Introduction
In this paper we are going to discuss the process of double diffractive production. In Figure 1 one can see the first Pomeron diagram for this process, in which we put all kinematic notations. We consider the scattering of two mesons,
Figure 1. The first Pomeron diagrams for double and single diffractive production. The wavy lines denote the BFKL Pomeron exchanges. The helix line denotes gluons while the arrow lines stand for heavy quarks. The produced masses are described by ln M 1 2 = Y − Y 01 ≡ Y m 1 \ln M_1^2 = Y - Y_{01} \equiv Y_{m1} ln M 1 2 = Y − Y 01 ≡ Y m 1 and ln M 2 2 = Y 02 ≡ Y m 2 \ln M_2^2 = Y_{02} \equiv Y_{m2} ln M 2 2 = Y 02 ≡ Y m 2 . Y 12 = Y 01 − Y 02 Y_{12} = Y_{01} - Y_{02} Y 12 = Y 01 − Y 02 is the value of the rapidity gap in the process of double diffraction. For single diffraction, Y m ≡ Y m 1 = ln M 1 2 = Y − Y 01 Y_m \equiv Y_{m1} = \ln M_1^2 = Y - Y_{01} Y m ≡ Y m 1 = ln M 1 2 = Y − Y 01 and Y 01 Y_{01} Y 01 denote the produced mass and the rapidity gap, respectively. The two cascades in Eq. (II.7) interact at rapidity Y = Y 0 Y = Y_0 Y = Y 0 (see Figure 2 and Figure 3 ).
each consisting of two heavy quarks, so that the perturbative QCD approach is applicable. We are interested in the production of two systems of hadrons with large masses M 1 M_1 M 1 and M 2 M_2 M 2 . In this case double diffractive production can be described by the exchange of BFKL Pomerons [1 ] .1 as shown in the second diagram of Figure 1 .
The equations for single diffraction have been derived in Refs. [2 , 3 ] , building on the dipole approach to high energy QCD [4 ] , which reproduces the nonlinear evolution equation for the scattering of a dilute parton system off a dense one [5 , 6 , 7 , 8 , 9 , 10 ] . However, despite considerable progress in the theoretical understanding of high energy QCD [1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 , 44 , 45 , 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 , 54 ] , the evolution equations for dilute-dilute and dense-dense parton scattering remain unknown. The central unsolved theoretical problem is the summation of the Pomeron loops, of which Figure 1 shows one example; we therefore need a way to handle these loops.
We sum the large Pomeron loops using t t t -channel unitarity, recast in a form convenient for the dipole approach to the CGC in Refs. [3 , 4 , 14 , 16 , 18 , 19 , 27 ] (see Figure 2 ). The analytic expression for the imaginary part of the scattering amplitude then takes the form [16 , 18 , 50 ] :
Figure 2. Summing large Pomeron loops. The wavy lines denote the BFKL Pomeron exchanges. The circles denote the amplitude γ \gamma γ .
N ( Y , r , r ′ ; b ) = ∑ n = 1 ∞ ( − 1 ) n + 1 n ! ∫ ∏ i = 1 n d 2 r i d 2 r i ′ d 2 b i ′ ∫ d 2 b i γ B A ( r i , r i ′ , b i − b i ′ ≡ δ b i ) ρ n ( Y − Y 0 , r , b ; { r i , b i } ) ρ n ( Y 0 , r ′ , b ′ ; { r i ′ , b i ′ } ) (1) N(Y,\mathbf{r},\mathbf{r}';b) = \sum_{n=1}^{\infty}(-1)^{n+1}n!\int\prod_{i=1}^{n}d^{2}r_i\,d^{2}r'_i\,d^{2}b'_i\int d^{2}b_i\,\gamma^{BA}(r_i,r'_i,b_i-b'_i\equiv\delta b_i)\rho_n(Y-Y_0,r,b;\{r_i,b_i\})\rho_n(Y_0,r',b';\{r'_i,b'_i\}) \tag*{(1)} N ( Y , r , r ′ ; b ) = n = 1 ∑ ∞ ( − 1 ) n + 1 n ! ∫ i = 1 ∏ n d 2 r i d 2 r i ′ d 2 b i ′ ∫ d 2 b i γ B A ( r i , r i ′ , b i − b i ′ ≡ δ b i ) ρ n ( Y − Y 0 , r , b ; { r i , b i }) ρ n ( Y 0 , r ′ , b ′ ; { r i ′ , b i ′ }) (1)
γ B A \gamma^{BA} γ B A is the scattering amplitude of two dipoles in the Born approximation of perturbative QCD. The dipole densities ρ n ( Y , r , b ; { r i , b i } ) \rho_n(Y,r,b;\{r_i,b_i\}) ρ n ( Y , r , b ; { r i , b i }) have been introduced in Ref. [16 ] as follows:
ρ n ( Y − Y 0 , r , b ; { r i , b i } ) = 1 n ! ∏ i = 1 n δ δ u i Z ( Y − Y 0 , r , b ; [ u i ] ) ∣ u i = 1 (2) \rho_n(Y-Y_0,r,b;\{r_i,b_i\})=\left.\frac{1}{n!}\prod_{i=1}^{n}\frac{\delta}{\delta u_i}Z(Y-Y_0,r,b;[u_i])\right|_{u_i=1} \tag*{(2)} ρ n ( Y − Y 0 , r , b ; { r i , b i }) = n ! 1 i = 1 ∏ n δ u i δ Z ( Y − Y 0 , r , b ; [ u i ]) u i = 1 (2)
where the generating functional Z Z Z is
Z ( Y , r , b ; [ u i ] ) = ∑ n = 1 ∞ ∫ P n ( Y , r , b ; { r i , b i } ) ∏ i = 1 n u ( r i , b i ) d 2 r i d 2 b i (3) Z(Y,r,b;[u_i])=\sum_{n=1}^{\infty}\int P_n(Y,r,b;\{r_i,b_i\})\prod_{i=1}^{n}u(r_i,b_i)\,d^{2}r_i\,d^{2}b_i \tag*{(3)} Z ( Y , r , b ; [ u i ]) = n = 1 ∑ ∞ ∫ P n ( Y , r , b ; { r i , b i }) i = 1 ∏ n u ( r i , b i ) d 2 r i d 2 b i (3)
where u ( r i , b i ) ≡ u i u(r_i,b_i)\equiv u_i u ( r i , b i ) ≡ u i is an arbitrary function and P n P_n P n is the probability of finding n n n dipoles with the given kinematics. The initial and boundary conditions for the BFKL cascade which stems from one dipole have the following form for the functional Z Z Z :
Z ( Y = 0 , r , b ; [ u i ] ) = u ( r , b ) ; Z ( Y , r , b ; [ u i = 1 ] ) = 1 ; (4) Z\left(Y=0,\mathbf{r},\mathbf{b};[u_i]\right)=u(\mathbf{r},\mathbf{b});\qquad Z\left(Y,\mathbf{r},\mathbf{b};[u_i=1]\right)=1; \tag*{(4)} Z ( Y = 0 , r , b ; [ u i ] ) = u ( r , b ) ; Z ( Y , r , b ; [ u i = 1 ] ) = 1 ; (4)
ρ 1 ( Y = 0 , r , b ; r 1 , b 1 ) = δ ( 2 ) ( r − r 1 ) δ ( 2 ) ( b − b 1 ) ; ρ n ( Y = 0 , r , b ; { r i , b i } ) = 0 at n ≥ 2 ; (5) \rho_1\left(Y=0,\mathbf{r},\mathbf{b};\mathbf{r}_1,\mathbf{b}_1\right)=\delta^{(2)}\left(\mathbf{r}-\mathbf{r}_1\right)\delta^{(2)}\left(\mathbf{b}-\mathbf{b}_1\right);\qquad\rho_n\left(Y=0,\mathbf{r},\mathbf{b};\{\mathbf{r}_i,\mathbf{b}_i\}\right)=0\text{ at }n\geq2; \tag*{(5)} ρ 1 ( Y = 0 , r , b ; r 1 , b 1 ) = δ ( 2 ) ( r − r 1 ) δ ( 2 ) ( b − b 1 ) ; ρ n ( Y = 0 , r , b ; { r i , b i } ) = 0 at n ≥ 2 ; (5)
In Eq. (I.1) b i = b − b i ′ \mathbf{b}_i=\mathbf{b}-\mathbf{b}'_i b i = b − b i ′ .
In Ref. [54 ] we found ρ n ( Y − Y 0 , r , b ; { r i , b i } ) \rho_n\left(Y-Y_0,\mathbf{r},\mathbf{b};\{\mathbf{r}_i,\mathbf{b}_i\}\right) ρ n ( Y − Y 0 , r , b ; { r i , b i } ) , which in the saturation region take the form:
ρ n ( Y − Y 0 , r , b ; { r i , b i } ) = C n ∏ i = 1 n G I P ( z i ) with z i = κ α ˉ S ( Y − Y 0 ) + ξ r i , r (6) \rho_n\left(Y-Y_0,\mathbf{r},\mathbf{b};\{\mathbf{r}_i,\mathbf{b}_i\}\right)=C_n\prod_{i=1}^{n}G_{\mathrm{IP}}(z_i)\quad\text{with}\quad z_i=\kappa\bar{\alpha}_S\left(Y-Y_0\right)+\xi_{r_i,r} \tag*{(6)} ρ n ( Y − Y 0 , r , b ; { r i , b i } ) = C n i = 1 ∏ n G IP ( z i ) with z i = κ α ˉ S ( Y − Y 0 ) + ξ r i , r (6)
where r r r is the size of the incoming dipole. The scaling variable z i z_i z i is equal to z i = ln ( r i 2 Q s 2 ( Y , b ) ) z_i=\ln\left(r_i^2Q_s^2(Y,b)\right) z i = ln ( r i 2 Q s 2 ( Y , b ) ) , where Q s Q_s Q s is a new dimensional scale: the saturation momentum [5 ] which has the following Y Y Y dependence [5 , 28 , 29 ] $:
Q s 2 ( Y , b ) = Q s 2 ( Y = 0 , b ) e α ˉ S κ Y − 3 2 γ ˉ ln ( α ˉ S Y ) (7) Q_s^2(Y,b)=Q_s^2(Y=0,b)e^{\bar{\alpha}_S\kappa Y-\frac{3}{2\bar{\gamma}}\ln(\bar{\alpha}_S Y)} \tag*{(7)} Q s 2 ( Y , b ) = Q s 2 ( Y = 0 , b ) e α ˉ S κY − 2 γ ˉ 3 l n ( α ˉ S Y ) (7)
Eq. (I.5) shows geometric scaling behaviour [30 , 69 ] as a function of the single variable z i z_i z i . The Green’s function of the BFKL Pomeron in this region is known [28 ] :
G I P ( z i ) = e γ ˉ z i (8) G_{\mathrm{IP}}(z_i)=e^{\bar{\gamma}z_i} \tag*{(8)} G IP ( z i ) = e γ ˉ z i (8)
The value of κ \kappa κ and γ ˉ \bar{\gamma} γ ˉ are determined by the BFKL kernel (see for example Ref. [3 ] ). However, in this paper we will not use Eq. (I.5); instead we will find ρ n \rho_n ρ n in Eq. (I.1) using the equation for single diffraction and the simple expression for the densities ρ n A \rho_n^A ρ n A for a nucleus with radius r ′ = R A = A 1 / 3 R N 2 r'=R_A=A^{1/3}R_N^2 r ′ = R A = A 1/3 R N 2 , assuming that the nucleus is a bag of dipoles with size r ′ r' r ′ that do not interact with each other at Y 0 = 0 Y_0=0 Y 0 = 0 . It has the form
ρ n A ( Y 0 = 0 , r ′ , b ′ ; { r i ′ , b i ′ } ) = S A n ( b ′ ) n ! ∏ i = 1 n δ ( 2 ) ( r ′ − r i ′ ) δ ( 2 ) ( b ′ − b i ) ; (9) \rho_n^A\left(Y_0=0,r',b';\{r'_i,b'_i\}\right)=\frac{S_A^n(b')}{n!}\prod_{i=1}^{n}\delta^{(2)}\left(r'-r'_i\right)\delta^{(2)}\left(b'-b_i\right); \tag*{(9)} ρ n A ( Y 0 = 0 , r ′ , b ′ ; { r i ′ , b i ′ } ) = n ! S A n ( b ′ ) i = 1 ∏ n δ ( 2 ) ( r ′ − r i ′ ) δ ( 2 ) ( b ′ − b i ) ; (9)
The factor S A n ( b ′ ) S_A^n(b') S A n ( b ′ ) gives the probability of finding n n n nucleons (dipoles) in a nucleus at impact parameter b ′ b' b ′ . Finally, G I P ( z i ) G_{\mathrm{IP}}(z_i) G IP ( z i ) is the solution of the BFKL equation (the Green’s function of the BFKL Pomeron) in the saturation region, where it exhibits geometric scaling behaviour [30 , 69 ] . The BFKL equation takes the following general form:
∂ G I P ( Y , x 01 , r i , b ) ∂ Y = α ˉ S 2 π ∫ d 2 x 2 x 01 2 x 02 2 x 12 2 × [ G I P ( Y , x 12 , r i , b − x 02 2 ) + G I P ( Y , x 02 , r i , b − x 12 2 ) − G I P ( Y , x 01 , r i , b ) ] (10) \begin{aligned}
\frac{\partial G_{\mathrm{IP}}(Y,x_{01},r_i,b)}{\partial Y}
&=\frac{\bar{\alpha}_S}{2\pi}\int d^2x_2\frac{x_{01}^2}{x_{02}^2x_{12}^2}\\
&\quad\times\left[G_{\mathrm{IP}}\left(Y,x_{12},r_i,b-\frac{x_{02}}{2}\right)+G_{\mathrm{IP}}\left(Y,x_{02},r_i,b-\frac{x_{12}}{2}\right)-G_{\mathrm{IP}}(Y,x_{01},r_i,b)\right] \tag*{(10)}
\end{aligned} ∂ Y ∂ G IP ( Y , x 01 , r i , b ) = 2 π α ˉ S ∫ d 2 x 2 x 02 2 x 12 2 x 01 2 × [ G IP ( Y , x 12 , r i , b − 2 x 02 ) + G IP ( Y , x 02 , r i , b − 2 x 12 ) − G IP ( Y , x 01 , r i , b ) ] (10)
with the initial condition which corresponds to the exchange of two gluons in t t t -channel (see Appendix A for its solution).
Summing large Pomeron loops: diffraction in the dipole-dipole scattering
The equation for the single diffraction in dipole-nucleus collisions has the following form in QCD [2 , 3 ] :
∂ S D ( x 0 , x 1 , Y , Y 01 ) ∂ Y = α ˉ s 2 π ∫ d 2 x 2 x 01 2 x 02 2 x 12 2 { S D ( x 1 , x 2 , Y , Y 01 ) S D ( x 0 , x 2 , Y , Y 01 ) − S D ( x 0 , x 1 , Y , Y 01 ) } (11) \frac{\partial S^D(x_0,x_1,Y,Y_{01})}{\partial Y} = \frac{\bar{\alpha}_s}{2\pi} \int d^2x_2 \frac{x_{01}^2}{x_{02}^2x_{12}^2} \left\{ S^D(x_1,x_2,Y,Y_{01})S^D(x_0,x_2,Y,Y_{01}) - S^D(x_0,x_1,Y,Y_{01}) \right\} \tag*{(11)} ∂ Y ∂ S D ( x 0 , x 1 , Y , Y 01 ) = 2 π α ˉ s ∫ d 2 x 2 x 02 2 x 12 2 x 01 2 { S D ( x 1 , x 2 , Y , Y 01 ) S D ( x 0 , x 2 , Y , Y 01 ) − S D ( x 0 , x 1 , Y , Y 01 ) } (11)
where
S D ( x 0 , x 1 , Y , Y 01 ) = 1 − 2 N ( x 0 , x 1 , Y ) + N D ( x 0 , x 1 , Y , Y 01 ) (12) S^D(x_0,x_1,Y,Y_{01}) = 1 - 2N(x_0,x_1,Y) + N^D(x_0,x_1,Y,Y_{01}) \tag*{(12)} S D ( x 0 , x 1 , Y , Y 01 ) = 1 − 2 N ( x 0 , x 1 , Y ) + N D ( x 0 , x 1 , Y , Y 01 ) (12)
where N N N is the amplitude for the elastic scattering and N D N^D N D is the single diffraction cross section with the rapidity gap ≥ Y 0 \ge Y_0 ≥ Y 0 . All vectors are two dimensional.
The Balitsky–Kovchegov (BK) nonlinear equation [7 , 8 ] for the scattering amplitude N ( x 0 , x 1 , Y ) N(x_0,x_1,Y) N ( x 0 , x 1 , Y ) has the same form as Eq. (II.1) but for
S ( x 0 , x 1 , Y ) = 1 − N ( x 0 , x 1 , Y ) (13) S(x_0,x_1,Y) = 1 - N(x_0,x_1,Y) \tag*{(13)} S ( x 0 , x 1 , Y ) = 1 − N ( x 0 , x 1 , Y ) (13)
The difference between them is only in the initial condition, which for Eq. (II.1) has the form:
S D ( x 0 , x 1 , Y = Y 01 , Y 01 ) = S 2 ( x 0 , x 1 , Y 01 ) (14) S^D(x_0,x_1,Y=Y_{01},Y_{01}) = S^2(x_0,x_1,Y_{01}) \tag*{(14)} S D ( x 0 , x 1 , Y = Y 01 , Y 01 ) = S 2 ( x 0 , x 1 , Y 01 ) (14)
To calculate the diffraction cross sections in QCD, we need to modify Eq. (I.1) by (i) introducing two scattering amplitudes: γ \gamma γ for the scattering amplitude in Figure 1 and γ ˉ \bar{\gamma} γ ˉ for the complex conjugate amplitude (see Ref. [24 ] for details); and (ii) setting a new definition for parton densities ρ n , k D ( Y − Y 0 , Y 01 , r , b ; r 1 , b 1 , … , r n , b n ; … , r k , b k ) \rho^D_{n,k}(Y-Y_0,Y_{01},r,b;r_1,b_1,\ldots,r_n,b_n;\ldots,r_k,b_k) ρ n , k D ( Y − Y 0 , Y 01 , r , b ; r 1 , b 1 , … , r n , b n ; … , r k , b k ) :
ρ n , k D ( Y − Y 0 , Y 01 , r , b ; { r i , b i } , { r j , b j } ) = 1 n ! ∏ i = 1 n δ δ u i 1 k ! ∏ j = 1 k δ δ u ˉ j Z D ( Y − Y 0 , Y 01 , r , b ; [ u i ] , [ u ˉ j ] ) ∣ u i = 1 , u ˉ j = 1 (15) \rho^D_{n,k}(Y-Y_0,Y_{01},r,b;\{r_i,b_i\},\{r_j,b_j\}) = \frac{1}{n!}\prod_{i=1}^{n}\frac{\delta}{\delta u_i}\frac{1}{k!}\prod_{j=1}^{k}\frac{\delta}{\delta\bar{u}_j}Z^D(Y-Y_0,Y_{01},r,b;[u_i],[\bar{u}_j])\bigg|_{u_i=1,\bar{u}_j=1} \tag*{(15)} ρ n , k D ( Y − Y 0 , Y 01 , r , b ; { r i , b i } , { r j , b j }) = n ! 1 i = 1 ∏ n δ u i δ k ! 1 j = 1 ∏ k δ u ˉ j δ Z D ( Y − Y 0 , Y 01 , r , b ; [ u i ] , [ u ˉ j ]) u i = 1 , u ˉ j = 1 (15)
with
Z D ( Y − Y 0 , Y 01 , r , b ; [ u i ] , [ u ˉ j ] ) = S D ( Y − Y 0 , Y 01 , r , b ; γ = 1 − u i , γ ˉ = 1 − u ˉ j ) . Z^D(Y-Y_0,Y_{01},r,b;[u_i],[\bar{u}_j]) = S^D(Y-Y_0,Y_{01},r,b;\gamma=1-u_i,\bar{\gamma}=1-\bar{u}_j). Z D ( Y − Y 0 , Y 01 , r , b ; [ u i ] , [ u ˉ j ]) = S D ( Y − Y 0 , Y 01 , r , b ; γ = 1 − u i , γ ˉ = 1 − u ˉ j ) .
Using these densities we can rewrite Eq. (I.1) in the following form for single diffraction in the dipole-dipole scattering (see Figure 3 -a):
Figure 3. Summing large Pomeron loops for diffraction in the dipole-dipole scattering. The wavy lines denote the BFKL Pomeron exchanges. The yellow circles denote the amplitude γ \gamma γ . The black circle denotes the triple Pomeron vertices. Fig. 3-a shows the single diffraction cross section. For simplicity we put only the BFKL Pomeron exchange for the production of the mass ln M 1 2 = Y − Y 01 \ln M_1^2 = Y - Y_{01} ln M 1 2 = Y − Y 01 . Fig. 3-b describes the double diffraction. Only the BFKL Pomeron exchanges are shown for production ln M 1 2 = Y − Y 01 \ln M_1^2 = Y - Y_{01} ln M 1 2 = Y − Y 01 and ln M 2 2 = Y 02 − 0 \ln M_2^2 = Y_{02} - 0 ln M 2 2 = Y 02 − 0 . S S S and S ∗ S^* S ∗ stand for the S S S -matrix and its conjugate.
