Skip to main content

Full text of "The Real Corrections to the Gamma* Impact Factor: First Numerical Results"

See other formats

Proceedings of the DIS'2004, Strbske Pleso, Slovakia 



Department of Physics & Astronomy, University of Manchester, 
Oxford Road, Manchester M13 9PL, U.K. 
E-mail: man. 

We have performed analytically the transverse momentum integrations in the real 
corrections to the longitudinal 7£ impact factor and carried out numerically the 
remaining integrations. I outline the analytical integration and present the nume- 
rical results: we have performed a numerical test and computed those parts of the 
impact factor that depend upon the energy scale sq. 

1 Introduction 

The 7*7* total cross section is a very attractive observable to be calculated in the 
framework of NLO BFKL. To perform this calculation and also to approach e.g. the 
problem of the color dipole picture at NLO, the 7* impact factor is needed at NLO. 
The NLO corrections of this impact factor are calculated from the photon- Reggeon 
vertices for qq and qqg production, respectively. The virtual corrections involve 
the vertex T^)_^ q q at one-loop level which has been calculated in |1I2| . As to the 

real corrections, the squared vertex |ri£? ^ | 2 is needed at tree level; it has been 
computed in lj (cf. also In 0] we have combined the infrared divergences 
of the virtual and of the real parts, and we have demonstrated their cancellation. 
What remains to complete the calculation of the NLO photon impact factor are 
the integrations over the qq and qqg phase space, respectively. A slightly different 
approach of calculating the NLO photon impact factor has been proposed in [5j . 

We have performed, for the case of longitudinal photon polarisation, the phase 
space integration in the real corrections jHj • After the introduction of Fcynman pa- 
rameters we carried out the integration over transverse momenta analytically. This 
will allow for further theoretical investigations of the photon impact factor. In par- 
ticular, the Mellin transform of the real corrections w.r.t the Reggeon momentum 
can be calculated. This representation (together with an analogous representa- 
tion of the virtual corrections) can be a starting point for the resummation of the 
next-to-leading logs(l/x) in the quark anomalous dimensions. 

The procedure of arriving at finite NLO corrections to the 7* impact factor has 
been described in 0|: |r^_^ ff | 2 has to be integrated over the qqg phase space. 
But there are two restrictions. First, we have to exclude that region of phase 
space where the gluon is separated in rapidity from the qq pair (central region); 
this configuration belongs to the LLA and has to be subtracted. To divide the qqg 
phase space an energy cutoff sq is introduced which plays the role of the energy scale. 

*The work presented in this talk was done in collaboration with J.Bartels, University Hamburg 
© Institute of Experimental Physics SAS, Kosice, Slovakia 1 


A. Kyrieleis 

The virtual corrections to the impact factor are independent of sq. As a result, the 
integration of the real corrections already allows to study the sq dependence of the 
NLO impact factor. Second, we need to take care of the infrared infinities. The 
divergences in |rl < ?_ >g g ff | 2 due to the gluon being either soft or collinear to either of 
the fermions are regularised by subtracting the approximation of the squared vertex 
in the corresponding limit. These expressions are then re-added and integrated in 
4— 2e space-time dimensions, giving rise to poles in e, which drop out in combination 
with the virtual corrections and to finite pieces. The subtraction of the collinear 
limit requires the introduction of a momentum cutoff parameter, A. The final NLO 
corrections must be independent of this auxiliary parameter. In our numerical 
analysis we performed this important test. 

q-k- I q 

Figure 1. The process -y* -f- — > qqg + q' to calculate Y 

2 The integration 

The qqg production vertex is calculated from the process 7* + q' — > qqg + q' at 
tree level. The integration of |r^7_^. g ^ ff | 2 therefore implies the integration of a sum 
of expressions, corresponding to products of Feynman diagrams that differ in their 
denominator structure (for simplicity we will in the following use 'diagram' rather 
than 'product of diagrams'). In order to introduce Feynman parameters we there- 
fore split this sum and treat each diagram (or pairs of them) independently. This 
gives rise to divergences, that in the sum of all diagrams cancel but show up in 
individual diagrams. The main task in the program of performing the integration 
analytically is the regularisation of these additional divergences in each individual 
diagram. We used the subtraction method: From each divergent diagram its ap- 
proximation in the limit of divergence is subtracted and re-added. The integrations 
in the re-added pieces are carried out analytically in 4 — 2e space-time dimensions; 
the poles in e are found to cancel. 

