2004f4lo.dat
and mrst2004f4nlo.dat
.
(
mrst2004f.uu tarred, gzipped and uuencoded version of the complete
package)
New MRST (NLO in pQCD) parton distributions, described in hep-ph/0411040 [15], that incorporate order(alpha) QED corrections in the DGLAP evolution equations. Although the quality of the fit is essentially unchanged, there are two important physical consequences. First, the different DGLAP evolution of u and d type quarks introduces isospin violation, i.e. $u^p \neq d^n$. A second consequence is the appearance of photon parton distributions $\gamma(x,Q^2)$ of the proton and the neutron. There are 2 pdf sets provided: one set for the proton distributions (mode = 1) and the other for the neutron distributions (mode = 2).
Note (January 2007):
The neutron parton
distributions have been corrected for a bug affecting the $\bar u$ and $\bar d$
distributions at high x.
Requires the grid files qed6-10gridp.dat
and qed6-10gridn.dat
.
(
mrst2004qed.uu tarred, gzipped and uuencoded version of the complete
package)
New MRST NLO and NNLO parton distributions, described in hep-ph/0410230 [14],
in which a more physical parametrisation of the gluon distribution is used. In
the new parametrisation, the gluon distribution at large x in the MSbar scheme
is driven by the valence quarks, which naturally produces a shoulder-like form
at large x, and hence produces a better description of the Tevatron inclusive
jet data. Some new Drell-Yan and DIS data are also included in the global fit.
Requires the grid files mrst2004nlo.dat
and mrst2004nnlo.dat
.
( mrst2004.uu
tarred, gzipped and uuencoded version of the complete package)
Variants of the MRST2002 NLO and NNLO parton distributions,
described in hep-ph/0308087
[13], in which the range of DIS structure function data fitted is restricted to
x > 0.005, W^2 > 15 GeV^2 and Q^2 > 10 (7) GeV^2 for the NLO (NNLO)
global fit. In this restricted domain, the parton distributions obtained from
the fits are stable with respect to further variations of these cuts. We can
therefore be confident that in this region, fixed-order leading-twist DGLAP
evolution is certainly valid, and hence we describe the resulting partons as
'conservative'.
Requires the grid files mrst2003cnlo.dat
and mrst2003cnnlo.dat
.
(
mrst2003c.uu tarred, gzipped and uuencoded version of the complete
package).
The new (2002) MRST2002 updated
parton distributions, as described in hep-ph/0211080 [12].
There are only minor changes from the 2001 MRST sets: a slightly different
gluon parameterisation is used, there is a modification of the treatment of the
Drell-Yan data errors, and new ZEUS high Q^2 data is included. The only
signficant change in the MRST 2002 partons, compared to those of MRST 2001, is
an increase in the gluon at high x, see Fig. 14 of [12]. The LO fit would give
essentially identical partons to the MRST2001 LO set, and so is not repeated.
Requires the grid files mrst2002nlo.dat
and mrst2002nnlo.dat
.
(
mrst2002.uu tarred, gzipped and uuencoded version of the complete
package).
*****************************************************************
*** NOTE: This must be used with mrst02f2.f to
calculate electromagnetic
*** and mrst02f2cc.f
to calculate charged current DIS structure functions!
*****************************************************************
The new (2002) MRST2001
"parton distributions with errors" package, described in hep-ph/0211080 [12],
which allows estimates of the pdf uncertainties for given physical quantities
according to the Hessian approach. There are 30 pdf "extremum" sets
("+" and "-" sets for each of the 15 eigenvectors in
parameter space) corresponding to mode numbers n = 1, ..,30 and a central
"best fit" set given by mode n = 0. The best fit set is very close to
the previous MRST2001 NLO set (see below).
The format is similar to the previous MRST series, except that the 31 grid
files have been concatenated in one large grid file mrst2001E_hessian.dat
(4.5Mb). The improved grid interpolation procedure based on bicubic
interpolation is used. As always, x times the parton distribution is returned,
q is the scale in GeV, MSbar factorization is assumed, and Lambda(MSbar,nf=4) =
323 MeV (corresponding to alpha_S(M_Z) = 0.119) for all sets.
Requires the grid file mrst2001E_hessian.dat
.
The sample FORTRAN programme ggh.f illustrates
the use of the package to compute a prediction and +- "pdf error" on
the leading-order Higgs production by gluon-gluon fusion cross section at
hadron colliders. It also contains the MRST alpha_s function subroutine which
must be used with the pdfs when calculating cross sections and structure
functions.
The new (2001) MRST2001 package described in hep-ph/0110215 [10]. The format is similar to the previous (1999) MRST series. A new feature is an improved grid interpolation procedure based on bicubic interpolation which delivers much smoother output without any significant time overhead (thanks to Jeppe Andersen for help with the design and implementation of this). There are four new (NLO) sets, the (default) overall best fit, two fits which span the allowed range in alpha_s, and a fit which gives a better fit to the Tevatron jet data at the expense of introducing "unphysical" structure in the input gluon distribution at large x. As before, x times the parton distribution is returned, q is the scale in GeV, MSbar factorization is assumed, and Lambda(MSbar,nf=4) is given below for each set.
mode set comment L(4)/MeV a_s(M_Z) grid#1
1 alf119 central gluon, a_s 323 0.119 0.00927
2 alf117 lower a_s 290 0.117 0.00953
3 alf121 higher a_s 362 0.121 0.00889
4 j121 better fit to jet data 353 0.121 0.00826
Requires the grid files alf119.dat ,
alf117.dat ,
alf121.dat ,
j121.dat .