S d d D ( Y − Y 01 , Y 0 , r , r ′ ; b ) = (16) S^D_{dd}(Y-Y_{01},Y_0,r,r';b) = \tag*{(16)} S dd D ( Y − Y 01 , Y 0 , r , r ′ ; b ) = (16)
∑ n = 0 ∞ ( − 1 ) n n ! ∑ k = 0 ∞ ( − 1 ) k k ! ∫ ∏ i = 1 n d 2 r i d 2 r i ′ d 2 b i ′ ∏ j = 1 k d 2 r j d 2 r j ′ d 2 b j ′ ρ n , k D ( Y − Y 0 , Y 01 , r , b ; { r i , b i } , { r j , b j } ) \sum_{n=0}^{\infty}(-1)^n n!\sum_{k=0}^{\infty}(-1)^k k!\int\prod_{i=1}^{n}d^2r_i\,d^2r'_i\,d^2b'_i\prod_{j=1}^{k}d^2r_j\,d^2r'_j\,d^2b'_j\,\rho^D_{n,k}(Y-Y_0,Y_{01},r,b;\{r_i,b_i\},\{r_j,b_j\}) n = 0 ∑ ∞ ( − 1 ) n n ! k = 0 ∑ ∞ ( − 1 ) k k ! ∫ i = 1 ∏ n d 2 r i d 2 r i ′ d 2 b i ′ j = 1 ∏ k d 2 r j d 2 r j ′ d 2 b j ′ ρ n , k D ( Y − Y 0 , Y 01 , r , b ; { r i , b i } , { r j , b j })
× ∫ d 2 δ b i γ B A ( r i , r i ′ , b i − b i ′ ≡ δ b i ) ∫ d 2 δ b j γ ˉ B A ( r j , r j ′ , b j − b j ′ ≡ δ b j ) ρ n ( Y 0 , r , b ; { r i ′ , b i ′ } ) ρ k ( Y 0 , r , b ; { r j ′ , b j ′ } ) . {}\times\int d^2\delta b_i\,\gamma^{BA}(r_i,r'_i,b_i-b'_i\equiv\delta b_i)\int d^2\delta b_j\,\bar{\gamma}^{BA}(r_j,r'_j,b_j-b'_j\equiv\delta b_j)\rho_n(Y_0,r,b;\{r'_i,b'_i\})\rho_k(Y_0,r,b;\{r'_j,b'_j\}). × ∫ d 2 δ b i γ B A ( r i , r i ′ , b i − b i ′ ≡ δ b i ) ∫ d 2 δ b j γ ˉ B A ( r j , r j ′ , b j − b j ′ ≡ δ b j ) ρ n ( Y 0 , r , b ; { r i ′ , b i ′ }) ρ k ( Y 0 , r , b ; { r j ′ , b j ′ }) .
For double diffraction we have:
S D D ( Y − Y 01 , Y 02 , Y 01 − Y 02 , r , r ′ ; b ) = (17) S^{DD}(Y-Y_{01},Y_{02},Y_{01}-Y_{02},r,r';b) = \tag*{(17)} S D D ( Y − Y 01 , Y 02 , Y 01 − Y 02 , r , r ′ ; b ) = (17)
∑ n = 0 ∞ ( − 1 ) n n ! ∑ k = 0 ∞ ( − 1 ) k k ! ∫ ∏ i = 1 n d 2 r i d 2 r i ′ d 2 b i ′ ∏ j = 1 k d 2 r j d 2 r j ′ d 2 b j ′ ρ n , k D ( Y − Y 0 , Y 01 , r , b ; { r i , b i } , { r j , b j } ) \sum_{n=0}^{\infty}(-1)^n n!\sum_{k=0}^{\infty}(-1)^k k!\int\prod_{i=1}^{n}d^2r_i\,d^2r'_i\,d^2b'_i\prod_{j=1}^{k}d^2r_j\,d^2r'_j\,d^2b'_j\,\rho^D_{n,k}(Y-Y_0,Y_{01},r,b;\{r_i,b_i\},\{r_j,b_j\}) n = 0 ∑ ∞ ( − 1 ) n n ! k = 0 ∑ ∞ ( − 1 ) k k ! ∫ i = 1 ∏ n d 2 r i d 2 r i ′ d 2 b i ′ j = 1 ∏ k d 2 r j d 2 r j ′ d 2 b j ′ ρ n , k D ( Y − Y 0 , Y 01 , r , b ; { r i , b i } , { r j , b j })
× ∫ d 2 δ b i γ B A ( r i , r i ′ , b i − b i ′ ≡ δ b i ) ∫ d 2 δ b j γ ˉ B A ( r j , r j ′ , b j − b j ′ ≡ δ b j ) ρ n , k D ( Y 0 , Y 02 , r ′ , b ′ ; { r i ′ , b i ′ } , { r j ′ , b j ′ } ) . {}\times\int d^2\delta b_i\,\gamma^{BA}(r_i,r'_i,b_i-b'_i\equiv\delta b_i)\int d^2\delta b_j\,\bar{\gamma}^{BA}(r_j,r'_j,b_j-b'_j\equiv\delta b_j)\rho^D_{n,k}(Y_0,Y_{02},r',b';\{r'_i,b'_i\},\{r'_j,b'_j\}). × ∫ d 2 δ b i γ B A ( r i , r i ′ , b i − b i ′ ≡ δ b i ) ∫ d 2 δ b j γ ˉ B A ( r j , r j ′ , b j − b j ′ ≡ δ b j ) ρ n , k D ( Y 0 , Y 02 , r ′ , b ′ ; { r i ′ , b i ′ } , { r j ′ , b j ′ }) .
To carry out calculations using Eq. (II.6) and Eq. (II.7), we note that t t t -channel unitarity for the BFKL Pomeron in QCD takes the following form [14 ] (see notations in Eq. (I.1)):
G I P ( Y , r , r ′ , b ) = ∫ d 2 r i d 2 b i d 2 r i ′ d 2 b i ′ G I P ( Y − Y 0 , r , b − b i , r i ) γ B A ( r i , r i ′ , δ b i ) G I P ( Y 0 ; r ′ , b i ′ , r i ′ ) (18) G_{IP}(Y,r,r',b) = \int d^2r_i\,d^2b_i\,d^2r'_i\,d^2b'_i\,G_{IP}(Y-Y_0,r,b-b_i,r_i)\gamma^{BA}(r_i,r'_i,\delta b_i)G_{IP}(Y_0;r',b'_i,r'_i) \tag*{(18)} G I P ( Y , r , r ′ , b ) = ∫ d 2 r i d 2 b i d 2 r i ′ d 2 b i ′ G I P ( Y − Y 0 , r , b − b i , r i ) γ B A ( r i , r i ′ , δ b i ) G I P ( Y 0 ; r ′ , b i ′ , r i ′ ) (18)
where γ B A ( r i , r i ′ , δ b i ) \gamma^{BA}(r_i,r'_i,\delta b_i) γ B A ( r i , r i ′ , δ b i ) is the scattering amplitude of two dipoles in Born approximation of perturbative QCD.
Equation for double diffraction
In this section we attempt to write the evolution equation for double diffraction production in dipole-dipole scattering, based on summing large Pomeron loops. We use the observation that in Eq. (I.1), for the dipole-dipole scattering amplitude, the structure of the dipole cascade is the same as in BK evolution, both for Y > Y 0 Y > Y_0 Y > Y 0 and for Y < Y 0 Y < Y_0 Y < Y 0 . Therefore, for the fastest and the slowest dipoles we can trust the BK equations.
With this in mind, for Y − Y 0 > Y m 1 > 0 Y-Y_0 > Y_{m1} > 0 Y − Y 0 > Y m 1 > 0 we can write the following equation for double diffraction3 :
∂ S D D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 ) ∂ Y m 1 = α ˉ S 2 π ∫ d 2 x 2 x 01 2 x 02 2 x 12 2 { S D D ( x 1 , x 2 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 ) S D D ( x 0 , x 2 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 ) − S D D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 ) } , (19) \frac{\partial S^{DD}(x_0,x_1,x'_0,x'_1,Y_{m1},Y_{12},Y_{m2})}{\partial Y_{m1}} = \frac{\bar{\alpha}_S}{2\pi}\int d^2x_2\frac{x_{01}^2}{x_{02}^2x_{12}^2}
\left\{S^{DD}(x_1,x_2,x'_0,x'_1,Y_{m1},Y_{12},Y_{m2})S^{DD}(x_0,x_2,x'_0,x'_1,Y_{m1},Y_{12},Y_{m2})-S^{DD}(x_0,x_1,x'_0,x'_1,Y_{m1},Y_{12},Y_{m2})\right\}, \tag*{(19)} ∂ Y m 1 ∂ S D D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 ) = 2 π α ˉ S ∫ d 2 x 2 x 02 2 x 12 2 x 01 2 { S D D ( x 1 , x 2 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 ) S D D ( x 0 , x 2 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 ) − S D D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 ) } , (19)
with the initial condition
S D D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 = 0 , Y 12 , Y m 2 ) = S d d D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y , Y m 2 ) (20) S^{DD}(x_0,x_1,x'_0,x'_1,Y_{m1}=0,Y_{12},Y_{m2})=S^D_{dd}(x_0,x_1,x'_0,x'_1,Y,Y_{m2}) \tag*{(20)} S D D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 = 0 , Y 12 , Y m 2 ) = S dd D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y , Y m 2 ) (20)
where S d d D S^D_{dd} S dd D describes single diffraction in dipole-dipole scattering.
For Y m 2 < Y 0 Y_{m2}<Y_0 Y m 2 < Y 0 the same equation takes the form
∂ S D D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 ) ∂ Y m 2 = − α ˉ S 2 π ∫ d 2 x 2 ′ x 01 ′ 2 x 02 ′ 2 x 12 ′ 2 { S D D ( x 0 , x 1 , x 1 ′ , x 2 ′ , Y m 1 , Y 12 , Y m 2 ) S D D ( x 0 , x 1 , x 0 ′ , x 2 ′ , Y m 1 , Y 12 , Y m 2 ) − S D D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 ) } , (21) \frac{\partial S^{DD}(x_0,x_1,x'_0,x'_1,Y_{m1},Y_{12},Y_{m2})}{\partial Y_{m2}}=-\frac{\bar{\alpha}_S}{2\pi}\int d^2x'_2\frac{x_{01}^{\prime2}}{x_{02}^{\prime2}x_{12}^{\prime2}}
\left\{S^{DD}(x_0,x_1,x'_1,x'_2,Y_{m1},Y_{12},Y_{m2})S^{DD}(x_0,x_1,x'_0,x'_2,Y_{m1},Y_{12},Y_{m2})-S^{DD}(x_0,x_1,x'_0,x'_1,Y_{m1},Y_{12},Y_{m2})\right\}, \tag*{(21)} ∂ Y m 2 ∂ S D D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 ) = − 2 π α ˉ S ∫ d 2 x 2 ′ x 02 ′2 x 12 ′2 x 01 ′2 { S D D ( x 0 , x 1 , x 1 ′ , x 2 ′ , Y m 1 , Y 12 , Y m 2 ) S D D ( x 0 , x 1 , x 0 ′ , x 2 ′ , Y m 1 , Y 12 , Y m 2 ) − S D D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 ) } , (21)
and the initial condition of Eq. (II.10) becomes
S D D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 = 0 ) = S d d D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y ) (22) S^{DD}(x_0,x_1,x'_0,x'_1,Y_{m1},Y_{12},Y_{m2}=0)=S^D_{dd}(x_0,x_1,x'_0,x'_1,Y_{m1},Y) \tag*{(22)} S D D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 , Y m 2 = 0 ) = S dd D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y ) (22)
These equations must be supplemented by the equations for single diffraction in dipole-dipole scattering, which have a similar form:
∂ S d d D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 ) ∂ Y m 1 = α ˉ S 2 π ∫ d 2 x 2 x 01 2 x 02 2 x 12 2 { S d d D ( x 1 , x 2 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 ) S d d D ( x 0 , x 2 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 ) − S d d D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 ) } (23) \frac{\partial S_{dd}^{D}(x_0,x_1,x'_0,x'_1,Y_{m1},Y_{12})}{\partial Y_{m1}}=\frac{\bar{\alpha}_S}{2\pi}\int d^2x_2\frac{x_{01}^2}{x_{02}^2x_{12}^2}\left\{S_{dd}^{D}(x_1,x_2,x'_0,x'_1,Y_{m1},Y_{12})S_{dd}^{D}(x_0,x_2,x'_0,x'_1,Y_{m1},Y_{12})-S_{dd}^{D}(x_0,x_1,x'_0,x'_1,Y_{m1},Y_{12})\right\} \tag*{(23)} ∂ Y m 1 ∂ S dd D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 ) = 2 π α ˉ S ∫ d 2 x 2 x 02 2 x 12 2 x 01 2 { S dd D ( x 1 , x 2 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 ) S dd D ( x 0 , x 2 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 ) − S dd D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 , Y 12 ) } (23)
− ∂ S d d D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y 12 , Y m 2 ) ∂ Y m 2 = − α ˉ S 2 π ∫ d 2 x 2 ′ x 01 ′ 2 x 02 ′ 2 x 12 ′ 2 { S d d D ( x 0 , x 1 , x 1 ′ , x 2 ′ , Y 12 , Y m 2 ) S d d D ( x 0 , x 1 , x 0 ′ , x 2 ′ , Y 12 , Y m 2 ) − S d d D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y 12 , Y m 2 ) } . (24) -\frac{\partial S_{dd}^{D}(x_0,x_1,x'_0,x'_1,Y_{12},Y_{m2})}{\partial Y_{m2}}=-\frac{\bar{\alpha}_S}{2\pi}\int d^2x'_2\frac{x'_{01}{}^2}{x'_{02}{}^2x'_{12}{}^2}\left\{S_{dd}^{D}(x_0,x_1,x'_1,x'_2,Y_{12},Y_{m2})S_{dd}^{D}(x_0,x_1,x'_0,x'_2,Y_{12},Y_{m2})-S_{dd}^{D}(x_0,x_1,x'_0,x'_1,Y_{12},Y_{m2})\right\}. \tag*{(24)} − ∂ Y m 2 ∂ S dd D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y 12 , Y m 2 ) = − 2 π α ˉ S ∫ d 2 x 2 ′ x 02 ′ 2 x 12 ′ 2 x 01 ′ 2 { S dd D ( x 0 , x 1 , x 1 ′ , x 2 ′ , Y 12 , Y m 2 ) S dd D ( x 0 , x 1 , x 0 ′ , x 2 ′ , Y 12 , Y m 2 ) − S dd D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y 12 , Y m 2 ) } . (24)
The initial conditions for both equations have the form
S d d D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y , Y m 2 = 0 ) = S d d D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 = 0 , Y ) = ( S d d ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y ) ) 2 , (25) S_{dd}^{D}(x_0,x_1,x'_0,x'_1,Y,Y_{m2}=0)=S_{dd}^{D}(x_0,x_1,x'_0,x'_1,Y_{m1}=0,Y)=\left(S_d^d(x_0,x_1,x'_0,x'_1,Y)\right)^2, \tag*{(25)} S dd D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y , Y m 2 = 0 ) = S dd D ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y m 1 = 0 , Y ) = ( S d d ( x 0 , x 1 , x 0 ′ , x 1 ′ , Y ) ) 2 , (25)
where S d d S_d^d S d d is the elastic S S S -matrix for dipole-dipole scattering, discussed in Appendix B.
In the rest of the paper we are going to discuss more detailed solution based of the direct application of Eq. (II.7).
Double diffraction in the one-dimensional model
Single diffraction
In the one-dimensional model [13 , 14 , 20 , 21 , 22 , 23 , 24 , 56 , 57 , 58 , 59 , 60 , 61 , 62 , 63 , 64 , 65 , 66 ] we neglect the dependence of the amplitude with respect to the transverse coordinate and Eq. (II.1) takes the form:
∂ S D ( Y , Y 01 ) ∂ Y = Δ I P { S D ( Y , Y 01 ) 2 − S D ( Y , Y 01 ) } (26) \frac{\partial S^D(Y,Y_{01})}{\partial Y}=\Delta_{IP}\left\{S^D(Y,Y_{01})^2-S^D(Y,Y_{01})\right\} \tag*{(26)} ∂ Y ∂ S D ( Y , Y 01 ) = Δ I P { S D ( Y , Y 01 ) 2 − S D ( Y , Y 01 ) } (26)
where Δ I P \Delta_{IP} Δ I P is the intercept of the BFKL Pomeron. Introducing a new variable Y ~ = Δ I P Y \widetilde{Y}=\Delta_{IP}Y Y = Δ I P Y one can see that Eq. (III.1) has the general solution:
S D ( Y ~ , Y ~ 01 ) = 1 1 + Φ ( Y ~ 01 ) e Y ~ − Y ~ 01 (27) S^D(\widetilde{Y},\widetilde{Y}_{01})=\frac{1}{1+\Phi(\widetilde{Y}_{01})e^{\widetilde{Y}-\widetilde{Y}_{01}}} \tag*{(27)} S D ( Y , Y 01 ) = 1 + Φ ( Y 01 ) e Y − Y 01 1 (27)
with the equation for Φ \Phi Φ :
1 1 + Φ ( Y ~ 01 ) = S 2 ( Y ~ 01 ) (28) \frac{1}{1+\Phi(\widetilde{Y}_{01})}=S^2(\widetilde{Y}_{01}) \tag*{(28)} 1 + Φ ( Y 01 ) 1 = S 2 ( Y 01 ) (28)
Solution for S S S of the same equation is
S ( γ , Y ~ ) = 1 − γ 1 + γ ( e Y ~ − 1 ) = 1 1 + γ R e Y ~ (29) S(\gamma,\widetilde{Y})=\frac{1-\gamma}{1+\gamma(e^{\widetilde{Y}}-1)}=\frac{1}{1+\gamma_R e^{\widetilde{Y}}} \tag*{(29)} S ( γ , Y ) = 1 + γ ( e Y − 1 ) 1 − γ = 1 + γ R e Y 1 (29)
with the initial condition S ( γ , Y ~ = 0 ) = 1 − γ S(\gamma,\widetilde{Y}=0)=1-\gamma S ( γ , Y = 0 ) = 1 − γ where γ \gamma γ is the amplitude of the interaction with the target at low energies, γ R = γ / ( 1 − γ ) \gamma_R=\gamma/(1-\gamma) γ R = γ / ( 1 − γ ) . Plugging this equation into Eq. (III.3) we obtain:
S D ( Y ~ , Y ~ 01 ) = 1 1 − e Y ~ m 1 + e Y ~ m 1 S − 2 ( γ , Y ~ 01 ) = 1 1 + γ R e Y ~ ( 2 + γ R e Y ~ 01 ) (30) S^D(\widetilde{Y},\widetilde{Y}_{01})=\frac{1}{1-e^{\widetilde{Y}_{m1}}+e^{\widetilde{Y}_{m1}}S^{-2}(\gamma,\widetilde{Y}_{01})}=\frac{1}{1+\gamma_R e^{\widetilde{Y}}\left(2+\gamma_R e^{\widetilde{Y}_{01}}\right)} \tag*{(30)} S D ( Y , Y 01 ) = 1 − e Y m 1 + e Y m 1 S − 2 ( γ , Y 01 ) 1 = 1 + γ R e Y ( 2 + γ R e Y 01 ) 1 (30)
The cross section for the diffractive production with fixed rapidity gap Y 01 Y_{01} Y 01 is equal to [2 ]
σ D ( Y ~ , Y ~ 01 ) = − ∂ N D ( Y ~ , Y ~ 01 ) ∂ Y ~ 01 = − ∂ S D ( Y ~ , Y ~ 01 ) ∂ Y ~ 01 (31) \sigma^D(\widetilde{Y},\widetilde{Y}_{01})=-\frac{\partial N^D(\widetilde{Y},\widetilde{Y}_{01})}{\partial\widetilde{Y}_{01}}=-\frac{\partial S^D(\widetilde{Y},\widetilde{Y}_{01})}{\partial\widetilde{Y}_{01}} \tag*{(31)} σ D ( Y , Y 01 ) = − ∂ Y 01 ∂ N D ( Y , Y 01 ) = − ∂ Y 01 ∂ S D ( Y , Y 01 ) (31)
In Eq. (III.6) we use Eq. (II.2) calculating ∂ N D ( Y ~ , Y ~ 01 ) ∂ Y ~ 01 \frac{\partial N^D(\tilde{Y},\tilde{Y}_{01})}{\partial\tilde{Y}_{01}} ∂ Y ~ 01 ∂ N D ( Y ~ , Y ~ 01 ) . For Eq. (III.5) it takes the form4
σ D ( Y ~ , Y ~ 01 ) = γ R 2 e Y ~ + Y ~ 01 ( 1 + γ R e Y ~ ( 2 + γ R e Y ~ 01 ) ) 2 (32) \sigma^D(\tilde{Y},\tilde{Y}_{01})=\frac{\gamma_R^2 e^{\tilde{Y}+\tilde{Y}_{01}}}{\left(1+\gamma_R e^{\tilde{Y}}\left(2+\gamma_R e^{\tilde{Y}_{01}}\right)\right)^2} \tag*{(32)} σ D ( Y ~ , Y ~ 01 ) = ( 1 + γ R e Y ~ ( 2 + γ R e Y ~ 01 ) ) 2 γ R 2 e Y ~ + Y ~ 01 (32)
where Y ~ = Y ~ m 1 + Y ~ 01 \tilde{Y}=\tilde{Y}_{m1}+\tilde{Y}_{01} Y ~ = Y ~ m 1 + Y ~ 01 .
We plot this cross section in Figure 4 for γ = 0.1 \gamma=0.1 γ = 0.1 ; one can see that σ D \sigma^D σ D decreases as a function of Y ~ 01 \tilde{Y}_{01} Y ~ 01 and also decreases with increasing Y ~ m 1 \tilde{Y}_{m1} Y ~ m 1 . The characteristic behaviour exhibits a maximum at Y 0 = Y m a x Y_0=Y_{max} Y 0 = Y ma x , and Y m a x Y_{max} Y ma x decreases as Y ~ m 1 \tilde{Y}_{m1} Y ~ m 1 increases. All these features have been discussed in Ref. [2 ] . Coming back to Eq. (III.5) one can see that this solution can be rewritten in the following form:
Figure 4. σ D \sigma^D σ D versus Y ~ 01 \tilde{Y}_{01} Y ~ 01 at fixed Y ~ m 1 \tilde{Y}_{m1} Y ~ m 1 . γ = 0.1 \gamma=0.1 γ = 0.1 .