As the result of the regularisation and of the analytical integration over trans- 
verse momenta we have obtained for each diagram (or pairs of them) a convergent 
integral over the Feynman parameters and the momentum fractions of the final 
state quark and gluon. We denote the sum of these integrals from all diagrams 
(including the finite pieces from the regularisation of the additional divergences) 
by $a and <&f according to the separation w.r.t. the color factors Ca,Cf- The 
integrand for each diagram is a sum of terms corresponding to the subtractions 
necessary to regularise all divergences of the diagram (see [B]). 

NLO photon impact factor 


3 Numerical results 

We have carried out numerically the integrations over the Feynman parameters 
and the momentum fractions of the quark and gluon for all diagrams. As the 
result we have obtained values of the real corrections, <j> rea/ j to the photon impact 
factor as a function of A/Q, sq/Q 2 and r 2 /Q 2 , where r is the transverse part of the 
Reggeon momentum and Q 2 is the photon virtuality. In the following we suppress 
the scaling with Q and label the dimensionless variables A, so,r 2 . Besides $>a,®f 
(see above) we include in $ rea ' those finite pieces from the regularisation of the soft 
and collinear divergences that depend upon A or so (A A and A So , respectively): 


peal = e 2 e 2 ^ + <f> F + A A + A So ) , A A = -3 C F -—r In A. 

denotes the LO 7* impact factor, ee/ is the charge of the quark. The full 
dependence of the NLO impact factor on A and so is now included in $ recU . As 
said above, the NLO impact factor and hence $ rea ' has to be independent of A. 
Fig|2 shows that, in fact, the A dependence of &a is very weak. As to the Cf 
terms, <5>f turns out to be proportional to In A. This growth with A, however, is 
fully compensated by Aa. Note that $f is a sum of many Feynman diagrams, 
whereas Aa ~ In A. The compensation of the A dependence, therefore, represents 
a rather stringent test of the calculation of the 7* impact factor. 

Next we address the dependence of the NLO 7* impact factor on the energy 
scale sq. Since, at the moment, we only know the real corrections, we compute, as 


A. Kyrieleis 

a part of the full LO and NLO impact factor: 

$' = g 2 $(°) + g 4 <j> rea/ 

where g 2 — 47ra s . We set e 2 e 2 — 1. For the photon virtuality we choose Q 2 = 15 
GeV 2 which leads a s (Q 2 ) = 0.18 or g 2 = 1.5. Fig compares $' to the LO impact 
factor g 2 $(°) as function of r 2 at different values of sq. The real corrections are 
negative and rather large. More important, <!>' becomes, in absolute terms, more 
significant for smaller values of s Q . Since we included all Sq dependent terms in 
this implies that the 7* impact factor tends to become smaller with decreasing 
so. A physical scattering amplitude (e.g. for the 7*7* scattering process), when 
consistently evaluated in the framework of NLO BFKL has to be invariant under 
changes of so- Since a decrease of so in the energy dependence will enhance 

the scattering amplitude, the combined sq dependence of the impact factors and 
the BFKL Green's function has to compensate this growth. Our result for the so 
behaviour of the 7* impact factor is therefore, at least, consistent with the general 

~ 500 




123456789 10 



Figure 3. <!>' at different different values of so 


1. J. Bartels, S. Gieseke and C. F. Qiao, Phys. Rev. D63 (2001) 056014 [Erratum- 
ibid. D65 (2001) 079902]. 

2. V. S. Fadin, D. Y. Ivanov and M. I. Kotsky, Phys. Atom. Nucl. 65 (2002) 1513 
[Yad. Fiz. 65 (2002) 1551]. 

3. J. Bartels, S. Gieseke and A. Kyrieleis, Phys. Rev. D65 (2002) 014006. 

4. J. Bartels, D. Colferai, S. Gieseke and A. Kyrieleis, Phys. Rev. D66 (2002) 

5. V. S. Fadin, D. Y. Ivanov a nd M. I. Kotsky, Nucl. Phys. B658 (2003) 156. 

6. J. Bartels and A. Kyrieleis, arXiv:hep-ph/0407051 

g 2 =1.5 

= ---- s„= 10 

- s = l 

= so = 0-1 

-1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1