(
mrst2001.uu tarred, gzipped and uuencoded version of the complete
package).
*****************************************************************
*** NOTE: This must be used with mrst01f2.f to
calculate electromagnetic
*** and mrst01f2cc.f
to calculate charged current DIS structure functions!
*****************************************************************
The new (2001) MRSTNNLO package described in hep-ph/0201127 [11]. The format is precisely the same as for MRST2001 (the NLO distributions). There are four new (NNLO) sets, the (default) overall best fit, two fits which span the range of NNLO splitting functions, and a fit which gives a better fit to the Tevatron jet data at the expense of introducing "unphysical" structure in the input gluon distribution at large x. As before, x times the parton distribution is returned, q is the scale in GeV, MSbar factorization is assumed, and Lambda(MSbar,nf=4) is given below for each set.
mode set comment L(4)/MeV a_s(M_Z) grid#1
1 vnvalf1155 `average' evolution 235 0.1155 0.00725
2 vnvalf1155a `fast' evolution 235 0.1155 0.00734
3 vnvalf1155b 'slow' evolution 235 0.1155 0.00739
4 vnvalf1180j better fit to jet data 267 0.118 0.00865
Requires the grid files vnvalf1155.dat
, vnvalf115a.dat
, vnvalf155b.dat
, vnvalf1180j.dat
.
(
mrstnnlo.uu tarred, gzipped and uuencoded version of the complete
package).
The new (2001) MRST2001LO package described in hep-ph/0201127 [11]. The format is the same as the NLO MRST2001. As before, x times the parton distribution is returned, q is the scale in GeV, and Lambda(LO.nf=4) is given.
mode set L(4)/MeV a_s(M_Z) grid#1
1 lo2002 220 0.130 0.02868
Requires the grid file lo2002.dat
(
mrst2001lo.uu tarred, gzipped and uuencoded version of the complete
package).
The new (1999) MRST99 package described in hep-ph/9907231 [9]. The format is similar to the previous (1998) MRST series. The first five sets are the corrected versions of the corresponding five MRST sets (there was a small bug in the evolution program), and the remaining sets are additional sets used to explore uncertainties in the predicted W and Z cross sections at the Tevatron and LHC colliders. As before, x times the parton distribution is returned, q is the scale in GeV, MSbar factorization is assumed, and Lambda(MSbar,nf=4) is given below for each set.
mode set comment L(4)/MeV a_s(M_Z) grid#1
1 cor01 central gluon, a_s 300 0.1175 0.00537
2 cor02 higher gluon 300 0.1175 0.00497
3 cor03 lower gluon 300 0.1175 0.00398
4 cor04 lower a_s 229 0.1125 0.00585
5 cor05 higher a_s 383 0.1225 0.00384
6 cor06 quarks up 303.3 0.1178 0.00497
7 cor07 quarks down 290.3 0.1171 0.00593
8 cor08 strange up 300 0.1175 0.00524
9 cor09 strange down 300 0.1175 0.00524
10 cor10 charm up 300 0.1175 0.00525
11 cor11 charm down 300 0.1175 0.00524
12 cor12 larger d/u 300 0.1175 0.00515
Requires the grid files cor01.dat ,
cor02.dat ,
cor03.dat ,
cor04.dat ,
cor05.dat ,
cor06.dat ,
cor07.dat ,
cor08.dat ,
cor09.dat ,
cor10.dat ,
cor11.dat ,
cor12.dat .
( mrs99.uu tarred,
gzipped and uuencoded version of the complete package (726kb)).
*****************************************************************
*** NOTE: This must be used with mrsf299.f to
calculate electromagnetic, or
*** with the new mrstcc00.f
(which replaces mrscc99.f) to calculate charged current,
*** DIS structure functions!
*** (contact r.g.roberts@rl.ac.uk if
you need the earlier charged current programme,)
*****************************************************************
The DIS-scheme version of the new MRST 1999 package of [9]. The format is similar to the MSbar-scheme versions:
mode set comment L(4)/MeV a_s(M_Z) grid#1
1 COR01DIS central gluon, a_s 300 0.1175 0.00643
2 COR02DIS higher gluon 300 0.1175 0.00619
3 COR03DIS lower gluon 300 0.1175 0.00515
4 COR04DIS lower a_s 229 0.1125 0.00697
5 COR05DIS higher a_s 383 0.1225 0.00510
6 COR06DIS quarks up 303.3 0.1178 0.00615
7 COR07DIS quarks down 290.3 0.1171 0.00717
8 COR08DIS strange up 300 0.1175 0.00643
9 COR09DIS strange down 300 0.1175 0.00643
10 C0R10DIS charm up 300 0.1175 0.00644
11 COR11DIS charm down 300 0.1175 0.00643
Requires the grid files cor01dis.dat ,
cor02dis.dat
, cor03dis.dat
, cor04dis.dat
, cor05dis.dat
,
cor06dis.dat
, cor07dis.dat
, cor08dis.dat
, cor09dis.dat
, cor10dis.dat
, cor11dis.dat
.
( mrs99dis.uu
tarred, gzipped and uuencoded version of the complete package (654kb)).
*****************************************************************
*** NOTE: This must be used with mrsf299dis.f to
calculate DIS structure functions!
*****************************************************************
The new MRST 1998 package of [8]. The format is similar to the previous (1996) MRS-R series. As before, x times the parton distribution is returned, q is the scale in GeV, MSbar factorization is assumed, and Lambda(MSbar,nf=4) is given below for each set.
mode set comment L(4)/MeV a_s(M_Z) grid#1