S D ( Y ~ , Y ~ 01 ) = 1 1 + γ R e Y ~ m 1 + Δ z ( Y ~ 01 ) ≡ S ( Y ~ m 1 + Δ z ( Y ~ 01 ) ) with Δ z ( Y ~ 01 ) = Y ~ 01 + ln ( 2 + γ R e Y ~ 01 ) ; (33) S^D(\tilde{Y},\tilde{Y}_{01})=\frac{1}{1+\gamma_R e^{\tilde{Y}_{m1}+\Delta z(\tilde{Y}_{01})}}\equiv S\left(\tilde{Y}_{m1}+\Delta z\left(\tilde{Y}_{01}\right)\right)\ \text{with}\ \Delta z\left(\tilde{Y}_{01}\right)=\tilde{Y}_{01}+\ln\left(2+\gamma_R e^{\tilde{Y}_{01}}\right); \tag*{(33)} S D ( Y ~ , Y ~ 01 ) = 1 + γ R e Y ~ m 1 + Δ z ( Y ~ 01 ) 1 ≡ S ( Y ~ m 1 + Δ z ( Y ~ 01 ) ) with Δ z ( Y ~ 01 ) = Y ~ 01 + ln ( 2 + γ R e Y ~ 01 ) ; (33)
The function Δ z ( Y ~ 01 ) \Delta z\left(\tilde{Y}_{01}\right) Δ z ( Y ~ 01 ) can be found from the following equation:
S ( Δ z ( Y ~ 01 ) ) = S 2 ( Y ~ 01 ) (34) S\left(\Delta z\left(\tilde{Y}_{01}\right)\right)=S^2\left(\tilde{Y}_{01}\right) \tag*{(34)} S ( Δ z ( Y ~ 01 ) ) = S 2 ( Y ~ 01 ) (34)
This is easy to understand: the equation for S D S^D S D has the same form as the Balitsky–Kovchegov equation for the elastic scattering matrix S S S . The two therefore have the same general solution (see Eq. (III.8)) and differ only in the initial condition (see Eq. (III.9)).
Double diffraction
It is well known that high energy QCD can be formulated as the BFKL Pomeron calculus: the exchange and interactions between BFKL Pomerons [2 , 5 , 6 , 34 , 36 , 37 , 39 , 41 , 46 , 48 ] . In particular, in Ref. [54 ] it is shown that ρ n \rho_n ρ n in Eq. (I.1) can be found in this approach. In Figure 5 we give an example of the Pomeron diagram that describes the single diffractive production for the interaction of the dipole with a nucleus. Our main idea is to use Eq. (I.8) for ρ n A ( Y 0 = 0 , r ′ , b ′ ; { r i , b i } ) \rho_n^A(Y_0=0,r',b';\{r_i,b_i\}) ρ n A ( Y 0 = 0 , r ′ , b ′ ; { r i , b i }) and Eq. (I.1) and find the Pomeron structure of ρ n \rho_n ρ n of the fast dipole for the diffractive production. An example of such Pomeron diagrams can be seen in Figure 5 -c.
Figure 5. Fig. 5-a and b: An example of the Pomeron diagrams that contribute to the single diffractive production. Wavy lines are Pomerons. The vertical dashed lines show the cut Pomerons. The black circle denotes the triple Pomeron vertices while the yellow circle shows the scattering amplitude with the nucleon. Fig. 5-c shows the contribution of these diagrams to Eq. (I.1).
After finding ρ n \rho_n ρ n for the fast dipoles we use Eq. (I.1) to find the double diffractive production cross section (see Figure 6 in which we show an example of our calculations for the simple diagram of Figure 5 ). However, as has been noted in Ref. [24 ] , we need to introduce two scattering amplitudes: γ \gamma γ for the scattering amplitude in Figure 1 and γ ˉ \bar{\gamma} γ ˉ for the complex conjugate amplitude (see Eq. (II.5)). Bearing this in mind Eq. (III.5) takes the form:
Figure 6. Fig. 6-a: An example of our calculation of double diffraction, using Eq. (I.1) for the Pomeron diagrams of Figure 5 . Fig. 6-b: An example of the diagrams for single diffraction in dipole-dipole scattering. The notations are the same as for Figure 5 .
S D ( Y ~ , Y ~ 01 ) = 1 1 − e Y ~ m + e Y ~ m S − 1 ( γ , Y ~ 01 ) S − 1 ( γ ˉ , Y ~ 01 ) = 1 1 − e Y ~ m + e Y ~ m ( 1 + γ R e Y ~ 01 ) ( 1 + γ ˉ R e Y ~ 01 ) (35) S^{D}\left(\widetilde{Y},\widetilde{Y}_{01}\right)=\frac{1}{1-e^{\widetilde{Y}_{m}}+e^{\widetilde{Y}_{m}}S^{-1}\left(\gamma,\widetilde{Y}_{01}\right)S^{-1}\left(\bar{\gamma},\widetilde{Y}_{01}\right)}=\frac{1}{1-e^{\widetilde{Y}_{m}}+e^{\widetilde{Y}_{m}}\left(1+\gamma_{R}e^{\widetilde{Y}_{01}}\right)\left(1+\bar{\gamma}_{R}e^{\widetilde{Y}_{01}}\right)} \tag*{(35)} S D ( Y , Y 01 ) = 1 − e Y m + e Y m S − 1 ( γ , Y 01 ) S − 1 ( γ ˉ , Y 01 ) 1 = 1 − e Y m + e Y m ( 1 + γ R e Y 01 ) ( 1 + γ ˉ R e Y 01 ) 1 (35)
= − 1 e Y ~ m − 1 ∑ n = 0 ∞ ( Z ( Y ~ m ) ( 1 + γ R e Y ~ 01 ) ( 1 + γ ˉ R e Y ~ 01 ) ) n ; with Z ( Y ~ m ) = e Y ~ m e Y ~ m − 1 ; (36) =\frac{-1}{e^{\widetilde{Y}_{m}}-1}\sum_{n=0}^{\infty}\left(\mathcal{Z}\left(\widetilde{Y}_{m}\right)\left(1+\gamma_{R}e^{\widetilde{Y}_{01}}\right)\left(1+\bar{\gamma}_{R}e^{\widetilde{Y}_{01}}\right)\right)^{n};\ \text{with }\mathcal{Z}\left(\widetilde{Y}_{m}\right)=\frac{e^{\widetilde{Y}_{m}}}{e^{\widetilde{Y}_{m}}-1}; \tag*{(36)} = e Y m − 1 − 1 n = 0 ∑ ∞ ( Z ( Y m ) ( 1 + γ R e Y 01 ) ( 1 + γ ˉ R e Y 01 ) ) n ; with Z ( Y m ) = e Y m − 1 e Y m ; (36)
= − 1 e Y ~ m − 1 ∑ n = 0 ∞ Z n ( Y ~ m ) ∑ l = 0 n ∑ k = 0 n n ! 2 ( n − l ) ! ( n − k ) ! ( γ R G I P ( Y ~ 01 ) ) l l ! ( γ ˉ R G I P ( Y ~ 01 ) ) k k ! (37) =\frac{-1}{e^{\widetilde{Y}_{m}}-1}\sum_{n=0}^{\infty}\mathcal{Z}^{n}\left(\widetilde{Y}_{m}\right)\sum_{l=0}^{n}\sum_{k=0}^{n}\frac{n!^{2}}{(n-l)!(n-k)!}\frac{\left(\gamma_{R}G_{IP}\left(\widetilde{Y}_{01}\right)\right)^{l}}{l!}\frac{\left(\bar{\gamma}_{R}G_{IP}\left(\widetilde{Y}_{01}\right)\right)^{k}}{k!} \tag*{(37)} = e Y m − 1 − 1 n = 0 ∑ ∞ Z n ( Y m ) l = 0 ∑ n k = 0 ∑ n ( n − l )! ( n − k )! n ! 2 l ! ( γ R G I P ( Y 01 ) ) l k ! ( γ ˉ R G I P ( Y 01 ) ) k (37)
Taking sum over n n n in Eq. (III.10c) and taking into account that ρ n ( Y ~ 01 ) = C n G I P n ( Y ~ 01 ) = ( γ R G I P ( Y ~ 01 ) ) n \rho_n\left(\widetilde{Y}_{01}\right)=C_nG_{IP}^{n}\left(\widetilde{Y}_{01}\right)=\left(\gamma_{R}G_{IP}\left(\widetilde{Y}_{01}\right)\right)^{n} ρ n ( Y 01 ) = C n G I P n ( Y 01 ) = ( γ R G I P ( Y 01 ) ) n (see Eq. (III.4)) we rewrite Eq. (III.10c) as
S D ( Y ~ , Y ~ 01 ) = − 1 e Y ~ m − 1 ∑ l = 0 ∞ ∑ k = 0 ∞ ( l ! ) 2 Z l ( Y ~ m ) 2 F 1 ( l + 1 , l + 1 ; − k + l + 1 ; Z ( Y ~ m ) ) ( l − k ) ! ρ l ( Y ~ 01 ) l ! ρ k ( Y ~ 01 ) k ! (38) S^{D}\left(\widetilde{Y},\widetilde{Y}_{01}\right)=\frac{-1}{e^{\widetilde{Y}_{m}}-1}\sum_{l=0}^{\infty}\sum_{k=0}^{\infty}\frac{(l!)^{2}\mathcal{Z}^{l}\left(\widetilde{Y}_{m}\right)\,{}_{2}F_{1}\left(l+1,l+1;-k+l+1;\mathcal{Z}\left(\widetilde{Y}_{m}\right)\right)}{(l-k)!}\frac{\rho_l\left(\widetilde{Y}_{01}\right)}{l!}\frac{\rho_k\left(\widetilde{Y}_{01}\right)}{k!} \tag*{(38)} S D ( Y , Y 01 ) = e Y m − 1 − 1 l = 0 ∑ ∞ k = 0 ∑ ∞ ( l − k )! ( l ! ) 2 Z l ( Y m ) 2 F 1 ( l + 1 , l + 1 ; − k + l + 1 ; Z ( Y m ) ) l ! ρ l ( Y 01 ) k ! ρ k ( Y 01 ) (38)
Recalling that S D S^{D} S D is the diffractive S-matrix of the dipole scattering with the nucleus at rest (Y 0 = 0 Y_{0}=0 Y 0 = 0 ) we can use Eq. (I.8) which has the following form for the one dimensional problem:
ρ n A ( Y 0 = 0 ) = γ R n n ! (39) \rho_n^{A}(Y_{0}=0)=\frac{\gamma_{R}^{n}}{n!} \tag*{(39)} ρ n A ( Y 0 = 0 ) = n ! γ R n (39)
Plugging Eq. (III.12) for the dipole densities in Eq. (II.6) with Y 0 = 0 Y_{0}=0 Y 0 = 0 we can extract ρ l , k D \rho_{l,k}^{D} ρ l , k D from S D S^{D} S D , viz.:
S D ( Y ~ m , Y ~ 01 ) = ∑ l = 0 ∞ ∑ k = 0 ∞ ( − 1 ) l + k ρ l , k D ( Y ~ m , Y ~ 01 ) γ R l γ ˉ R k (40) S^{D}\left(\widetilde{Y}_{m},\widetilde{Y}_{01}\right)=\sum_{l=0}^{\infty}\sum_{k=0}^{\infty}(-1)^{l+k}\rho_{l,k}^{D}\left(\widetilde{Y}_{m},\widetilde{Y}_{01}\right)\gamma_{R}^{l}\bar{\gamma}_{R}^{k} \tag*{(40)} S D ( Y m , Y 01 ) = l = 0 ∑ ∞ k = 0 ∑ ∞ ( − 1 ) l + k ρ l , k D ( Y m , Y 01 ) γ R l γ ˉ R k (40)
Comparing Eq. (III.11) with Eq. (III.13) we obtain the following expression for the diffractive parton densities ρ l , k D ( Y ~ m , Y ~ 01 ) \rho_{l,k}^{D}\left(\widetilde{Y}_{m},\widetilde{Y}_{01}\right) ρ l , k D ( Y m , Y 01 ) :
ρ l , k D ( Y ~ m , Y ~ 01 ) = ( − 1 ) l − k + 1 l ! Z l ( Y ~ m ) 2 F 1 ( l + 1 , l + 1 ; − k + l + 1 ; Z ( Y ~ m ) ) e Y ~ m − 1 G I P l ( Y ~ 01 ) ( l − k ) ! k ! G I P k ( Y ~ 01 ) (41) \rho_{l,k}^{D}\left(\widetilde{Y}_{m},\widetilde{Y}_{01}\right)=\frac{(-1)^{l-k+1}l!\mathcal{Z}^{l}\left(\widetilde{Y}_{m}\right)\,{}_{2}F_{1}\left(l+1,l+1;-k+l+1;\mathcal{Z}\left(\widetilde{Y}_{m}\right)\right)}{e^{\widetilde{Y}_{m}}-1}\frac{G_{IP}^{l}\left(\widetilde{Y}_{01}\right)}{(l-k)!k!}G_{IP}^{k}\left(\widetilde{Y}_{01}\right) \tag*{(41)} ρ l , k D ( Y m , Y 01 ) = e Y m − 1 ( − 1 ) l − k + 1 l ! Z l ( Y m ) 2 F 1 ( l + 1 , l + 1 ; − k + l + 1 ; Z ( Y m ) ) ( l − k )! k ! G I P l ( Y 01 ) G I P k ( Y 01 ) (41)
where 2 F 1 ( l + 1 , l + 1 ; − k + l + 1 ; Z ) {}_{2}F_{1}\left(l+1,l+1;-k+l+1;\mathcal{Z}\right) 2 F 1 ( l + 1 , l + 1 ; − k + l + 1 ; Z ) is the hypergeometric function (see Ref. [68 ] , formulae 9.2).
For large values of Y ~ m \widetilde{Y}_m Y m , Z ( Y ~ m ) − 1 → 1 / G I P ( Y ~ m ) ≪ 1 Z\left(\widetilde{Y}_m\right)-1 \to1/G_{IP}\left(\widetilde{Y}_m\right) \ll1 Z ( Y m ) − 1 → 1/ G I P ( Y m ) ≪ 1 , and therefore 2 F 1 ( l + 1 , l + 1 ; − k + l + 1 ; Z ( Y ~ m ) ) ≃ ( l − k ) ! ( l + k ) ! ( l ! ) 2 ( 1 − Z ( Y ~ m ) ) − l − k − 1 {}_2F_1\left(l+1,l+1;-k+l+1;Z\left(\widetilde{Y}_m\right)\right) \simeq\frac{(l-k)!(l+k)!}{(l!)^2}\left(1-Z\left(\widetilde{Y}_m\right)\right)^{-l-k-1} 2 F 1 ( l + 1 , l + 1 ; − k + l + 1 ; Z ( Y m ) ) ≃ ( l ! ) 2 ( l − k )! ( l + k )! ( 1 − Z ( Y m ) ) − l − k − 1 . In this limit, Eq. (III.14) takes a simple form:
ρ l , k D ( Y ~ m , Y ~ 01 ) = 1 e Y ~ m − 1 ( l + k ) ! ( Z ( Y ~ m ) − 1 ) l + k + 1 G I P l ( Y ~ 01 ) l ! G I P k ( Y ~ 01 ) k ! ( 1 + O ( e − Y ~ m ) ) . (42) \rho^D_{l,k}\left(\widetilde{Y}_m,\widetilde{Y}_{01}\right)=\frac{1}{e^{\widetilde{Y}_m}-1}\frac{(l+k)!}{\left(Z\left(\widetilde{Y}_m\right)-1\right)^{l+k+1}}\frac{G_{IP}^{l}\left(\widetilde{Y}_{01}\right)}{l!}\frac{G_{IP}^{k}\left(\widetilde{Y}_{01}\right)}{k!}\left(1+\mathcal{O}\left(e^{-\widetilde{Y}_m}\right)\right). \tag*{(42)} ρ l , k D ( Y m , Y 01 ) = e Y m − 1 1 ( Z ( Y m ) − 1 ) l + k + 1 ( l + k )! l ! G I P l ( Y 01 ) k ! G I P k ( Y 01 ) ( 1 + O ( e − Y m ) ) . (42)
Eq. (II.8) equation takes a simple form in the one-dimensional model:
G I P ( Y ~ ) = G I P ( Y ~ − Y ~ 0 ) G I P ( Y ~ 0 ) (43) G_{IP}\left(\widetilde{Y}\right)=G_{IP}\left(\widetilde{Y}-\widetilde{Y}_0\right)G_{IP}\left(\widetilde{Y}_0\right) \tag*{(43)} G I P ( Y ) = G I P ( Y − Y 0 ) G I P ( Y 0 ) (43)
since G I P ( Y ) G_{IP}(Y) G I P ( Y ) is equal to e Δ I P Y e^{\Delta_{IP}Y} e Δ I P Y . Hence we can put γ B A = 1 \gamma^{BA}=1 γ B A = 1 in Eq. (I.1).
Now we are ready to compute S D D S^{DD} S D D in the one dimensional model, which takes the form
S D D ( Y ~ m 1 , Y ~ m 2 ; Y ~ 01 − Y ~ 02 ) = ∑ l = 0 ∞ ( − 1 ) l l ! ∑ k = 0 ∞ ( − 1 ) k k ! ρ l , k D ( Y ~ m 1 , Y ~ 01 − Y ~ 0 ) ρ l , k D ( Y ~ m 2 , Y ~ 0 − Y ~ 02 ) (44) S^{DD}\left(\widetilde{Y}_{m1},\widetilde{Y}_{m2};\widetilde{Y}_{01}-\widetilde{Y}_{02}\right)=\sum_{l=0}^{\infty}(-1)^l l!\sum_{k=0}^{\infty}(-1)^k k!\rho^D_{l,k}\left(\widetilde{Y}_{m1},\widetilde{Y}_{01}-\widetilde{Y}_0\right)\rho^D_{l,k}\left(\widetilde{Y}_{m2},\widetilde{Y}_0-\widetilde{Y}_{02}\right) \tag*{(44)} S D D ( Y m 1 , Y m 2 ; Y 01 − Y 02 ) = l = 0 ∑ ∞ ( − 1 ) l l ! k = 0 ∑ ∞ ( − 1 ) k k ! ρ l , k D ( Y m 1 , Y 01 − Y 0 ) ρ l , k D ( Y m 2 , Y 0 − Y 02 ) (44)
Using this equation and Eq. (III.15) one can see that S D D S^{DD} S D D can be written as follows from Eq. (II.7) with j = k + l j=k+l j = k + l :
S D D ( Y ~ m 1 , Y ~ m 2 ; Y ~ 01 − Y ~ 02 ) = 1 e Y ~ m 1 − 1 1 e Y ~ m 2 − 1 ∑ j = 0 ∞ ∑ k = 0 j j ! 2 ( j − k ) ! k ! ( − G I P ( Y ~ 12 ) ) j ( ( Z ( Y ~ m 1 ) − 1 ) ( Z ( Y ~ m 2 ) − 1 ) ) j + 1 = 1 e Y ~ m 1 − 1 1 e Y ~ m 2 − 1 1 ( Z ( Y ~ m 1 ) − 1 ) ( Z ( Y ~ m 2 ) − 1 ) ∑ j = 0 ∞ j ! ( − ξ D D ) j ⏟ = 1 . (45) \begin{aligned}
S^{DD}\left(\widetilde{Y}_{m1},\widetilde{Y}_{m2};\widetilde{Y}_{01}-\widetilde{Y}_{02}\right)
&=\frac{1}{e^{\widetilde{Y}_{m1}}-1}\frac{1}{e^{\widetilde{Y}_{m2}}-1}\sum_{j=0}^{\infty}\sum_{k=0}^{j}\frac{j!^2}{(j-k)!k!}\frac{\left(-G_{IP}\left(\widetilde{Y}_{12}\right)\right)^j}{\left(\left(Z\left(\widetilde{Y}_{m1}\right)-1\right)\left(Z\left(\widetilde{Y}_{m2}\right)-1\right)\right)^{j+1}}\\
&=\frac{1}{e^{\widetilde{Y}_{m1}}-1}\frac{1}{e^{\widetilde{Y}_{m2}}-1}\underbrace{\frac{1}{\left(Z\left(\widetilde{Y}_{m1}\right)-1\right)\left(Z\left(\widetilde{Y}_{m2}\right)-1\right)}\sum_{j=0}^{\infty}j!\left(-\xi_{DD}\right)^j}_{=1}. \tag*{(45)}
\end{aligned} S D D ( Y m 1 , Y m 2 ; Y 01 − Y 02 ) = e Y m 1 − 1 1 e Y m 2 − 1 1 j = 0 ∑ ∞ k = 0 ∑ j ( j − k )! k ! j ! 2 ( ( Z ( Y m 1 ) − 1 ) ( Z ( Y m 2 ) − 1 ) ) j + 1 ( − G I P ( Y 12 ) ) j = e Y m 1 − 1 1 e Y m 2 − 1 1 = 1 ( Z ( Y m 1 ) − 1 ) ( Z ( Y m 2 ) − 1 ) 1 j = 0 ∑ ∞ j ! ( − ξ D D ) j . (45)
where ξ D D = 2 G I P ( Y ~ 12 ) ( Z ( Y ~ m 1 ) − 1 ) ( Z ( Y ~ m 2 ) − 1 ) \xi_{DD}=\frac{2G_{IP}\left(\widetilde{Y}_{12}\right)}{\left(Z\left(\widetilde{Y}_{m1}\right)-1\right)\left(Z\left(\widetilde{Y}_{m2}\right)-1\right)} ξ D D = ( Z ( Y m 1 ) − 1 ) ( Z ( Y m 2 ) − 1 ) 2 G I P ( Y 12 ) .
Summing over j j j we obtain:
S D D ( Y ~ m 1 , Y ~ m 2 ; Y ~ 01 − Y ~ 02 = Y ~ 12 ) = 1 ξ D D e 1 ξ D D Γ ( 0 , 1 ξ D D ) ⟶ ξ D D ≫ 1 ln ξ D D ξ D D (46) S^{DD}\left(\widetilde{Y}_{m1},\widetilde{Y}_{m2};\widetilde{Y}_{01}-\widetilde{Y}_{02}=\widetilde{Y}_{12}\right)=\frac{1}{\xi_{DD}}e^{\frac{1}{\xi_{DD}}}\Gamma\left(0,\frac{1}{\xi_{DD}}\right)\underset{\xi_{DD}\gg1}{\longrightarrow}\frac{\ln\xi_{DD}}{\xi_{DD}} \tag*{(46)} S D D ( Y m 1 , Y m 2 ; Y 01 − Y 02 = Y 12 ) = ξ D D 1 e ξ D D 1 Γ ( 0 , ξ D D 1 ) ξ D D ≫ 1 ⟶ ξ D D ln ξ D D (46)
For large Y ~ m 1 \widetilde{Y}_{m1} Y m 1 , Y ~ m 2 \widetilde{Y}_{m2} Y m 2 and Y ~ 12 \widetilde{Y}_{12} Y 12 Eq. (III.19) leads to the following expression:
S D D ( Y ~ m 1 , Y ~ m 2 ; Y ~ 12 ) = 1 2 ( Y ~ m 1 + Y ~ m 2 + Y ~ 12 ) e − Y ~ m 1 − Y ~ m 2 − Y ~ 12 (47) S^{DD}\left(\widetilde{Y}_{m1},\widetilde{Y}_{m2};\widetilde{Y}_{12}\right)=\frac{1}{2}\left(\widetilde{Y}_{m1}+\widetilde{Y}_{m2}+\widetilde{Y}_{12}\right)e^{-\widetilde{Y}_{m1}-\widetilde{Y}_{m2}-\widetilde{Y}_{12}} \tag*{(47)} S D D ( Y m 1 , Y m 2 ; Y 12 ) = 2 1 ( Y m 1 + Y m 2 + Y 12 ) e − Y m 1 − Y m 2 − Y 12 (47)
For single diffraction in dipole-dipole scattering, Eq. (II.6) takes the following form:
S d d D ( Y ~ m , Y ~ 01 ) = ∑ n = 0 ∞ ( − 1 ) n n ! ∑ k = 0 ∞ ( − 1 ) k k ! ρ n , k D ( Y ~ m , Y ~ 01 − Y ~ 0 ) ρ n ( Y ~ 0 ) ρ k ( Y ~ 0 ) (48) S^D_{dd}\left(\widetilde{Y}_m,\widetilde{Y}_{01}\right)=\sum_{n=0}^{\infty}(-1)^n n!\sum_{k=0}^{\infty}(-1)^k k!\rho^D_{n,k}\left(\widetilde{Y}_m,\widetilde{Y}_{01}-\widetilde{Y}_0\right)\rho_n\left(\widetilde{Y}_0\right)\rho_k\left(\widetilde{Y}_0\right) \tag*{(48)} S dd D ( Y m , Y 01 ) = n = 0 ∑ ∞ ( − 1 ) n n ! k = 0 ∑ ∞ ( − 1 ) k k ! ρ n , k D ( Y m , Y 01 − Y 0 ) ρ n ( Y 0 ) ρ k ( Y 0 ) (48)
Plugging Eq. (III.15) for ρ n , k D ( Y ~ m , Y ~ 01 − Y ~ 0 ) \rho^D_{n,k}\left(\widetilde{Y}_m,\widetilde{Y}_{01}-\widetilde{Y}_0\right) ρ n , k D ( Y m , Y 01 − Y 0 ) and Eq. (III.4) for ρ n \rho_n ρ n and taking sum of n + k n+k n + k we obtain
S d d D ( Y ~ m , Y ~ 01 ) = 1 ξ D 2 e 1 ξ D Γ ( − 1 , 1 ξ D ) ⟶ Y ~ m ≫ 1 1 ξ D (49) S^D_{dd}\left(\widetilde{Y}_m,\widetilde{Y}_{01}\right)=\frac{1}{\xi_D^2}e^{\frac{1}{\xi_D}}\Gamma\left(-1,\frac{1}{\xi_D}\right)\underset{\widetilde{Y}_m\gg1}{\longrightarrow}\frac{1}{\xi_D} \tag*{(49)} S dd D ( Y m , Y 01 ) = ξ D 2 1 e ξ D 1 Γ ( − 1 , ξ D 1 ) Y m ≫ 1 ⟶ ξ D 1 (49)
where ξ D = G I P ( Y ~ 01 ) ( Z ( Y ~ m ) − 1 ) \xi_D=\frac{G_{IP}\left(\widetilde{Y}_{01}\right)}{\left(Z\left(\widetilde{Y}_m\right)-1\right)} ξ D = ( Z ( Y m ) − 1 ) G I P ( Y 01 ) .
Double diffraction in high energy QCD
In QCD we need to find ρ n , k D \rho^{D}_{n,k} ρ n , k D dipole densities to obtain the double diffraction cross section using Eq. (II.7). As has been mentioned above, we start with diffractive production in dipole-nucleus scattering. The equation for single diffraction (see Eq. (II.1)) for the scattering of a dipole with a nucleus has the same form as the BK equation. The difference is only in the initial conditions of Eq. (II.4) for S D S^{D} S D . Since BK equation in the saturation region shows the geometric scaling behaviour [30 , 69 ] , the solution to Eq. (II.1) for the single diffraction can be generally written as follows [2 ] :
S D ( z m , z d 01 ) = S ( z m + Δ z ( z d 01 ) ) ; (50) S^{D}(z_m,z_d^{01}) = S\left(z_m+\Delta z\left(z_d^{01}\right)\right); \tag*{(50)} S D ( z m , z d 01 ) = S ( z m + Δ z ( z d 01 ) ) ; (50)
where function Δ z ( z d 01 ) \Delta z\left(z_d^{01}\right) Δ z ( z d 01 ) is determined by the following equation:
S ( Δ z ( z d 01 ) ) = S 2 ( z d 01 ) or Ω ( Δ z ( z d 01 ) ) = 2 Ω ( z d 01 ) (51) S\left(\Delta z\left(z_d^{01}\right)\right)=S^2\left(z_d^{01}\right)\quad\text{or}\quad\Omega\left(\Delta z\left(z_d^{01}\right)\right)=2\Omega\left(z_d^{01}\right) \tag*{(51)} S ( Δ z ( z d 01 ) ) = S 2 ( z d 01 ) or Ω ( Δ z ( z d 01 ) ) = 2Ω ( z d 01 ) (51)
with S D = exp ( − Ω D ) S^{D}=\exp\left(-\Omega^{D}\right) S D = exp ( − Ω D ) and S = exp ( − Ω ) S=\exp(-\Omega) S = exp ( − Ω ) . The variables z z z , which we will use, are (see Figure 1 , Figure 3 and Figure 6 )
z = κ α ˉ S Y + ξ r , R ; z d 01 = κ α ˉ S Y 01 + ξ r , R ; z d 12 = κ α ˉ S ( Y 01 − Y 02 ) + ξ r , r ′ (52) z=\kappa\bar{\alpha}_{S}Y+\xi_{r,R};\qquad z_d^{01}=\kappa\bar{\alpha}_{S}Y_{01}+\xi_{r,R};\qquad z_d^{12}=\kappa\bar{\alpha}_{S}\left(Y_{01}-Y_{02}\right)+\xi_{r,r'} \tag*{(52)} z = κ α ˉ S Y + ξ r , R ; z d 01 = κ α ˉ S Y 01 + ξ r , R ; z d 12 = κ α ˉ S ( Y 01 − Y 02 ) + ξ r , r ′ (52)
In Eq. (IV.3) the corresponding ξ \xi ξ 's are determined in Eq. (A.2b) of the Appendix A.
Single diffraction for the leading-twist BFKL kernel
At the moment we are not ready to discuss the solution of Eq. (IV.1) for the general BFKL kernel. However, we are able to find a solution in the entire region of z z z in the saturation region for the simplified BFKL kernel suggested in Ref. [15 ] . This kernel describes the high-energy asymptotic solution of the nonlinear BK equation and leads to geometric scaling behaviour:
χ ( γ ) = { 1 1 − γ for z = κ α ˉ S Y + ξ r , R + > 0 , summing ( z ) n ; 1 γ for z < 0 , summing ( ξ ) n ; (53) \chi(\gamma)=
\begin{cases}
\dfrac{1}{1-\gamma} & \text{for } z=\kappa\bar{\alpha}_{S}Y+\xi_{r,R+}>0,\quad\text{summing }(z)^n;\\
\dfrac{1}{\gamma} & \text{for } z<0,\quad\text{summing }(\xi)^n;
\end{cases} \tag*{(53)} χ ( γ ) = ⎩ ⎨ ⎧ 1 − γ 1 γ 1 for z = κ α ˉ S Y + ξ r , R + > 0 , summing ( z ) n ; for z < 0 , summing ( ξ ) n ; (53)
where κ = 4 \kappa=4 κ = 4 for this kernel.
Since this kernel sums logarithmic contributions, it corresponds to the leading-twist term of the full BFKL kernel. It has a very simple form in the coordinate representation [15 ] :
∫ K ( r ′ , r − r ′ ∣ r ) d 2 r ′ ⟶ α ˉ S 2 ∫ 1 / Q s 2 ( Y , b ) r 2 d r ′ 2 r ′ 2 + α ˉ S 2 ∫ 1 / Q s 2 ( Y , b ) r 2 d ∣ r − r ′ ∣ ∣ r − r ′ ∣ 2 = α ˉ S 2 ∫ ξ s ξ d ξ r ′ + α ˉ S 2 ∫ ξ s ξ d ξ r − r ′ (54) \int K\left(r',r-r'|r\right)\,\mathrm{d}^2r'
\longrightarrow
\frac{\bar{\alpha}_{S}}{2}\int_{1/Q_s^2(Y,b)}^{r^2}\frac{\mathrm{d}r'^2}{r'^2}
+
\frac{\bar{\alpha}_{S}}{2}\int_{1/Q_s^2(Y,b)}^{r^2}\frac{\mathrm{d}|r-r'|}{|r-r'|^2}
=
\frac{\bar{\alpha}_{S}}{2}\int_{\xi_s}^{\xi}\mathrm{d}\xi_{r'}
+
\frac{\bar{\alpha}_{S}}{2}\int_{\xi_s}^{\xi}\mathrm{d}\xi_{r-r'} \tag*{(54)} ∫ K ( r ′ , r − r ′ ∣ r ) d 2 r ′ ⟶ 2 α ˉ S ∫ 1/ Q s 2 ( Y , b ) r 2 r ′2 d r ′2 + 2 α ˉ S ∫ 1/ Q s 2 ( Y , b ) r 2 ∣ r − r ′ ∣ 2 d ∣ r − r ′ ∣ = 2 α ˉ S ∫ ξ s ξ d ξ r ′ + 2 α ˉ S ∫ ξ s ξ d ξ r − r ′ (54)
where ξ r ′ = ln ( r ′ 2 r 1 2 b 4 ) \xi_{r'}=\ln\left(\frac{r'^2r_1^2}{b^4}\right) ξ r ′ = ln ( b 4 r ′2 r 1 2 ) for b > r , r 1 b>r,r_1 b > r , r 1 and ξ s = κ α ˉ S Y \xi_s=\kappa\bar{\alpha}_S Y ξ s = κ α ˉ S Y for the scattering of the dipole r ′ r' r ′ with the dipole r 1 r_1 r 1 . Note that the logarithms originate from the decay of a large dipole into one small dipole and one large dipole [15 ] . However, the size of the small dipole remains larger than 1 / Q s 2 = b 4 r 1 2 e − ξ s 1/Q_s^2=\frac{b^4}{r_1^2}e^{-\xi_s} 1/ Q s 2 = r 1 2 b 4 e − ξ s . Using Eq. (II.1) we can obtain the nonlinear equation for Ω D ( x 0 , x 1 , Y , Y 01 ) \Omega^D(x_0,x_1,Y,Y_{01}) Ω D ( x 0 , x 1 , Y , Y 01 ) , which determines the S S S -matrix for diffraction in dipole-nucleus interaction S d A D ( x 0 , x 1 , Y , Y 01 ) = exp ( − Ω D ( x 0 , x 1 , Y , Y 01 ) ) S_{dA}^D(x_0,x_1,Y,Y_{01})=\exp(-\Omega^D(x_0,x_1,Y,Y_{01})) S d A D ( x 0 , x 1 , Y , Y 01 ) = exp ( − Ω D ( x 0 , x 1 , Y , Y 01 )) :
∂ 2 Ω D ( ξ s , ξ ) ∂ ξ s ∂ ξ = 1 4 ( 1 − e − Ω D ( ξ s , ξ ) ) (55) \frac{\partial^2\Omega^D(\xi_s,\xi)}{\partial\xi_s\,\partial\xi}=\frac{1}{4}\left(1-e^{-\Omega^D(\xi_s,\xi)}\right) \tag*{(55)} ∂ ξ s ∂ ξ ∂ 2 Ω D ( ξ s , ξ ) = 4 1 ( 1 − e − Ω D ( ξ s , ξ ) ) (55)
The initial conditions for Eq. (IV.6) are:
Ω D ( x 0 , x 1 , Y = Y 01 , Y 01 ) = 2 Ω ( x 0 , x 1 , Y = Y 01 ) (56) \Omega^D(x_0,x_1,Y=Y_{01},Y_{01})=2\Omega(x_0,x_1,Y=Y_{01}) \tag*{(56)} Ω D ( x 0 , x 1 , Y = Y 01 , Y 01 ) = 2Ω ( x 0 , x 1 , Y = Y 01 ) (56)
where Ω ( x 0 , x 1 , Y = Y 01 ) \Omega(x_0,x_1,Y=Y_{01}) Ω ( x 0 , x 1 , Y = Y 01 ) is defined through the elastic scattering matrix S = e − Ω S=e^{-\Omega} S = e − Ω . Eq. (IV.6) has the geometric scaling solution [30 , 69 ] (see formula 3.5.3 in Ref. [72 ] ), which has the following implicit form [70 , 71 ] :
2 ∫ Ω D ( z m = 0 , z d 01 ) Ω D ( z m , z d 01 ) d Ω ′ Ω ′ − 1 + exp ( − Ω ′ ) = z m (57) \sqrt{2}\int_{\Omega^D(z_m=0,z_d^{01})}^{\Omega^D(z_m,z_d^{01})}\frac{d\Omega'}{\sqrt{\Omega'-1+\exp(-\Omega')}}=z_m \tag*{(57)} 2 ∫ Ω D ( z m = 0 , z d 01 ) Ω D ( z m , z d 01 ) Ω ′ − 1 + exp ( − Ω ′ ) d Ω ′ = z m (57)
In Eq. (IV.8) Ω \Omega Ω is the initial condition for Ω D ( z m = 0 , z d 01 ) \Omega^D(z_m=0,z_d^{01}) Ω D ( z m = 0 , z d 01 ) : Ω D ( z m = 0 , z d 01 ) = 2 Ω ( z d 01 ) \Omega^D(z_m=0,z_d^{01})=2\Omega(z_d^{01}) Ω D ( z m = 0 , z d 01 ) = 2Ω ( z d 01 ) , where the elastic S S S -matrix is equal to S = e − Ω ( z d 01 ) S=e^{-\Omega(z_d^{01})} S = e − Ω ( z d 01 ) .
Let us consider both Ω D ( z m , z d 01 ) \Omega^D(z_m,z_d^{01}) Ω D ( z m , z d 01 ) and Ω D ( z m = 0 , z d 01 ) \Omega^D(z_m=0,z_d^{01}) Ω D ( z m = 0 , z d 01 ) being small. In this case Eq. (IV.8) reduces to
2 ∫ Ω D ( z m = 0 , z d 01 ) Ω D ( z m , z d 01 ) d Ω ′ 1 2 Ω ′ 2 = z m ; Ω D ( z m , z d 01 ) = Ω D ( z m = 0 , z d 01 ) e 1 2 z m (58) \sqrt{2}\int_{\Omega^D(z_m=0,z_d^{01})}^{\Omega^D(z_m,z_d^{01})}\frac{d\Omega'}{\sqrt{\frac{1}{2}\Omega'^2}}=z_m\ ;\qquad\Omega^D(z_m,z_d^{01})=\Omega^D(z_m=0,z_d^{01})e^{\frac{1}{2}z_m} \tag*{(58)} 2 ∫ Ω D ( z m = 0 , z d 01 ) Ω D ( z m , z d 01 ) 2 1 Ω ′2 d Ω ′ = z m ; Ω D ( z m , z d 01 ) = Ω D ( z m = 0 , z d 01 ) e 2 1 z m (58)
For small z m z_m z m and z d 01 z_d^{01} z d 01 the main contribution to the scattering amplitude comes from the exchange of the BFKL Pomerons. Using Eq. (II.2) and assuming the small values of Ω \Omega Ω 's we obtain:
Ω D ( z m = 0 , z d 01 ) = 2 N 0 e 1 2 z d 01 − N 0 2 e z d 01 (59) \Omega^D(z_m=0,z_d^{01})=2N_0e^{\frac{1}{2}z_d^{01}}-N_0^2e^{z_d^{01}} \tag*{(59)} Ω D ( z m = 0 , z d 01 ) = 2 N 0 e 2 1 z d 01 − N 0 2 e z d 01 (59)
where N 0 N_0 N 0 is the value of the scattering amplitude for the BFKL Pomeron exchange, which we consider as being small. Finally, we have from Eq. (IV.10) that
Ω D ( z m , z d 01 ) = 2 N 0 e 1 2 z ⏟ BFKL Pomeron exchange with Y − N 0 2 e 1 2 z m e z d 01 ⏟ Triple BFKL Pomeron contribution (60) \Omega^D(z_m,z_d^{01})=\underbrace{2N_0e^{\frac{1}{2}z}}_{\text{BFKL Pomeron exchange with }Y}-\underbrace{N_0^2e^{\frac{1}{2}z_m}e^{z_d^{01}}}_{\text{Triple BFKL Pomeron contribution}} \tag*{(60)} Ω D ( z m , z d 01 ) = BFKL Pomeron exchange with Y 2 N 0 e 2 1 z − Triple BFKL Pomeron contribution N 0 2 e 2 1 z m e z d 01 (60)
This confirms that Eq. (IV.8) correctly describes the Pomeron contributions to the scattering amplitude, and that all constants in the solution of Eq. (IV.6) have been chosen correctly.
Now we consider Ω ( z d 01 ) ≫ 1 \Omega(z_d^{01}) \gg1 Ω ( z d 01 ) ≫ 1 . For such Ω ( z d 01 ) \Omega(z_d^{01}) Ω ( z d 01 ) , Ω ′ \Omega' Ω ′ in Eq. (IV.8) turns out to be large and Eq. (IV.8) reduces to the following one:
2 ∫ 2 Ω ( z d 01 ) Ω D ( z m , z d 01 ) d Ω ′ Ω ′ = z m ; 2 2 ( Ω D ( z m , z d 01 ) − 2 Ω ( z d 01 ) ) = z m ; (61) \sqrt{2}\int_{2\Omega(z_d^{01})}^{\Omega^D(z_m,z_d^{01})}\frac{d\Omega'}{\sqrt{\Omega'}}=z_m;\qquad2\sqrt{2}\left(\sqrt{\Omega^D(z_m,z_d^{01})}-\sqrt{2\Omega(z_d^{01})}\right)=z_m; \tag*{(61)} 2 ∫ 2Ω ( z d 01 ) Ω D ( z m , z d 01 ) Ω ′ d Ω ′ = z m ; 2 2 ( Ω D ( z m , z d 01 ) − 2Ω ( z d 01 ) ) = z m ; (61)
From this equation
Ω D ( z m , z d 01 ) = 1 8 ( z m + 4 Ω ( z d 01 ) ) 2 = 1 2 κ ( z m + 2 κ Ω ( z d 01 ) ) 2 (62) \Omega^D\left(z_m,z_d^{01}\right)=\frac{1}{8}\left(z_m+4\sqrt{\Omega(z_d^{01})}\right)^2=\frac{1}{2\kappa}\left(z_m+2\sqrt{\kappa\Omega(z_d^{01})}\right)^2 \tag*{(62)} Ω D ( z m , z d 01 ) = 8 1 ( z m + 4 Ω ( z d 01 ) ) 2 = 2 κ 1 ( z m + 2 κ Ω ( z d 01 ) ) 2 (62)
Therefore, the solution of Eq. (IV.13) has the form of Eq. (IV.1) with Δ z ( z d 01 ) = κ Ω ( z d 01 ) \Delta z (z_d^{01})=\kappa\sqrt{\Omega(z_d^{01})} Δ z ( z d 01 ) = κ Ω ( z d 01 ) . For large z d 01 z_d^{01} z d 01 , Ω ( z d 01 ) = 1 2 κ ( z d 01 ) 2 \Omega(z_d^{01})=\frac{1}{2\kappa}(z_d^{01})^2 Ω ( z d 01 ) = 2 κ 1 ( z d 01 ) 2 and Eq. (IV.13) takes the form:
Ω D ( z m , z d 01 ) = 1 8 ( z m + 2 z d 01 ) 2 (63) \Omega^D\left(z_m,z_d^{01}\right)=\frac{1}{8}\left(z_m+\sqrt{2}z_d^{01}\right)^2 \tag*{(63)} Ω D ( z m , z d 01 ) = 8 1 ( z m + 2 z d 01 ) 2 (63)
with Δ z ( z d 01 ) = 2 z d 01 \Delta z (z_d^{01})=\sqrt{2}z_d^{01} Δ z ( z d 01 ) = 2 z d 01 . This solution shows that our Eq. (IV.1) is correct at least at large values of z d 01 z_d^{01} z d 01 .
Using solution of Eq. (IV.8) we can solve numerically Eq. (IV.2) and find function Δ z ( z d 01 ) \Delta z (z_d^{01}) Δ z ( z d 01 ) . In Figure 8 we show this solution: the curve labelled “exact” is the solution of Eq. (IV.2) with Eq. (IV.5) for the elastic scattering amplitude, while Ω a p p r o x ( z ) = ( z − 4 ) 2 8 + ln ( 1 2 z ) \Omega_{approx}(z)=\frac{(z-4)^2}{8}+\ln\left(\frac{1}{2}z\right) Ω a pp r o x ( z ) = 8 ( z − 4 ) 2 + ln ( 2 1 z ) is the approximate solution for the elastic amplitude taken from Ref. [54 ] .
Figure 8. Function Δ z ( z d 01 ) \Delta z (z_d^{01}) Δ z ( z d 01 ) versus z d 01 z_d^{01} z d 01 for small (Fig. 8-a) and large (Fig. 8-b) z d 01 z_d^{01} z d 01 .
From Figure 8 one can see that the approximate Ω \Omega Ω reproduces the function Δ z \Delta z Δ z quite well for z d 01 ≥ 4 z_d^{01}\geq4 z d 01 ≥ 4 , and for these values of z d 01 z_d^{01} z d 01 , Δ z ∝ z d 01 \Delta z\propto z_d^{01} Δ z ∝ z d 01 .
Using Eq. (IV.1) and the series in Eq. (B.1) for the elastic S S S -matrix we rewrite S D S^D S D in the form:5
S D ( z m , z d 01 ) = S ( z m + Δ z ( z d 01 ) ) = ∑ n = 0 ∞ ( − 1 ) n C n e 1 2 n 2 ( z m + Δ z ( z d 01 ) ) = ∑ n = 0 ∞ ( − 1 ) n n e 1 2 n 2 − 2 n e 1 2 n ( z m + Δ z ( z d 01 ) ) (64) S^D\left(z_m,z_d^{01}\right)=S\left(z_m+\Delta z\left(z_d^{01}\right)\right)=\sum_{n=0}^{\infty}(-1)^n C_n e^{\frac{1}{2}n^2}\left(z_m+\Delta z\left(z_d^{01}\right)\right)=\sum_{n=0}^{\infty}\frac{(-1)^n}{n}e^{\frac{1}{2}n^2-2n}e^{\frac{1}{2}n\left(z_m+\Delta z\left(z_d^{01}\right)\right)} \tag*{(64)} S D ( z m , z d 01 ) = S ( z m + Δ z ( z d 01 ) ) = n = 0 ∑ ∞ ( − 1 ) n C n e 2 1 n 2 ( z m + Δ z ( z d 01 ) ) = n = 0 ∑ ∞ n ( − 1 ) n e 2 1 n 2 − 2 n e 2 1 n ( z m + Δ z ( z d 01 ) ) (64)
with C n = 1 n exp ( 1 2 n 2 − 2 n ) C_n=\frac{1}{n}\exp\left(\frac{1}{2}n^2-2n\right) C n = n 1 exp ( 2 1 n 2 − 2 n ) . In Appendix B we explain why C n ∝ exp ( 1 2 n 2 ) C_n\propto\exp\left(\frac{1}{2}n^2\right) C n ∝ exp ( 2 1 n 2 ) . Other factors in Eq. (IV.15) have been discussed in Ref. [54 ] . They allow us to expand Eq. (IV.15) to the larger region of z z z ’s.
It is worth mentioning that the leading-twist BFKL kernel allows us to prove that the nonlinear equation for diffraction has a solution that shows geometric scaling behaviour with the variable z m + 2 κ Ω ( z d 01 ) z_m+2\sqrt{\kappa\Omega(z_d^{01})} z m + 2 κ Ω ( z d 01 ) . It is instructive to recall that for the BK equation we also got the proof for the scaling behaviour of the scattering amplitude using the leading-twist BFKL kernel (see Ref. [15 ] ).
Double diffraction for the leading-twist kernel
We need to introduce two kinds of dipoles in S S S and S ∗ S^* S ∗ (see Sec. III and Figure 3 ) to find ρ n , k D ( r 1 , b 1 , … , r n , b n ; … , r k , b k ; Y − Y 0 , Y 01 ) \rho^D_{n,k}(r_1,b_1,\ldots,r_n,b_n;\ldots,r_k,b_k;Y-Y_0,Y_{01}) ρ n , k D ( r 1 , b 1 , … , r n , b n ; … , r k , b k ; Y − Y 0 , Y 01 ) of Eq. (II.7). Bearing this in mind we rewrite S D S^D S D in the following form:
S D ( z m , z d 01 ) = ∑ n = 0 ∞ ( − 1 ) n n e 1 2 n 2 − 2 n 2 − n ∑ k = 0 n n ! k ! ( n − k ) ! e 1 2 k ( z m + Δ z ( z d 01 ) ) e 1 2 ( n − k ) ( z m + Δ z ( z d 01 ) ) (65) S^D(z_m,z_d^{01})=\sum_{n=0}^{\infty}\frac{(-1)^n}{n}e^{\frac{1}{2}n^2-2n^2-n}\sum_{k=0}^{n}\frac{n!}{k!(n-k)!}e^{\frac{1}{2}k\left(z_m+\Delta z\left(z_d^{01}\right)\right)}e^{\frac{1}{2}(n-k)\left(z_m+\Delta z\left(z_d^{01}\right)\right)} \tag*{(65)} S D ( z m , z d 01 ) = n = 0 ∑ ∞ n ( − 1 ) n e 2 1 n 2 − 2 n 2 − n k = 0 ∑ n k ! ( n − k )! n ! e 2 1 k ( z m + Δ z ( z d 01 ) ) e 2 1 ( n − k ) ( z m + Δ z ( z d 01 ) ) (65)
Using Eq. (IV.16) and Eq. (I.8) we can find ρ n , k D \rho^D_{n,k} ρ n , k D which take the form (see notation in Figure 3 -b):
ρ n , k D ( Y − Y 0 , Y 01 , r , b ; r 1 , b 1 , … , r n , b n ; … , r k , b k ) = C n + k 2 n − k ( n + k ) ! n ! k ! ∏ i = 1 n G P ( z m 1 + Δ z ( z i , d 01 ) ) ∏ j = 1 k G P ( z m 1 + Δ z ( z j , d 01 ) ) (66) \rho^D_{n,k}(Y-Y_0,Y_{01},r,b;r_1,b_1,\ldots,r_n,b_n;\ldots,r_k,b_k)=C_{n+k}2^{n-k}\frac{(n+k)!}{n!k!}\prod_{i=1}^{n}G_P\left(z_{m1}+\Delta z\left(z_{i,d}^{01}\right)\right)\prod_{j=1}^{k}G_P\left(z_{m1}+\Delta z\left(z_{j,d}^{01}\right)\right) \tag*{(66)} ρ n , k D ( Y − Y 0 , Y 01 , r , b ; r 1 , b 1 , … , r n , b n ; … , r k , b k ) = C n + k 2 n − k n ! k ! ( n + k )! i = 1 ∏ n G P ( z m 1 + Δ z ( z i , d 01 ) ) j = 1 ∏ k G P ( z m 1 + Δ z ( z j , d 01 ) ) (66)
Plugging Eq. (IV.17) in Eq. (II.7) we obtain:
S D D ( Y − Y 01 , Y 02 , Y 01 − Y 02 , r , r ′ ; b ) = ∑ n = 0 ∞ ( − 1 ) n ∑ k = 0 n n ! 2 k ! ( n − k ) ! C n 2 2 − 2 n G P n ( z m 1 + Δ z ( z d 01 ) + z m 2 + Δ z ( z d 02 ) ) (67) S^{DD}(Y-Y_{01},Y_{02},Y_{01}-Y_{02},r,r';b)=\sum_{n=0}^{\infty}(-1)^n\sum_{k=0}^{n}\frac{n!^2}{k!(n-k)!}C_n^2 2^{-2n}G_P^n\left(z_{m1}+\Delta z\left(z_d^{01}\right)+z_{m2}+\Delta z\left(z_d^{02}\right)\right) \tag*{(67)} S D D ( Y − Y 01 , Y 02 , Y 01 − Y 02 , r , r ′ ; b ) = n = 0 ∑ ∞ ( − 1 ) n k = 0 ∑ n k ! ( n − k )! n ! 2 C n 2 2 − 2 n G P n ( z m 1 + Δ z ( z d 01 ) + z m 2 + Δ z ( z d 02 ) ) (67)
As discussed above, Δ z ( z d 01 ) ∝ z d 01 \Delta z\left(z_d^{01}\right)\propto z_d^{01} Δ z ( z d 01 ) ∝ z d 01 ; hence we can rewrite Eq. (IV.18) in a more transparent form:
S D D ( Y − Y 01 , Y 02 , Y 01 − Y 02 , r , r ′ ; b ) = ∑ n = 0 ∞ ( − 1 ) n C n 2 2 − n n ! G P n ( z m 1 + z m 2 + Δ z ( z d 12 ) ) = ∑ n = 0 ∞ ( − 1 ) n n ! n 2 e n 2 − ( 4 + ln 2 ) n G P n ( z m 1 + z m 2 + Δ z ( z d 12 ) ) (68) \begin{aligned}
S^{DD}(Y-Y_{01},Y_{02},Y_{01}-Y_{02},r,r';b)
&=\sum_{n=0}^{\infty}(-1)^n C_n^2 2^{-n}n!G_P^n\left(z_{m1}+z_{m2}+\Delta z\left(z_d^{12}\right)\right)\\
&=\sum_{n=0}^{\infty}(-1)^n\frac{n!}{n^2}e^{n^2-(4+\ln2)n}G_P^n\left(z_{m1}+z_{m2}+\Delta z\left(z_d^{12}\right)\right) \tag*{(68)}
\end{aligned} S D D ( Y − Y 01 , Y 02 , Y 01 − Y 02 , r , r ′ ; b ) = n = 0 ∑ ∞ ( − 1 ) n C n 2 2 − n n ! G P n ( z m 1 + z m 2 + Δ z ( z d 12 ) ) = n = 0 ∑ ∞ ( − 1 ) n n 2 n ! e n 2 − ( 4 + l n 2 ) n G P n ( z m 1 + z m 2 + Δ z ( z d 12 ) ) (68)
Rewriting the sum of n n n as the counter integral as has been done in Eq. (B.8) and Eq. (B.9) and calculating the sum over poles using contour C ′ ′ C'' C ′′ (see Figure 7 ) we obtain:
Figure 7. The contours of integration in Eq. (B.3). The red points denote the poles in n n n for the right semi-plane. The black points show the poles in S d d ( z ) S_d^d(z) S d d ( z ) where S d d ( z ) S_d^d(z) S d d ( z ) has the double poles.
S D D ( z m 1 , z m 2 , z d 12 ) = ∮ C ′ ′ d n 2 π i π Γ ( 1 + n ) sin ( π n ) 1 n 2 e n 2 − ( 4 + ln 2 ) n e 1 2 n ( z m 1 + z m 2 + Δ z ( z d 12 ) ) (69) S^{DD}(z_{m1},z_{m2},z_d^{12})=\oint_{C''}\frac{dn}{2\pi i}\frac{\pi\Gamma(1+n)}{\sin(\pi n)}\frac{1}{n^2}e^{n^2-(4+\ln2)n}e^{\frac{1}{2}n\left(z_{m1}+z_{m2}+\Delta z\left(z_d^{12}\right)\right)} \tag*{(69)} S D D ( z m 1 , z m 2 , z d 12 ) = ∮ C ′′ 2 π i d n sin ( π n ) π Γ ( 1 + n ) n 2 1 e n 2 − ( 4 + l n 2 ) n e 2 1 n ( z m 1 + z m 2 + Δ z ( z d 12 ) ) (69)
Using Γ ( 1 − x ) Γ ( x ) = π sin ( π x ) \Gamma(1-x)\Gamma(x)=\frac{\pi}{\sin(\pi x)} Γ ( 1 − x ) Γ ( x ) = s i n ( π x ) π we reduce Eq. (IV.20) to the form:
S D D ( z m 1 , z m 2 , z d 12 ) = ∮ C ′ ′ d n 2 π i π 2 sin 2 ( π n ) Γ ( − n ) 1 n 2 e n 2 − ( 4 + ln 2 ) n e 1 2 n ( z m 1 + z m 2 + Δ z ( z d 12 ) ) = ∑ k = 1 ∞ 1 k 2 Γ ( k ) ( ψ ( 0 ) ( k ) − 2 k + 2 k − ( 4 + ln 2 ) + 1 2 ( z m 1 + z m 2 + Δ z ( z d 12 ) ) ) e k 2 + ( 4 + ln 2 ) k − 1 2 k ( z m 1 + z m 2 + Δ z ( z d 12 ) ) (70) \begin{aligned}
S^{DD}(z_{m1},z_{m2},z_d^{12})
&=\oint_{C''}\frac{dn}{2\pi i}\frac{\pi^2}{\sin^2(\pi n)\Gamma(-n)}\frac{1}{n^2}e^{n^2-(4+\ln2)n}e^{\frac{1}{2}n\left(z_{m1}+z_{m2}+\Delta z\left(z_d^{12}\right)\right)}\\
&=\sum_{k=1}^{\infty}\frac{1}{k^2\Gamma(k)}\left(\psi^{(0)}(k)-2k+\frac{2}{k}-(4+\ln2)+\frac{1}{2}\left(z_{m1}+z_{m2}+\Delta z\left(z_d^{12}\right)\right)\right)e^{k^2+(4+\ln2)k-\frac{1}{2}k\left(z_{m1}+z_{m2}+\Delta z\left(z_d^{12}\right)\right)} \tag*{(70)}
\end{aligned} S D D ( z m 1 , z m 2 , z d 12 ) = ∮ C ′′ 2 π i d n sin 2 ( π n ) Γ ( − n ) π 2 n 2 1 e n 2 − ( 4 + l n 2 ) n e 2 1 n ( z m 1 + z m 2 + Δ z ( z d 12 ) ) = k = 1 ∑ ∞ k 2 Γ ( k ) 1 ( ψ ( 0 ) ( k ) − 2 k + k 2 − ( 4 + ln 2 ) + 2 1 ( z m 1 + z m 2 + Δ z ( z d 12 ) ) ) e k 2 + ( 4 + l n 2 ) k − 2 1 k ( z m 1 + z m 2 + Δ z ( z d 12 ) ) (70)
Repeating all the steps of the calculation (see Eq. (B.8)) with the saddle point at large k k k : k S P = 1 4 ( z m 1 + z m 2 + Δ z ( z d 12 ) ) − 1 2 ( 4 + ln 2 ) k_{SP}=\frac{1}{4}\left(z_{m1}+z_{m2}+\Delta z\left(z_d^{12}\right)\right)-\frac{1}{2}(4+\ln2) k S P = 4 1 ( z m 1 + z m 2 + Δ z ( z d 12 ) ) − 2 1 ( 4 + ln 2 ) , we obtain the following answer:
S D D ( z m 1 + z m 2 + Δ z ( z d 12 ) ) = Φ ( z m 1 + z m 2 + Δ z ( z d 12 ) ) Γ ( z m 1 + z m 2 + Δ z ( z d 12 ) 4 − 1 2 ( 4 + ln 2 ) ) exp ( − ( z m 1 + z m 2 + Δ z ( z d 12 ) − 8 − 2 ln 2 ) 2 16 ) (71) S^{DD}\left(z_{m1}+z_{m2}+\Delta z\left(z_d^{12}\right)\right)=\frac{\Phi\left(z_{m1}+z_{m2}+\Delta z\left(z_d^{12}\right)\right)}{\Gamma\left(\frac{z_{m1}+z_{m2}+\Delta z\left(z_d^{12}\right)}{4}-\frac{1}{2}(4+\ln2)\right)}\exp\left(-\frac{\left(z_{m1}+z_{m2}+\Delta z\left(z_d^{12}\right)-8-2\ln2\right)^2}{16}\right) \tag*{(71)} S D D ( z m 1 + z m 2 + Δ z ( z d 12 ) ) = Γ ( 4 z m 1 + z m 2 + Δ z ( z d 12 ) − 2 1 ( 4 + ln 2 ) ) Φ ( z m 1 + z m 2 + Δ z ( z d 12 ) ) exp ( − 16 ( z m 1 + z m 2 + Δ z ( z d 12 ) − 8 − 2 ln 2 ) 2 ) (71)
where Φ \Phi Φ is a smooth function of n S P n_{SP} n S P . In Eq. (IV.22) this function is equal to Φ ( k S P ) − π k S P 2 ( ψ ( 0 ) ( k S P ) + 2 k S P ) \Phi(k_{SP})-\frac{\sqrt{\pi}}{k_{SP}^2}\left(\psi^{(0)}(k_{SP})+\frac{2}{k_{SP}}\right) Φ ( k S P ) − k S P 2 π ( ψ ( 0 ) ( k S P ) + k S P 2 ) .
It is easy to see that S D D S^{DD} S D D of Eq. (IV.22) satisfies the following initial and boundary conditions:
S D D ( z m 1 , z m 2 = 0 , z d 12 ) = S d D D ( z m 1 , z d 12 ) ; (72) S^{DD}\left(z_{m1},z_{m2}=0,z_d^{12}\right)=S_d^{DD}\left(z_{m1},z_d^{12}\right); \tag*{(72)} S D D ( z m 1 , z m 2 = 0 , z d 12 ) = S d D D ( z m 1 , z d 12 ) ; (72)
S D D ( z m 1 = 0 , z m 2 , z d 12 ) = S d D D ( z m 2 , z d 12 ) ; (73) S^{DD}\left(z_{m1}=0,z_{m2},z_d^{12}\right)=S_d^{DD}\left(z_{m2},z_d^{12}\right); \tag*{(73)} S D D ( z m 1 = 0 , z m 2 , z d 12 ) = S d D D ( z m 2 , z d 12 ) ; (73)
S D D ( z m 1 = 0 , z m 2 = 0 , z d 12 ) = ( S d ( z d 12 ) ) 2 ; (74) S^{DD}\left(z_{m1}=0,z_{m2}=0,z_d^{12}\right)=\left(S_d\left(z_d^{12}\right)\right)^2; \tag*{(74)} S D D ( z m 1 = 0 , z m 2 = 0 , z d 12 ) = ( S d ( z d 12 ) ) 2 ; (74)
Double diffraction for the full BFKL kernel
In the previous section we worked with the simplified leading-twist BFKL kernel. It is instructive to note, however, that Eq. (IV.1) and Eq. (IV.2) can be generalized to double diffraction for the regular BFKL kernel. Indeed,
S D D ( z m 1 + z m 2 + Δ z ( z d 12 ) ) = S d d ( z m 1 + z m 2 + Δ z ( z d 12 ) ) (75) S^{DD}\left(z_{m1}+z_{m2}+\Delta z\left(z_d^{12}\right)\right)=S_d^{d}\left(z_{m1}+z_{m2}+\Delta z\left(z_d^{12}\right)\right) \tag*{(75)} S D D ( z m 1 + z m 2 + Δ z ( z d 12 ) ) = S d d ( z m 1 + z m 2 + Δ z ( z d 12 ) ) (75)
where S d d S_d^{d} S d d is the scattering matrix for dipole-dipole interaction. This solution has the initial conditions of Eq. (IV.23a)-Eq. (IV.23c) as well as Eq. (IV.22) for the leading-twist BFKL kernel. Δ z ( z d 12 ) \Delta z\left(z_d^{12}\right) Δ z ( z d 12 ) is the solution of the following equation:
S ( Δ z ( z d 12 ) ) = S 2 ( z d 12 ) or Ω ( Δ z ( z d 12 ) ) = 2 Ω ( z d 12 ) (76) S\left(\Delta z\left(z_d^{12}\right)\right)=S^2\left(z_d^{12}\right)\quad\text{or}\quad\Omega\left(\Delta z\left(z_d^{12}\right)\right)=2\Omega\left(z_d^{12}\right) \tag*{(76)} S ( Δ z ( z d 12 ) ) = S 2 ( z d 12 ) or Ω ( Δ z ( z d 12 ) ) = 2Ω ( z d 12 ) (76)
This equation is the Schröder-type doubling equation. Unfortunately we cannot solve Eq. (IV.25) analytically in the entire kinematic range of the variable, but we found a numerical solution for the equation:
Ω ( U ( Y , x 01 ) , x 01 ) = 2 Ω ( Y , x 01 ) (77) \Omega\left(U(Y,x_{01}),x_{01}\right)=2\Omega(Y,x_{01}) \tag*{(77)} Ω ( U ( Y , x 01 ) , x 01 ) = 2Ω ( Y , x 01 ) (77)
which is plotted in Figure 9 . The numerical solution for U ( ( Y , x 01 ) , x 01 ) U((Y,x_{01}),x_{01}) U (( Y , x 01 ) , x 01 ) can be parameterize as
Figure 9. Function U ( Y , x 01 ) U(Y,x_{01}) U ( Y , x 01 ) at fixed ξ = ln ∣ x 01 ∣ 2 = − 8 \xi=\ln|x_{01}|^2=-8 ξ = ln ∣ x 01 ∣ 2 = − 8 (see Eq. (IV.26)). The parametrization is given by Eq. (IV.27). The initial conditions are given by McLerran–Venugopalan formula [12 ] with Q s ( Y = Y A ) = 1 Q_s(Y=Y_A)=1 Q s ( Y = Y A ) = 1 GeV.
U ( Y , x 01 ) = Y + Δ ~ I P + ( 2 − 1 ) w ln ( 1 + exp ( ( Y − Y s a t ( x 01 ) ) / w ) ) (78) U(Y,x_{01})=Y+\widetilde{\Delta}_{IP}+(\sqrt{2}-1)w\ln\left(1+\exp\left((Y-Y_{\mathrm{sat}}(x_{01}))/w\right)\right) \tag*{(78)} U ( Y , x 01 ) = Y + Δ I P + ( 2 − 1 ) w ln ( 1 + exp ( ( Y − Y sat ( x 01 )) / w ) ) (78)
where ln Q s 2 ( Y s a t ) = ln ( 1 / x 01 2 ) \ln Q_s^2(Y_{\mathrm{sat}})=\ln(1/x_{01}^2) ln Q s 2 ( Y sat ) = ln ( 1/ x 01 2 ) , Δ ~ I P = ( ln 2 ) / Δ I P \widetilde{\Delta}_{IP}=(\ln2)/\Delta_{IP} Δ I P = ( ln 2 ) / Δ I P and w = 3.99 w=3.99 w = 3.99 is the only parameter which was fitted to the data. From Figure 9 one can see that Eq. (IV.27) describe the solution with accuracy about 11%. The main lesson from this equation is that the solution reproduces the large Y Y Y limit for Y ≥ Y s a t Y\geq Y_{\mathrm{sat}} Y ≥ Y sat .
It is instructive to notice, that σ D = − ∂ S D / ∂ Y 01 \sigma^D=-\partial S^D/\partial Y_{01} σ D = − ∂ S D / ∂ Y 01 (see Eq. (III.6)) shows the same main features as in the one dimensional model (see Figure 4 ): each curve is a smooth bump. The peak sits at Y 01 ≈ 9.3 Y_{01}\approx9.3 Y 01 ≈ 9.3 , 8.3, 7.9 for Y m = 0 Y_m=0 Y m = 0 , 1, 1.56 at α S = 0.2 \alpha_S = 0.2 α S = 0.2 , so a larger Y m Y_m Y m moves the peak to smaller Y 01 Y_{01} Y 01 (see Figure 10 -a). The value of σ D \sigma^D σ D at the peaks are the same in contrast with the one dimensional model (see Figure 4 ). It worthwhile mentioning that for the parameterization of Eq. (IV.27) the values of σ D \sigma^D σ D in the peaks changes.
(a) Fig. 10-a
(b) Fig. 10-b
Figure 10. σ D ( Y ~ m , Y ~ 01 ) = − ∂ S D ( Y ~ m , Y ~ 01 ) ∂ Y ~ 01 \sigma^D(\widetilde{Y}_m,\widetilde{Y}_{01}) = -\frac{\partial S^D(\widetilde{Y}_m,\widetilde{Y}_{01})}{\partial\widetilde{Y}_{01}} σ D ( Y m , Y 01 ) = − ∂ Y 01 ∂ S D ( Y m , Y 01 ) versus Y ~ 01 \widetilde{Y}_{01} Y 01 at fixed Y ~ m \widetilde{Y}_m Y m and fixed ξ = ln ∣ x 01 ∣ 2 = − 4 \xi= \ln\lvert x_{01}\rvert^2 = -4 ξ = ln ∣ x 01 ∣ 2 = − 4 . Fig. 10-a: Numerical solution to the KL equation. Fig. 10-b: Numerical solution to the KL equation with U ( Y ~ , ξ ) U(\widetilde{Y},\xi) U ( Y , ξ ) for parameterization of Eq. (IV.27). The initial conditions are given by McLerran–Venugopalan formula [12 ] with Q s ( Y = Y A ) = 1 Q_s(Y = Y_A) = 1 Q s ( Y = Y A ) = 1 GeV. The rescaled rapidities Y ~ m = Δ P Y m \widetilde{Y}_m = \Delta_{\mathbb{P}}Y_m Y m = Δ P Y m (Y ~ 01 = Δ P Y 01 \widetilde{Y}_{01} = \Delta_{\mathbb{P}}Y_{01} Y 01 = Δ P Y 01 ) are analogous to those defined in Section III for the one-dimensional model.
However, we are able to find the analytic solution in the region where z m 1 z_{m1} z m 1 , z m 2 z_{m2} z m 2 and z d 12 z_d^{12} z d 12 are large. In this region this equation reduces [15 ] to the simple linear evolution equation:
∂ S D ( x 1 , x 0 , Y m , Y 01 ) ∂ Y m = − α ˉ S 2 π ∫ d 2 x 02 x 01 2 x 02 2 x 12 2 S D ( x 1 , x 0 , Y m , Y 01 ) (79) \frac{\partial S^D(x_1,x_0,Y_m,Y_{01})}{\partial Y_m} = -\frac{\bar{\alpha}_S}{2\pi}\int d^2x_{02}\,\frac{x_{01}^2}{x_{02}^2x_{12}^2}S^D(x_1,x_0,Y_m,Y_{01}) \tag*{(79)} ∂ Y m ∂ S D ( x 1 , x 0 , Y m , Y 01 ) = − 2 π α ˉ S ∫ d 2 x 02 x 02 2 x 12 2 x 01 2 S D ( x 1 , x 0 , Y m , Y 01 ) (79)
In the saturation region where we have the geometric scaling behaviour [30 , 69 ] , Eq. (IV.28) takes the simple form:
κ d S D ( z m , z d 01 ) d z m = − z m S D ( z m , z d 01 ) (80) \kappa\frac{dS^D(z_m,z_d^{01})}{dz_m}=-z_m S^D(z_m,z_d^{01}) \tag*{(80)} κ d z m d S D ( z m , z d 01 ) = − z m S D ( z m , z d 01 ) (80)
where r r r and R R R are the sizes of incoming dipoles and the dipoles inside of the nucleus. The variables z m z_m z m and z d 01 z_d^{01} z d 01 have been defined in Eq. (IV.3). For Ω ( z m , z d 01 ) \Omega(z_m,z_d^{01}) Ω ( z m , z d 01 ) Eq. (IV.29) takes the form:
κ d 2 Ω D ( z m , z d 01 ) d z m 2 = 1 (81) \kappa\frac{d^2\Omega^D(z_m,z_d^{01})}{dz_m^2}=1 \tag*{(81)} κ d z m 2 d 2 Ω D ( z m , z d 01 ) = 1 (81)
The general solution of this equation is
Ω D ( z m , z d 01 ) = 1 κ ( 1 2 z m 2 + f ( z d 01 ) z m + p ( z d 01 ) ) (82) \Omega^D(z_m,z_d^{01})=\frac{1}{\kappa}\left(\frac{1}{2}z_m^2+f(z_d^{01})z_m+p(z_d^{01})\right) \tag*{(82)} Ω D ( z m , z d 01 ) = κ 1 ( 2 1 z m 2 + f ( z d 01 ) z m + p ( z d 01 ) ) (82)
Functions f f f and p p p have to be found from the initial conditions: Ω D ( z m = 0 , z d 01 ) = 2 Ω ( z d 01 ) \Omega^D(z_m=0,z_d^{01})=2\Omega(z_d^{01}) Ω D ( z m = 0 , z d 01 ) = 2Ω ( z d 01 ) , where Ω \Omega Ω describes the elastic scattering. One can see that
f ( z d 01 ) = 2 κ Ω ( z d 01 ) ; p ( z d 01 ) = 2 κ Ω ( z d 01 ) ; (83) f(z_d^{01})=2\sqrt{\kappa\Omega(z_d^{01})};\qquad p(z_d^{01})=2\kappa\Omega(z_d^{01}); \tag*{(83)} f ( z d 01 ) = 2 κ Ω ( z d 01 ) ; p ( z d 01 ) = 2 κ Ω ( z d 01 ) ; (83)
leads to
Ω D ( z m , z d 01 ) = Ω D ( z m + 2 κ Ω ( z d 01 ) ) = 1 2 κ ( z m + 2 κ Ω ( z d 01 ) ) 2 (84) \Omega^D(z_m,z_d^{01})=\Omega^D\left(z_m+2\sqrt{\kappa\Omega(z_d^{01})}\right)=\frac{1}{2\kappa}\left(z_m+2\sqrt{\kappa\Omega(z_d^{01})}\right)^2 \tag*{(84)} Ω D ( z m , z d 01 ) = Ω D ( z m + 2 κ Ω ( z d 01 ) ) = 2 κ 1 ( z m + 2 κ Ω ( z d 01 ) ) 2 (84)
This solution satisfies the initial condition for Ω D \Omega^{D} Ω D
For large z d 01 z_{d}^{01} z d 01 , Ω ( z d 01 ) = 1 2 κ ( z d 01 ) 2 \Omega\left(z_{d}^{01}\right)=\frac{1}{2\kappa}\left(z_{d}^{01}\right)^{2} Ω ( z d 01 ) = 2 κ 1 ( z d 01 ) 2 reducing Eq. (IV.33) to the form:
S D ( z m , z d 01 ) = H ( ( z m + 2 z d 01 ) 2 2 κ ) exp ( − ( z m + 2 z d 01 ) 2 2 κ ) with z m = α ˉ S κ Y m + ξ r , R (85) S^{D}\left(z_{m},z_{d}^{01}\right)=H\left(\frac{\left(z_{m}+\sqrt{2}z_{d}^{01}\right)^{2}}{2\kappa}\right)\exp\left(-\frac{\left(z_{m}+\sqrt{2}z_{d}^{01}\right)^{2}}{2\kappa}\right)
\quad\text{with}\quad z_{m}=\bar{\alpha}_{S}\kappa Y_{m}+\xi_{r,R} \tag*{(85)} S D ( z m , z d 01 ) = H ( 2 κ ( z m + 2 z d 01 ) 2 ) exp ( − 2 κ ( z m + 2 z d 01 ) 2 ) with z m = α ˉ S κ Y m + ξ r , R (85)
where H H H is a smooth function which leads to a small correction to Eq. (IV.29). In Eq. (IV.34) we use the initial condition of Eq. (II.4) and assume that z d 01 z_{d}^{01} z d 01 is large and we have the geometric scaling behaviour for the elastic S S S . We have thus shown that at large values of z z z we obtain the same solution as for the leading-twist BFKL kernel. The double diffraction therefore takes the form:
S D D ( z m 1 + z m 2 + Δ z ( z d 12 ) ) = H ( z m 1 + z m 2 + Δ z ( z d 12 ) ) exp ( − ( z m 1 + z m 2 + Δ z ( z d 12 ) − 8 − 2 ln 2 ) 2 4 κ ) (86) S^{DD}\left(z_{m1}+z_{m2}+\Delta z\left(z_{d}^{12}\right)\right)=H\left(z_{m1}+z_{m2}+\Delta z\left(z_{d}^{12}\right)\right)\exp\left(-\frac{\left(z_{m1}+z_{m2}+\Delta z\left(z_{d}^{12}\right)-8-2\ln2\right)^{2}}{4\kappa}\right) \tag*{(86)} S D D ( z m 1 + z m 2 + Δ z ( z d 12 ) ) = H ( z m 1 + z m 2 + Δ z ( z d 12 ) ) exp ( − 4 κ ( z m 1 + z m 2 + Δ z ( z d 12 ) − 8 − 2 ln 2 ) 2 ) (86)
with Δ z ( z d 12 ) = 2 z d 12 \Delta z\left(z_{d}^{12}\right)=\sqrt{2}z_{d}^{12} Δ z ( z d 12 ) = 2 z d 12 . This solution is shown in Figure 9 by dotted line.
Possible relation to the ``rare" fluctuations approach
In this section, we present our results in a form close to that of the "rare" fluctuation approach [27 ] . This allows us to discuss how the two approaches relate to each other.
Elastic dipole-dipole scattering
Let us consider the scattering of a dipole r r r at rapidity Y Y Y with the dipole r d r_{d} r d at rapidity Y 0 Y_{0} Y 0 in the BK approach. This amplitude satisfies the BK equation and can be written as the following series [16 ] .7
S ( Y − Y 0 , r , b , r d , b d ) = ∑ n = 0 ∞ ( − 1 ) n ∫ ∏ i = 1 n d 2 r i d 2 b i γ B A ( r i , b i , r d , b d ) ρ n ( Y − Y 0 , r , b ; { r i , b i } ) = ∑ n = 0 ∞ ( − 1 ) n C n e n γ ˉ z d (87) S\left(Y-Y_{0},r,b,r_{d},b_{d}\right)=\sum_{n=0}^{\infty}(-1)^{n}\int\prod_{i=1}^{n}d^{2}r_{i}\,d^{2}b_{i}\,\gamma^{BA}\left(r_{i},b_{i},r_{d},b_{d}\right)\rho_{n}\left(Y-Y_{0},r,b;\{r_{i},b_{i}\}\right)=\sum_{n=0}^{\infty}(-1)^{n}C_{n}e^{n\bar{\gamma}z_{d}} \tag*{(87)} S ( Y − Y 0 , r , b , r d , b d ) = n = 0 ∑ ∞ ( − 1 ) n ∫ i = 1 ∏ n d 2 r i d 2 b i γ B A ( r i , b i , r d , b d ) ρ n ( Y − Y 0 , r , b ; { r i , b i } ) = n = 0 ∑ ∞ ( − 1 ) n C n e n γ ˉ z d (87)
where z d = α ˉ S κ ( Y − Y 0 ) + ξ r , r d z_{d}=\bar{\alpha}_{S}\kappa\left(Y-Y_{0}\right)+\xi_{r,r_{d}} z d = α ˉ S κ ( Y − Y 0 ) + ξ r , r d . In the second equation we use the features of the BFKL Pomeron in the saturation region that has been discussed in Eq. (II.8). For the geometric scaling solution the coefficients C n C_{n} C n can be found as follows:
( − 1 ) n C n = ∫ ϵ − i ∞ ϵ + i ∞ d γ ˉ z d 2 π i e − n γ ˉ z d S ( z d ) (88) (-1)^{n}C_{n}=\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\bar{\gamma}z_{d}}{2\pi i}e^{-n\bar{\gamma}z_{d}}S\left(z_{d}\right) \tag*{(88)} ( − 1 ) n C n = ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d γ ˉ z d e − n γ ˉ z d S ( z d ) (88)
For the scattering amplitude of the dipole r d r_{d} r d at rapidity Y 0 Y_{0} Y 0 with the dipole r ′ r^{\prime} r ′ at rest we have Eq. (IV.37) in the form:
( − 1 ) n C n = ∫ ϵ − i ∞ ϵ + i ∞ d γ ˉ z ˉ d 2 π i e − n γ ˉ z ˉ d S ( z ˉ d ) (89) (-1)^{n}C_{n}=\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\bar{\gamma}\bar{z}_{d}}{2\pi i}e^{-n\bar{\gamma}\bar{z}_{d}}S\left(\bar{z}_{d}\right) \tag*{(89)} ( − 1 ) n C n = ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d γ ˉ z ˉ d e − n γ ˉ z ˉ d S ( z ˉ d ) (89)
with z ˉ d = α ˉ S κ Y 0 + ξ r d , r ′ \bar{z}_{d}=\bar{\alpha}_{S}\kappa Y_{0}+\xi_{r_{d},r^{\prime}} z ˉ d = α ˉ S κ Y 0 + ξ r d , r ′ .
Using these equations we can rewrite Eq. (B.7) for S d d S_{d}^{d} S d d as follows:
S d d ( z d 12 ) = ∑ n = 0 ∞ ( − 1 ) n C n 2 Γ ( n + 1 ) e n γ ˉ z d 12 = ∑ n = 0 ∞ ( − 1 ) n n ! ∫ ϵ − i ∞ ϵ + i ∞ d γ ˉ z d 2 π i ∫ ϵ − i ∞ ϵ + i ∞ d γ ˉ z ˉ d 2 π i e n γ ˉ ( z d 12 − z d − z ˉ d ) S ( z d ) S ( z ˉ d ) (90) S_{d}^{d}\left(z_{d}^{12}\right)=\sum_{n=0}^{\infty}(-1)^{n}C_{n}^{2}\Gamma(n+1)e^{n\bar{\gamma}z_{d}^{12}}=\sum_{n=0}^{\infty}(-1)^{n}n!\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\bar{\gamma}z_{d}}{2\pi i}\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\bar{\gamma}\bar{z}_{d}}{2\pi i}e^{n\bar{\gamma}\left(z_{d}^{12}-z_{d}-\bar{z}_{d}\right)}S\left(z_{d}\right)S\left(\bar{z}_{d}\right) \tag*{(90)} S d d ( z d 12 ) = n = 0 ∑ ∞ ( − 1 ) n C n 2 Γ ( n + 1 ) e n γ ˉ z d 12 = n = 0 ∑ ∞ ( − 1 ) n n ! ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d γ ˉ z d ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d γ ˉ z ˉ d e n γ ˉ ( z d 12 − z d − z ˉ d ) S ( z d ) S ( z ˉ d ) (90)
First, let us neglect the factor ( − 1 ) n n ! (-1)^n n! ( − 1 ) n n ! in Eq. (IV.39). This simplification gives:
S d d ( z d 12 ) = ∑ n = 0 ∞ ∫ ϵ − i ∞ ϵ + i ∞ d γ ˉ 2 π i ∫ ϵ − i ∞ ϵ + i ∞ d z ˉ d 2 π i e n γ ˉ ( z d 12 − z d − z ˉ d ) S ( z d ) S ( z ˉ d ) → ∫ ϵ − i ∞ ϵ + i ∞ d z ˉ d 2 π i S ( z d 12 − z ˉ d ) S ( z ˉ d ) (91) S_d^d\left(z_d^{12}\right)=\sum_{n=0}^{\infty}\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\bar{\gamma}}{2\pi i}\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\bar{z}_d}{2\pi i}e^{n\bar{\gamma}\left(z_d^{12}-z_d-\bar{z}_d\right)}S(z_d)S(\bar{z}_d)\rightarrow\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\bar{z}_d}{2\pi i}S\left(z_d^{12}-\bar{z}_d\right)S(\bar{z}_d) \tag*{(91)} S d d ( z d 12 ) = n = 0 ∑ ∞ ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d γ ˉ ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d z ˉ d e n γ ˉ ( z d 12 − z d − z ˉ d ) S ( z d ) S ( z ˉ d ) → ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d z ˉ d S ( z d 12 − z ˉ d ) S ( z ˉ d ) (91)
We can easily take into account ( − 1 ) n n ! (-1)^n n! ( − 1 ) n n ! considering Eq. (III.20) and the definition of the Borel transform [55 ] (see also Appendix § 4.3 ):
S ( e γ ˉ z ) = ∫ 0 ∞ d t e − t B S ( − t e γ ˉ z ) (92) S\left(e^{\bar{\gamma}z}\right)=\int_0^\infty dt\,e^{-t}\mathcal{B}S\left(-te^{\bar{\gamma}z}\right) \tag*{(92)} S ( e γ ˉ z ) = ∫ 0 ∞ d t e − t B S ( − t e γ ˉ z ) (92)
Using this equation and Eq. (IV.40) we obtain for the scattering matrix our master equation:
S d d ( z d 12 ) = ∫ 0 ∞ d t e − t 1 γ ˉ ∫ ϵ − i ∞ ϵ + i ∞ d z ˉ d 2 π i B S ( ln ( − t ) γ ˉ + z d 12 − z ˉ d ) ⏟ BK equation S ( z ˉ d ) ⏟ BK equation (93) S_d^d\left(z_d^{12}\right)=\int_0^\infty dt\,e^{-t}\frac{1}{\bar{\gamma}}\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\bar{z}_d}{2\pi i}\underbrace{\mathcal{B}S\left(\frac{\ln(-t)}{\bar{\gamma}}+z_d^{12}-\bar{z}_d\right)}_{\text{BK equation}}\underbrace{S(\bar{z}_d)}_{\text{BK equation}} \tag*{(93)} S d d ( z d 12 ) = ∫ 0 ∞ d t e − t γ ˉ 1 ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d z ˉ d BK equation B S ( γ ˉ ln ( − t ) + z d 12 − z ˉ d ) BK equation S ( z ˉ d ) (93)
First, we check how Eq. (B.9) appears from this equation. In the saturation regions for both B S ( ln ( − t ) / γ ˉ + z d 12 − z ˉ d ) \mathcal{B}S\left(\ln(-t)/\bar{\gamma}+z_d^{12}-\bar{z}_d\right) B S ( ln ( − t ) / γ ˉ + z d 12 − z ˉ d ) and S ( z ˉ d ) S(\bar{z}_d) S ( z ˉ d ) we reduce the master equation to the following integral:
S d d ( z d 12 ) = ∫ 0 ∞ d t e − t 1 γ ˉ ∫ − ∞ ∞ d z ˉ d 2 π exp ( − ( ln ( − t ) / γ ˉ + z d 12 − z ˉ d ) 2 + z ˉ d 2 2 κ ) (94) S_d^d\left(z_d^{12}\right)=\int_0^\infty dt\,e^{-t}\frac{1}{\bar{\gamma}}\int_{-\infty}^{\infty}\frac{d\bar{z}_d}{2\pi}\exp\left(-\frac{\left(\ln(-t)/\bar{\gamma}+z_d^{12}-\bar{z}_d\right)^2+\bar{z}_d^2}{2\kappa}\right) \tag*{(94)} S d d ( z d 12 ) = ∫ 0 ∞ d t e − t γ ˉ 1 ∫ − ∞ ∞ 2 π d z ˉ d exp ( − 2 κ ( ln ( − t ) / γ ˉ + z d 12 − z ˉ d ) 2 + z ˉ d 2 ) (94)
Taking the integral over z ˉ d \bar{z}_d z ˉ d using the method of steepest descent with the saddle point z ˉ d , S P = 1 2 ( ln ( − t ) / γ ˉ + z d 12 ) \bar{z}_{d,SP}=\frac{1}{2}\left(\ln(-t)/\bar{\gamma}+z_d^{12}\right) z ˉ d , S P = 2 1 ( ln ( − t ) / γ ˉ + z d 12 ) we reproduce Eq. (B.9):
S d d ( z d 12 ) = κ 2 π γ ˉ ∫ 0 ∞ d t e − t exp ( − ( ln ( − t ) / γ ˉ + z d 12 ) 2 4 κ ) → z d 12 ≫ 1 H ( z d 12 2 κ γ ˉ ) Γ ( 1 − z d 12 2 κ γ ˉ ) exp ( − ( z d 12 ) 2 4 κ ) (95) S_d^d\left(z_d^{12}\right)=\frac{\sqrt{\kappa}}{2\sqrt{\pi}\bar{\gamma}}\int_0^\infty dt\,e^{-t}\exp\left(-\frac{\left(\ln(-t)/\bar{\gamma}+z_d^{12}\right)^2}{4\kappa}\right)\xrightarrow[z_d^{12}\gg1]{}H\left(\frac{z_d^{12}}{2\kappa\bar{\gamma}}\right)\Gamma\left(1-\frac{z_d^{12}}{2\kappa\bar{\gamma}}\right)\exp\left(-\frac{\left(z_d^{12}\right)^2}{4\kappa}\right) \tag*{(95)} S d d ( z d 12 ) = 2 π γ ˉ κ ∫ 0 ∞ d t e − t exp ( − 4 κ ( ln ( − t ) / γ ˉ + z d 12 ) 2 ) z d 12 ≫ 1 H ( 2 κ γ ˉ z d 12 ) Γ ( 1 − 2 κ γ ˉ z d 12 ) exp ( − 4 κ ( z d 12 ) 2 ) (95)
where H H H is a smooth function of its argument. More detailed calculations can be found in Appendix § 4.3 .
These estimates can be interpreted in a way that is close to the “rare” fluctuation estimates of Ref. [27 ] . Indeed, Eq. (IV.42) has the physical meaning of a “rare” fluctuation of size Y Y Y : it is the product of two probabilities, the probability of a fluctuation of size Y − Y 0 Y-Y_0 Y − Y 0 in the BK equation for the upper amplitude, and the probability of a small fluctuation of size Y 0 Y_0 Y 0 for the lower amplitude. The equation itself is the sum over all such possible small fluctuations, and Eq. (IV.44) reproduces the same result as Ref. [27 ] . In this discussion we have used the interpretation of the BK solution at large rapidities [15 ] as a “rare” fluctuation, following Ref. [27 ] . We believe that Eq. (IV.42), together with its connection to the “rare” fluctuation approach, will allow us to estimate the range of validity of both approaches, since summing the large Pomeron loops can only be trusted for the restricted range of rapidities Y ≤ 1 / α ˉ s 4 Y\leq1/\bar{\alpha}_s^4 Y ≤ 1/ α ˉ s 4 [73 ] .
Expressing the double diffraction in terms of the single ones
Generalizing the master equation of Eq. (IV.42) to double diffraction is straightforward. Indeed, Eq. (IV.37) and Eq. (IV.38) can be rewritten in terms of S D S^D S D as follows:8
( − 1 ) n C n = ∫ ϵ − i ∞ ϵ + i ∞ d γ ˉ 2 π i e − n γ ˉ z 01 D S D ( z ˉ 01 D ) ⏟ KL equation ; ( − 1 ) n C n = ∫ ϵ − i ∞ ϵ + i ∞ d γ ˉ 2 π i e − n γ ˉ z 02 D S D ( z ˉ 02 D ) ⏟ KL equation ; (96) (-1)^n C_n=\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\bar{\gamma}}{2\pi i}e^{-n\bar{\gamma}z_{01}^D}\underbrace{S^D(\bar{z}_{01}^D)}_{\text{KL equation}} \ ;\qquad(-1)^n C_n=\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\bar{\gamma}}{2\pi i}e^{-n\bar{\gamma}z_{02}^D}\underbrace{S^D(\bar{z}_{02}^D)}_{\text{KL equation}}\ ; \tag*{(96)} ( − 1 ) n C n = ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d γ ˉ e − n γ ˉ z 01 D KL equation S D ( z ˉ 01 D ) ; ( − 1 ) n C n = ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d γ ˉ e − n γ ˉ z 02 D KL equation S D ( z ˉ 02 D ) ; (96)
where z 01 D = z m 1 + Δ z ( z d 01 ) z^{D}_{01}=z_{m1}+\Delta z(z^{01}_{d}) z 01 D = z m 1 + Δ z ( z d 01 ) and z ˉ 02 D = z m 2 + Δ z ( z ˉ d 02 ) \bar{z}^{D}_{02}=z_{m2}+\Delta z(\bar{z}^{02}_{d}) z ˉ 02 D = z m 2 + Δ z ( z ˉ d 02 ) . Using these equations, we can rewrite Eq. (IV.19) for double diffraction in the form:
S D D ( z m 1 , z m 2 , z d 12 ) = ∑ n = 0 ∞ ( − 1 ) n 2 − n n ! ∫ ϵ − i ∞ ϵ + i ∞ d γ z 01 D 2 π i ∫ ϵ − i ∞ ϵ + i ∞ d γ ˉ z ˉ 02 D 2 π i e n γ ˉ ( z 12 D − z 01 D − z ˉ 02 D ) S D ( z 01 D ) S D ( z ˉ 02 D ) (97) S^{DD}(z_{m1},z_{m2},z^{12}_{d})=\sum_{n=0}^{\infty}(-1)^n2^{-n}n!\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\gamma\,z^{D}_{01}}{2\pi i}\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\bar{\gamma}\,\bar{z}^{D}_{02}}{2\pi i}e^{n\bar{\gamma}(z^{D}_{12}-z^{D}_{01}-\bar{z}^{D}_{02})}S^{D}(z^{D}_{01})S^{D}(\bar{z}^{D}_{02}) \tag*{(97)} S D D ( z m 1 , z m 2 , z d 12 ) = n = 0 ∑ ∞ ( − 1 ) n 2 − n n ! ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d γ z 01 D ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d γ ˉ z ˉ 02 D e n γ ˉ ( z 12 D − z 01 D − z ˉ 02 D ) S D ( z 01 D ) S D ( z ˉ 02 D ) (97)
where z d 12 = z m 1 + z m 2 + Δ z ( z d 12 ) z^{12}_{d}=z_{m1}+z_{m2}+\Delta z(z^{12}_{d}) z d 12 = z m 1 + z m 2 + Δ z ( z d 12 ) . Neglecting the factor ( − 1 ) n 2 − n n ! (-1)^n2^{-n}n! ( − 1 ) n 2 − n n ! and summing over n n n in Eq. (IV.46), we obtain:
S D D ( z 12 D ) = ∫ ϵ − i ∞ ϵ + i ∞ d z ˉ 02 D 2 π i S D ( z 12 D − z ˉ 02 D ) S D ( z ˉ 02 D ) (98) S^{DD}(z^{D}_{12})=\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\bar{z}^{D}_{02}}{2\pi i}S^{D}(z^{D}_{12}-\bar{z}^{D}_{02})S^{D}(\bar{z}^{D}_{02}) \tag*{(98)} S D D ( z 12 D ) = ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d z ˉ 02 D S D ( z 12 D − z ˉ 02 D ) S D ( z ˉ 02 D ) (98)
As discussed in the previous section, we need to perform the Borel transform for one of the S D S^{D} S D in Eq. (IV.47) (see Eq. (IV.41)). Introducing
S D ( e γ ˉ z 01 D ) = ∫ 0 ∞ d t e − t B S D ( − t e γ ˉ z 01 D ) (99) S^{D}\left(e^{\bar{\gamma}z^{D}_{01}}\right)=\int_{0}^{\infty}dt\,e^{-t}\mathcal{B}S^{D}\left(-te^{\bar{\gamma}z^{D}_{01}}\right) \tag*{(99)} S D ( e γ ˉ z 01 D ) = ∫ 0 ∞ d t e − t B S D ( − t e γ ˉ z 01 D ) (99)
we can rewrite Eq. (IV.47) in the form:
S D D ( z 12 D ) = ∫ 0 ∞ d t e − t 1 γ ˉ ∫ ϵ − i ∞ ϵ + i ∞ d z ˉ 02 D 2 π i B S D ( ln ( − t / 2 ) / γ ˉ + z 12 D − z ˉ 02 D ) ⏟ KL equation S D ( z ˉ 02 D ) ⏟ KL equation (100) S^{DD}(z^{D}_{12})=\int_{0}^{\infty}dt\,e^{-t}\frac{1}{\bar{\gamma}}\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\bar{z}^{D}_{02}}{2\pi i}\underbrace{\mathcal{B}S^{D}\left(\ln(-t/2)/\bar{\gamma}+z^{D}_{12}-\bar{z}^{D}_{02}\right)}_{\text{KL equation}}\underbrace{S^{D}(\bar{z}^{D}_{02})}_{\text{KL equation}} \tag*{(100)} S D D ( z 12 D ) = ∫ 0 ∞ d t e − t γ ˉ 1 ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d z ˉ 02 D KL equation B S D ( ln ( − t /2 ) / γ ˉ + z 12 D − z ˉ 02 D ) KL equation S D ( z ˉ 02 D ) (100)
This equation shows that for double diffraction we obtain an expression very similar to Eq. (IV.42), with S → S D S\to S^{D} S → S D and z d 01 → z 01 D z^{01}_{d}\to z^{D}_{01} z d 01 → z 01 D (z ˉ d 02 → z ˉ 02 D \bar{z}_{d}^{02}\to\bar{z}^{D}_{02} z ˉ d 02 → z ˉ 02 D ).
Conclusions
This paper presents a first attempt to investigate double diffractive production in dipole-dipole scattering within high energy QCD. Our approach is based on summing large BFKL Pomeron loops, which are responsible for producing two clusters of hadrons separated by a rapidity gap. Notably, in our framework both clusters can carry large masses — something that is not possible within the Balitsky-Kovchegov (BK) approach.
To sum the large loops, we utilize the dipole density extracted from the solution to the single diffraction equation [2 ] (see Eq. (II.1)), along with the simple nuclear model given by Eq. (I.8).
We develop two model approaches to double diffractive production: (i) a one dimensional model, and (ii) QCD with a simplified leading-twist BFKL kernel. These two models clarify the main features of double diffractive production and yield analytical results across the entire kinematic region of produced masses.
Our first main result is that the single diffraction cross section σ D \sigma^{D} σ D exhibits geometric scaling: it depends only on the single variable z m + 2 κ Ω ( z d 01 ) z_m+2\sqrt{\kappa\Omega(z^{01}_{d})} z m + 2 κ Ω ( z d 01 ) (see Eq. (IV.12) and Eq. (IV.13)). The double diffraction cross section σ D D \sigma^{DD} σ D D shows the same behaviour, depending only on the combination z m 1 + z m 2 + Δ z ( z d 12 ) z_{m1}+z_{m2}+\Delta z(z^{12}_{d}) z m 1 + z m 2 + Δ z ( z d 12 ) , where Δ z ( z d 12 ) = 2 z d 12 \Delta z(z^{12}_{d})=\sqrt{2}z^{12}_{d} Δ z ( z d 12 ) = 2 z d 12 for large z d 12 z^{12}_{d} z d 12 (see Eq. (IV.35)).
Our second main result is the following expression for S D D S^{DD} S D D :
S D D ( z m 1 + z m 2 + Δ z ( z d 12 ) ) = Φ ( z m 1 + z m 2 + Δ z ( z d 12 ) ) exp ( − ( z m 1 + z m 2 + Δ z ( z d 12 ) − 8 − 2 ln 2 ) 2 4 κ ) (101) S^{DD}\left(z_{m1}+z_{m2}+\Delta z(z^{12}_{d})\right)=\Phi\left(z_{m1}+z_{m2}+\Delta z(z^{12}_{d})\right)\exp\left(-\frac{\left(z_{m1}+z_{m2}+\Delta z(z^{12}_{d})-8-2\ln2\right)^2}{4\kappa}\right) \tag*{(101)} S D D ( z m 1 + z m 2 + Δ z ( z d 12 ) ) = Φ ( z m 1 + z m 2 + Δ z ( z d 12 ) ) exp ( − 4 κ ( z m 1 + z m 2 + Δ z ( z d 12 ) − 8 − 2 ln 2 ) 2 ) (101)
where Δ z ( z d 12 ) = 2 κ Ω ( z d 12 ) \Delta z(z^{12}_{d})=2\sqrt{\kappa\Omega(z^{12}_{d})} Δ z ( z d 12 ) = 2 κ Ω ( z d 12 ) and Φ \Phi Φ is a smooth function of its argument.
For the regular BFKL kernel, we find that the cross section is given by Eq. (V.1) with Δ z ( z d 12 ) = 2 z d 12 \Delta z(z^{12}_{d})=\sqrt{2}z^{12}_{d} Δ z ( z d 12 ) = 2 z d 12 for large values of z m 1 z_{m1} z m 1 , z m 2 z_{m2} z m 2 and z d 12 z^{12}_{d} z d 12 . We also propose an equation for double diffractive production that satisfies the appropriate initial conditions.
Crucially, we have verified each step of our approach against the one dimensional model, which admits an exact solution, and found that all the intuition gained from this simple model carries over to our approach in QCD.
The process of double diffractive production into two high-mass clusters lies beyond the reach of the BK approach, so experimental data on this process offer a genuine test of how well the summed large Pomeron loops describe it. Confronting the predicted scaling behaviour with data from high-energy colliders such as the LHC [74 ] will therefore be an important step towards establishing whether large Pomeron loops play a significant role in high energy QCD dynamics, and in particular in the unitarization of the scattering amplitude.
Acknowledgements
We thank our colleagues at Tel Aviv University and UTFSM for encouraging discussions. This research was supported by Fondecyt (Chile) grant No. 1231829. J. G. expresses his gratitude to the Institute of Physics of PUCV (IFIS).
Solution to the BFKL equation
The BFKL equation for the elastic scattering amplitude (N N N ) has the form:
∂ N B F K L ( x 0 , x 1 , Y ) ∂ Y = α ˉ S 2 π ∫ d 2 x 2 x 01 2 x 02 2 x 12 2 { N B F K L ( x 1 , x 2 , Y ) + N B F K L ( x 0 , x 2 , Y ) − N B F K L ( x 0 , x 1 , Y ) } (102) \frac{\partial N^{\mathrm{BFKL}}(x_0,x_1,Y)}{\partial Y}=\frac{\bar{\alpha}_S}{2\pi}\int d^2x_2\frac{x_{01}^2}{x_{02}^2x_{12}^2}\left\{N^{\mathrm{BFKL}}(x_1,x_2,Y)+N^{\mathrm{BFKL}}(x_0,x_2,Y)-N^{\mathrm{BFKL}}(x_0,x_1,Y)\right\} \tag*{(102)} ∂ Y ∂ N BFKL ( x 0 , x 1 , Y ) = 2 π α ˉ S ∫ d 2 x 2 x 02 2 x 12 2 x 01 2 { N BFKL ( x 1 , x 2 , Y ) + N BFKL ( x 0 , x 2 , Y ) − N BFKL ( x 0 , x 1 , Y ) } (102)
The eigenfunctions of these equations have the following form [17 ] :
ϕ γ ( r , r P , b ) = ( r 2 r P 2 ( b − 1 2 ( r − r P ) ) 2 ( b + 1 2 ( r − r P ) ) 2 ) γ = e γ ξ r , r P ; (103) \phi_\gamma(r,r_P,b)=\left(\frac{r^2r_P^2}{\left(b-\frac{1}{2}(r-r_P)\right)^2\left(b+\frac{1}{2}(r-r_P)\right)^2}\right)^\gamma=e^{\gamma\xi_{r,r_P}}; \tag*{(103)} ϕ γ ( r , r P , b ) = ( ( b − 2 1 ( r − r P ) ) 2 ( b + 2 1 ( r − r P ) ) 2 r 2 r P 2 ) γ = e γ ξ r , r P ; (103)
with ξ r , r P = ln ( r 2 r P 2 ( b − 1 2 ( r − r P ) ) 2 ( b + 1 2 ( r − r P ) ) 2 ) ; (104) \text{with }\xi_{r,r_P}=\ln\left(\frac{r^2r_P^2}{\left(b-\frac{1}{2}(r-r_P)\right)^2\left(b+\frac{1}{2}(r-r_P)\right)^2}\right); \tag*{(104)} with ξ r , r P = ln ( ( b − 2 1 ( r − r P ) ) 2 ( b + 2 1 ( r − r P ) ) 2 r 2 r P 2 ) ; (104)
In fact, it is proven in Ref. [17 ] that this eigenfunction holds for any kernel that respects conformal symmetry. The eigenvalue for the BFKL equation has the form9 :
ω ( α ˉ S , γ ) = α ˉ S χ ( γ ) = α ˉ S ( 2 ψ ( 1 ) − ψ ( γ ) − ψ ( 1 − γ ) ) → γ → 1 2 4 ln 2 α ˉ S − ψ ( 2 ) ( 1 2 ) α ˉ S ( γ − 1 2 ) 2 + … (105) \omega(\bar{\alpha}_S,\gamma)=\bar{\alpha}_S\chi(\gamma)=\bar{\alpha}_S\left(2\psi(1)-\psi(\gamma)-\psi(1-\gamma)\right)\xrightarrow{\gamma\to\frac{1}{2}}4\ln2\bar{\alpha}_S-\psi^{(2)}\left(\frac{1}{2}\right)\bar{\alpha}_S\left(\gamma-\frac{1}{2}\right)^2+\ldots \tag*{(105)} ω ( α ˉ S , γ ) = α ˉ S χ ( γ ) = α ˉ S ( 2 ψ ( 1 ) − ψ ( γ ) − ψ ( 1 − γ ) ) γ → 2 1 4 ln 2 α ˉ S − ψ ( 2 ) ( 2 1 ) α ˉ S ( γ − 2 1 ) 2 + … (105)
where ψ ( z ) \psi(z) ψ ( z ) denotes the Euler psi-function: ψ ( z ) = d ln Γ ( z ) / d z \psi(z)=d\ln\Gamma(z)/dz ψ ( z ) = d ln Γ ( z ) / d z (see [68 ] ). In Eq. (A.1) we denote by x 1 x_1 x 1 and x 0 x_0 x 0 the coordinates of the quark and antiquark in the colourless dipole, respectively. In Eq. (A.2a) and Eq. (A.2b) we introduce: (i) the size of the dipole: r = x 1 − x 0 \mathbf{r}=\mathbf{x}_1-\mathbf{x}_0 r = x 1 − x 0 , (ii) the size of the dipole in the projectile (or target) r P = x 1 , P − x 0 , P \mathbf{r}_P=\mathbf{x}_{1,P}-\mathbf{x}_{0,P} r P = x 1 , P − x 0 , P and (iii) the impact parameter b = 1 2 ( x 1 + x 0 − x 1 , P − x 0 , P ) \mathbf{b}=\frac{1}{2}(\mathbf{x}_1+\mathbf{x}_0-\mathbf{x}_{1,P}-\mathbf{x}_{0,P}) b = 2 1 ( x 1 + x 0 − x 1 , P − x 0 , P ) .
For the interaction with nucleus, which we consider in this paper, the scattering amplitude of the dipole with a nucleus for the exchange of the BFKL Pomeron has the following form:
A d A B F K L ( r , R N , b ; Y ) = ∫ ϵ − i ∞ ϵ + i ∞ d γ 2 π i ∫ d 2 c e ω ( α ˉ S , γ ) Y ϕ γ ( r , R N , c ) S A ( b − c ) (106) A^{\mathrm{BFKL}}_{dA}(r,R_N,b;Y)=\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\gamma}{2\pi i}\int d^2c\,e^{\omega(\bar{\alpha}_S,\gamma)Y}\phi_\gamma(r,R_N,c)S_A(b-c) \tag*{(106)} A d A BFKL ( r , R N , b ; Y ) = ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d γ ∫ d 2 c e ω ( α ˉ S , γ ) Y ϕ γ ( r , R N , c ) S A ( b − c ) (106)
where R N R_N R N is the size of a nucleon, c c c is the impact parameter for dipole-nucleon amplitude and b − c b-c b − c is the position of the nucleon with respect to the center of the nucleus. S A ( b − c ) S_A(b-c) S A ( b − c ) is the number of nucleons in a nucleus. Since c ≪ b c\ll b c ≪ b we can integrate over c c c replacing b − c \mathbf{b}-\mathbf{c} b − c by b \mathbf{b} b and obtain the expression:
A d A B F K L ( r , R N , b ; Y ) = ∫ ϵ − i ∞ ϵ + i ∞ d γ 2 π i ∫ d 2 c e ω ( α ˉ S , γ ) Y ϕ γ ( r , R N , c ) S A ( b − c ) = ∫ ϵ − i ∞ ϵ + i ∞ d γ 2 π i ∫ d 2 c e ω ( α ˉ S , γ ) Y ϕ γ ( r , R N , c ) S A ( b ) = ( e ω ( α ˉ S , γ ) γ ˉ Y r 2 R N 2 ) γ ˉ R N 2 S A ( b ) = ( r 2 Q s 2 ( Y , R N , b ) ) γ ˉ (107) \begin{aligned}
A_{dA}^{\mathrm{BFKL}}(r,R_N,b;Y)=\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\gamma}{2\pi i}\int d^2c\,e^{\omega(\bar{\alpha}_S,\gamma)Y}\phi_\gamma(r,R_N,c)S_A(b-c)=\int_{\epsilon-i\infty}^{\epsilon+i\infty}\frac{d\gamma}{2\pi i}\int d^2c\,e^{\omega(\bar{\alpha}_S,\gamma)Y}\phi_\gamma(r,R_N,c)S_A(b) \\
&=\left(e^{\frac{\omega(\bar{\alpha}_S,\gamma)}{\bar{\gamma}}Y}\frac{r^2}{R_N^2}\right)^{\bar{\gamma}}R_N^2S_A(b)=\left(r^2Q_s^2(Y,R_N,b)\right)^{\bar{\gamma}} \tag*{(107)}
\end{aligned} A d A BFKL ( r , R N , b ; Y ) = ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d γ ∫ d 2 c e ω ( α ˉ S , γ ) Y ϕ γ ( r , R N , c ) S A ( b − c ) = ∫ ϵ − i ∞ ϵ + i ∞ 2 π i d γ ∫ d 2 c e ω ( α ˉ S , γ ) Y ϕ γ ( r , R N , c ) S A ( b ) = ( e γ ˉ ω ( α ˉ S , γ ) Y R N 2 r 2 ) γ ˉ R N 2 S A ( b ) = ( r 2 Q s 2 ( Y , R N , b ) ) γ ˉ (107)
Therefore, in this case we can absorb all dependence on the impact parameter in the b b b dependence of the saturation scale. In Eq. (A.5) we implicitly assume that r ≤ R N r\leq R_N r ≤ R N . One can see that Eq. (A.5) reduces to Eq. (I.8) which we use to obtain the dipole densities from the dipole-nucleus scattering. In Eq. (IV.3) the size of large dipole R = R A = A 1 / 3 R N R=R_A=A^{1/3}R_N R = R A = A 1/3 R N .
Elastic dipole-dipole scattering
First, we consider the elastic scattering of a dipole r r r with a nucleus, which satisfies the BK equation. The BK equation has the same form as Eq. (III.1) but for S S S , which is given by Eq. (II.4). S A d ( z d 01 ) S_A^d(z_d^{01}) S A d ( z d 01 ) is a series of multi-BFKL Pomeron exchanges, viz.:
S A d ( z d 01 ) = ∑ n = 0 ∞ ( − 1 ) n C n e n γ ˉ z d 01 e n γ ˉ z d 01 = Green’s function of n -BFKL Pomeron exchanges (108) S_A^d(z_d^{01})=\sum_{n=0}^{\infty}(-1)^n C_n e^{n\bar{\gamma}z_d^{01}}e^{n\bar{\gamma}z_d^{01}}=\text{Green's function of }n\text{-BFKL Pomeron exchanges} \tag*{(108)} S A d ( z d 01 ) = n = 0 ∑ ∞ ( − 1 ) n C n e n γ ˉ z d 01 e n γ ˉ z d 01 = Green’s function of n -BFKL Pomeron exchanges (108)
Plugging this equation in Eq. (IV.28) for the elastic S S S we obtain:
∑ n = 0 ∞ ( − 1 ) n n C n e n γ ˉ z d 01 = − z d 01 κ ˉ γ ˉ ∑ n = 0 ∞ ( − 1 ) n C n e n γ ˉ z d 01 or ⟨ ∣ n ∣ ⟩ = − z d 01 κ ˉ γ ˉ (109) \sum_{n=0}^{\infty}(-1)^n n C_n e^{n\bar{\gamma}z_d^{01}}=-\frac{z_d^{01}}{\bar{\kappa}\bar{\gamma}}\sum_{n=0}^{\infty}(-1)^n C_n e^{n\bar{\gamma}z_d^{01}}\quad\text{or}\quad\langle|n|\rangle=-\frac{z_d^{01}}{\bar{\kappa}\bar{\gamma}} \tag*{(109)} n = 0 ∑ ∞ ( − 1 ) n n C n e n γ ˉ z d 01 = − κ ˉ γ ˉ z d 01 n = 0 ∑ ∞ ( − 1 ) n C n e n γ ˉ z d 01 or ⟨ ∣ n ∣ ⟩ = − κ ˉ γ ˉ z d 01 (109)
We can understand this equation if we rewrite the sum over n n n as the integration over contour C C C in Figure 7 , viz.:
S A d ( z d 01 ) = ∮ C d n 2 π i C n π sin ( π n ) G P n ( z d 01 ) (110) S_A^d(z_d^{01})=\oint_C\frac{dn}{2\pi i}C_n\frac{\pi}{\sin(\pi n)}G_P^n(z_d^{01}) \tag*{(110)} S A d ( z d 01 ) = ∮ C 2 π i d n C n sin ( π n ) π G P n ( z d 01 ) (110)
Assuming that we can open the contour C C C to the contour C ′ C' C ′ we can take the integral using the method of steepest descent. Denoting C n = exp ( ϕ ( n ) ) C_n=\exp(\phi(n)) C n = exp ( ϕ ( n )) we obtain the equation for the saddle point n S P n_{SP} n S P :
ϕ n ′ ( n S P ) = − γ ˉ z d 01 (111) \phi'_n(n_{SP})=-\bar{\gamma}z_d^{01} \tag*{(111)} ϕ n ′ ( n S P ) = − γ ˉ z d 01 (111)
From Eq. (B.1) n S P = − z γ ˉ κ ˉ n_{SP}=-\frac{z}{\bar{\gamma}\bar{\kappa}} n S P = − γ ˉ κ ˉ z and Eq. (B.4) gives ϕ ( n ) = γ ˉ 2 κ ˉ n 2 2 \phi(n)=\bar{\gamma}^2\bar{\kappa}\frac{n^2}{2} ϕ ( n ) = γ ˉ 2 κ ˉ 2 n 2 . Therefore, Eq. (I.5) for ρ n ( Y 01 − Y 0 , r , b ; { r i , b i } ) \rho_n(Y_{01}-Y_0,r,b;\{r_i,b_i\}) ρ n ( Y 01 − Y 0 , r , b ; { r i , b i }) has the form:
ρ n ( Y 01 − Y 0 , r , b ; { r i , b i } ) = e γ ˉ 2 κ ˉ n 2 2 ∏ i = 1 n G I P ( z i , d 01 ) with z i , d 01 = κ ˉ α ˉ S ( Y 01 − Y 0 ) + ξ r i , r (112) \rho_n(Y_{01}-Y_0,r,b;\{r_i,b_i\})=e^{\bar{\gamma}^2\bar{\kappa}\frac{n^2}{2}}\prod_{i=1}^{n}G_{IP}(z_{i,d}^{01})\quad\text{with}\quad z_{i,d}^{01}=\bar{\kappa}\bar{\alpha}_S(Y_{01}-Y_0)+\xi_{r_i,r} \tag*{(112)} ρ n ( Y 01 − Y 0 , r , b ; { r i , b i }) = e γ ˉ 2 κ ˉ 2 n 2 i = 1 ∏ n G I P ( z i , d 01 ) with z i , d 01 = κ ˉ α ˉ S ( Y 01 − Y 0 ) + ξ r i , r (112)
Calculating S d d ( z d 12 ) S_d^d(z_d^{12}) S d d ( z d 12 ) using Eq. (I.1), Eq. (II.8) and Eq. (B.5) we obtain
S d d ( z d 12 ) = ∫ 0 ∞ d t e − t ∑ n = 0 ∞ ( − 1 ) n C n 2 t n e n γ ˉ z d 12 = ∫ 0 ∞ d t e − t ∑ n = 0 ∞ ( − 1 ) n e 2 ϕ ( n ) t n e n γ ˉ z d 12 (113) S_d^d(z_d^{12})=\int_0^\infty dte^{-t}\sum_{n=0}^{\infty}(-1)^n C_n^2t^n e^{n\bar{\gamma}z_d^{12}}=\int_0^\infty dte^{-t}\sum_{n=0}^{\infty}(-1)^n e^{2\phi(n)}t^n e^{n\bar{\gamma}z_d^{12}} \tag*{(113)} S d d ( z d 12 ) = ∫ 0 ∞ d t e − t n = 0 ∑ ∞ ( − 1 ) n C n 2 t n e n γ ˉ z d 12 = ∫ 0 ∞ d t e − t n = 0 ∑ ∞ ( − 1 ) n e 2 ϕ ( n ) t n e n γ ˉ z d 12 (113)
In this equation we use z i , d 01 = κ ˉ α ˉ S ( Y 01 − Y 0 ) + ξ r i , r z_{i,d}^{01}=\bar{\kappa}\bar{\alpha}_S(Y_{01}-Y_0)+\xi_{r_i,r} z i , d 01 = κ ˉ α ˉ S ( Y 01 − Y 0 ) + ξ r i , r , z i , d 02 = κ ˉ α ˉ S ( Y 0 − Y 02 ) + ξ r i ′ , r ′ z_{i,d}^{02}=\bar{\kappa}\bar{\alpha}_S(Y_0-Y_{02})+\xi_{r_i',r'} z i , d 02 = κ ˉ α ˉ S ( Y 0 − Y 02 ) + ξ r i ′ , r ′ and z d 12 z_d^{12} z d 12 from Eq. (IV.3). We use Eq. (II.8) to make the integrals over r i r_i r i and b i b_i b i .
Taking integral over t t t we have:
S d d ( z d 12 ) = ∑ n = 0 ∞ ( − 1 ) n C n 2 Γ ( n + 1 ) e n γ ˉ z d 12 = ∑ n = 0 ∞ ( − 1 ) n e 2 ϕ ( n ) n ! e n γ ˉ z d 12 (114) S_d^d(z_d^{12})=\sum_{n=0}^{\infty}(-1)^n C_n^2\Gamma(n+1)e^{n\bar{\gamma}z_d^{12}}=\sum_{n=0}^{\infty}(-1)^n e^{2\phi(n)}n!e^{n\bar{\gamma}z_d^{12}} \tag*{(114)} S d d ( z d 12 ) = n = 0 ∑ ∞ ( − 1 ) n C n 2 Γ ( n + 1 ) e n γ ˉ z d 12 = n = 0 ∑ ∞ ( − 1 ) n e 2 ϕ ( n ) n ! e n γ ˉ z d 12 (114)
Rewriting this integral as a contour integral (see Eq. (B.3)), and taking the integral over n n n using contour C ′ ′ C^{\prime\prime} C ′′ in Figure 7 , we can write Eq. (B.7) in the form:
S d d ( z d 12 ) = ∮ C ′ ′ d n 2 π i C n 2 π Γ ( 1 + n ) sin ( π n ) e n γ ˉ z d 12 = ∑ k = 1 ∞ d d k [ 1 Γ ( k ) e 2 ϕ ( − k ) − k γ ˉ z d 12 ] = ∑ k = 1 ∞ Φ ( k ) Γ ( k ) e γ ˉ 2 κ k 2 − k γ ˉ z d 12 (115) S_{d}^{d}\left(z_{d}^{12}\right)=\oint_{C^{\prime\prime}}\frac{dn}{2\pi i}C_{n}^{2}\frac{\pi\Gamma(1+n)}{\sin(\pi n)}e^{n\bar{\gamma}z_{d}^{12}}=\sum_{k=1}^{\infty}\frac{d}{dk}\left[\frac{1}{\Gamma(k)}e^{2\phi(-k)-k\bar{\gamma}z_{d}^{12}}\right]=\sum_{k=1}^{\infty}\frac{\Phi(k)}{\Gamma(k)}e^{\bar{\gamma}^{2}\kappa k^{2}-k\bar{\gamma}z_{d}^{12}} \tag*{(115)} S d d ( z d 12 ) = ∮ C ′′ 2 π i d n C n 2 sin ( π n ) π Γ ( 1 + n ) e n γ ˉ z d 12 = k = 1 ∑ ∞ d k d [ Γ ( k ) 1 e 2 ϕ ( − k ) − k γ ˉ z d 12 ] = k = 1 ∑ ∞ Γ ( k ) Φ ( k ) e γ ˉ 2 κ k 2 − k γ ˉ z d 12 (115)
where Φ ( k ) = ψ ( 0 ) ( k ) − 2 γ ˉ 2 κ k − γ ˉ z d 12 \Phi(k)=\psi^{(0)}(k)-2\bar{\gamma}^{2}\kappa k-\bar{\gamma}z_{d}^{12} Φ ( k ) = ψ ( 0 ) ( k ) − 2 γ ˉ 2 κk − γ ˉ z d 12 is the residue of the double pole. In Eq. (B.8) we use that Γ ( n → − k ) = ( − 1 ) k k ! ( n + k ) \Gamma(n\to-k)=\frac{(-1)^{k}}{k!(n+k)} Γ ( n → − k ) = k ! ( n + k ) ( − 1 ) k (see Ref. [68 ] , formula 8.310), where k k k is a positive integer. In Eq. (B.8) we also used that ϕ ( − k ) = ϕ ( k ) \phi(-k)=\phi(k) ϕ ( − k ) = ϕ ( k ) . Taking the sum in this equation by the method of steepest descent, we find that
S d d ( z d 12 ) = Φ ( z d 12 ) Γ ( z d 12 2 γ ˉ κ ) exp ( − ( z d 12 ) 2 4 κ ) (116) S_{d}^{d}\left(z_{d}^{12}\right)=\frac{\Phi\left(z_{d}^{12}\right)}{\Gamma\left(\frac{z_{d}^{12}}{2\bar{\gamma}\kappa}\right)}\exp\left(-\frac{\left(z_{d}^{12}\right)^{2}}{4\kappa}\right) \tag*{(116)} S d d ( z d 12 ) = Γ ( 2 γ ˉ κ z d 12 ) Φ ( z d 12 ) exp ( − 4 κ ( z d 12 ) 2 ) (116)
where Φ \Phi Φ is a smooth function of z d 12 z_{d}^{12} z d 12 .
For a function A ( z ) = ∑ n = 0 ∞ a n z n A(z)=\sum_{n=0}^{\infty}a_{n}z^{n} A ( z ) = ∑ n = 0 ∞ a n z n , the Borel representation reads [55 ]
A ( z ) = ∑ n = 0 ∞ a n z n = ∫ 0 ∞ d t e − t B A ( t z ) , B A ( t z ) = ∑ n = 0 ∞ a n ( t z ) n n ! , (117) A(z)=\sum_{n=0}^{\infty}a_{n}z^{n}=\int_{0}^{\infty}dt\,e^{-t}\mathcal{B}A(tz),\qquad\mathcal{B}A(tz)=\sum_{n=0}^{\infty}a_{n}\frac{(tz)^{n}}{n!}, \tag*{(117)} A ( z ) = n = 0 ∑ ∞ a n z n = ∫ 0 ∞ d t e − t B A ( t z ) , B A ( t z ) = n = 0 ∑ ∞ a n n ! ( t z ) n , (117)
which follows from ∫ 0 ∞ d t e − t t n = n ! \int_{0}^{\infty}dt\,e^{-t}t^{n}=n! ∫ 0 ∞ d t e − t t n = n ! .
We apply this to the elastic S S S -matrix, taking G P ( z ) = e γ ˉ z G_{P}(z)=e^{\bar{\gamma}z} G P ( z ) = e γ ˉ z and C n = e γ ˉ 2 κ n 2 / 2 C_{n}=e^{\bar{\gamma}^{2}\kappa n^{2}/2} C n = e γ ˉ 2 κ n 2 /2 :
S ( z ) = ∑ n = 0 ∞ ( − 1 ) n C n G P n ( z ) = ∑ n = 0 ∞ C n ( − G P ( z ) ) n = ∫ 0 ∞ d t e − t B S ( − t e γ ˉ z ) , (118) S(z)=\sum_{n=0}^{\infty}(-1)^{n}C_{n}G_{P}^{n}(z)=\sum_{n=0}^{\infty}C_{n}\left(-G_{P}(z)\right)^{n}=\int_{0}^{\infty}dt\,e^{-t}\mathcal{B}S\left(-te^{\bar{\gamma}z}\right), \tag*{(118)} S ( z ) = n = 0 ∑ ∞ ( − 1 ) n C n G P n ( z ) = n = 0 ∑ ∞ C n ( − G P ( z ) ) n = ∫ 0 ∞ d t e − t B S ( − t e γ ˉ z ) , (118)
where the Borel image is
B S ( − t e γ ˉ z ) = ∑ n = 0 ∞ C n ( − t e γ ˉ z ) n n ! = ∑ n = 0 ∞ ( − 1 ) n n ! t n e γ ˉ 2 κ n 2 / 2 e n γ ˉ z . (119) \mathcal{B}S\left(-te^{\bar{\gamma}z}\right)=\sum_{n=0}^{\infty}C_{n}\frac{\left(-te^{\bar{\gamma}z}\right)^{n}}{n!}=\sum_{n=0}^{\infty}\frac{(-1)^{n}}{n!}t^{n}e^{\bar{\gamma}^{2}\kappa n^{2}/2}e^{n\bar{\gamma}z}. \tag*{(119)} B S ( − t e γ ˉ z ) = n = 0 ∑ ∞ C n n ! ( − t e γ ˉ z ) n = n = 0 ∑ ∞ n ! ( − 1 ) n t n e γ ˉ 2 κ n 2 /2 e n γ ˉ z . (119)
To perform the sum over n n n , we linearize the quadratic exponent with the Gaussian (Hubbard–Stratonovich) identity, valid for a > 0 a>0 a > 0 :
e a n 2 = 1 4 π a ∫ − ∞ ∞ d λ e − λ 2 4 a + n λ . (120) e^{an^{2}}=\frac{1}{\sqrt{4\pi a}}\int_{-\infty}^{\infty}d\lambda\,e^{-\frac{\lambda^{2}}{4a}+n\lambda}. \tag*{(120)} e a n 2 = 4 π a 1 ∫ − ∞ ∞ d λ e − 4 a λ 2 + nλ . (120)
Setting a = γ ˉ 2 κ / 2 a=\bar{\gamma}^{2}\kappa/2 a = γ ˉ 2 κ /2 and summing the resulting exponential series in n n n , we obtain
B S ( − t e γ ˉ z ) = 1 2 π γ ˉ 2 κ ∫ − ∞ ∞ d λ exp ( − 1 γ ˉ 2 κ ( λ 2 2 + η ( z ) e λ ) ) , η ( z ) = γ ˉ 2 κ t e γ ˉ z . (121) \mathcal{B}S\left(-te^{\bar{\gamma}z}\right)=\frac{1}{\sqrt{2\pi\bar{\gamma}^{2}\kappa}}\int_{-\infty}^{\infty}d\lambda\,\exp\left(-\frac{1}{\bar{\gamma}^{2}\kappa}\left(\frac{\lambda^{2}}{2}+\eta(z)e^{\lambda}\right)\right),\qquad\eta(z)=\bar{\gamma}^{2}\kappa t e^{\bar{\gamma}z}. \tag*{(121)} B S ( − t e γ ˉ z ) = 2 π γ ˉ 2 κ 1 ∫ − ∞ ∞ d λ exp ( − γ ˉ 2 κ 1 ( 2 λ 2 + η ( z ) e λ ) ) , η ( z ) = γ ˉ 2 κ t e γ ˉ z . (121)
The integral is evaluated by the saddle-point method, which is justified by the large factor 1 / ( γ ˉ 2 κ ) 1/(\bar{\gamma}^{2}\kappa) 1/ ( γ ˉ 2 κ ) in the exponent. The saddle-point condition is
λ S P + η ( z ) e λ S P = 0 , (122) \lambda_{SP}+\eta(z)e^{\lambda_{SP}}=0, \tag*{(122)} λ S P + η ( z ) e λ S P = 0 , (122)
with the solution
λ S P = − W ( η ( z ) ) , (123) \lambda_{SP}=-W(\eta(z)), \tag*{(123)} λ S P = − W ( η ( z )) , (123)
where W W W is the Lambert W W W function, defined by W ( η ) e W ( η ) = η W(\eta)e^{W(\eta)}=\eta W ( η ) e W ( η ) = η . For large η \eta η it behaves as W ( η ) ≃ ln η − ln ( ln η ) W(\eta)\simeq\ln\eta-\ln(\ln\eta) W ( η ) ≃ ln η − ln ( ln η ) . Since ln η = γ ˉ z + ln ( γ ˉ 2 κ t ) \ln\eta=\bar{\gamma}z+\ln(\bar{\gamma}^{2}\kappa t) ln η = γ ˉ z + ln ( γ ˉ 2 κ t ) , at large z z z we have λ S P ( z ) ≃ − γ ˉ z \lambda_{SP}(z)\simeq-\bar{\gamma}z λ S P ( z ) ≃ − γ ˉ z , up to terms that depend only logarithmically on z z z and t t t .
Evaluating the Gaussian integral around the saddle point, and using η e λ S P = − λ S P \eta e^{\lambda_{SP}}=-\lambda_{SP} η e λ S P = − λ S P together with ∂ λ 2 ( λ 2 / 2 + η e λ ) ∣ S P = 1 − λ S P \left.\partial_{\lambda}^{2}(\lambda^{2}/2+\eta e^{\lambda})\right|_{SP}=1-\lambda_{SP} ∂ λ 2 ( λ 2 /2 + η e λ ) S P = 1 − λ S P , gives
B S ( − t e γ ˉ z ) ≃ 1 1 − λ S P exp ( − λ S P 2 − 2 λ S P 2 γ ˉ 2 κ ) ≃ Ψ ( z ) exp ( − z 2 2 κ ) (124) \mathcal{B}S(-t e^{\bar{\gamma}z})\simeq\frac{1}{\sqrt{1-\lambda_{SP}}}\exp\left(-\frac{\lambda_{SP}^{2}-2\lambda_{SP}}{2\bar{\gamma}^{2}\kappa}\right)\simeq\Psi(z)\exp\left(-\frac{z^{2}}{2\kappa}\right) \tag*{(124)} B S ( − t e γ ˉ z ) ≃ 1 − λ S P 1 exp ( − 2 γ ˉ 2 κ λ S P 2 − 2 λ S P ) ≃ Ψ ( z ) exp ( − 2 κ z 2 ) (124)
where Ψ ( z ) \Psi(z) Ψ ( z ) is a smooth function of z z z . Substituting λ S P ≃ − γ ˉ z \lambda_{SP}\simeq-\bar{\gamma}z λ S P ≃ − γ ˉ z shows that it collects the prefactor and the term linear in z z z of the exponent, Ψ ( z ) ≃ ( 1 + γ ˉ z ) − 1 / 2 e − z / ( γ ˉ κ ) \Psi(z)\simeq(1+\bar{\gamma}z)^{-1/2}e^{-z/(\bar{\gamma}\kappa)} Ψ ( z ) ≃ ( 1 + γ ˉ z ) − 1/2 e − z / ( γ ˉ κ ) . Thus the Gaussian fall-off e − z 2 / 2 κ e^{-z^{2}/2\kappa} e − z 2 /2 κ dominates the large-z z z behavior of S S S .
References