1: \chapter{Hadronic two-body decays II: {Real~part}}
2: \label{ch:RePart}
3:
4: The calculation of the real part of the NNLO vertex corrections in
5: hadronic two-body decays proceeds along the same lines as the one of
6: the imaginary part which we presented in the previous chapter.
7: However, we will see below that the calculation of the real part is
8: far more complicated involving many additional MIs and the full NNLO
9: complexity concerning e.g.~the issue of renormalization and the
10: treatment of evanescent operators.
11:
12: In this chapter we present a preliminary result for the real part of
13: the colour-allowed tree amplitude $\alpha_1$ (in the CMM operator
14: basis). Similar to what we have seen in the previous chapter, the
15: colour-suppressed tree amplitude $\alpha_2$ can then be derived from
16: $\alpha_1$ after the transformation into a Fierz-symmetric operator
17: basis (which we called QCDF basis in~Chapter~\ref{ch:ImPart}). In
18: NNLO we thus have to extend the basis from (\ref{eq:Basis:QCDF}) to
19: include 2-loop evanescent operators with an appropriate definition
20: of $\eps$- and $\eps^2$-terms which guarantees manifest Fierz
21: symmetry in this operator basis. As we have not yet worked out the
22: details of this last step, we refer to~\cite{GBRePart} for the
23: analysis of the colour-suppressed tree amplitude.
24:
25: \section{2-loop calculation}
26:
27: The 2-loop calculation will be performed in the CMM operator basis
28: (cf.~Section~\ref{eq:BtoPiPi:OpBasis}). Apart from the operators in
29: (\ref{eq:Basis:CMM}), we have to take into account 2-loop evanescent
30: operators which are defined by
31: \begin{align}
32: \hat E_1' &= \left[\bar u
33: \gamma^\mu\gamma^\nu\gamma^\rho\gamma^\sigma\gamma^\tau L\, T^A
34: b\right] \;
35: \left[\bar d \gamma_\mu\gamma_\nu\gamma_\rho\gamma_\sigma\gamma_\tau \,L\, T^A u\right]
36: -20 \hat E_1-256 \,\hat Q_1,\no\\
37: \hat E_2' &= \left[\bar u
38: \gamma^\mu\gamma^\nu\gamma^\rho\gamma^\sigma\gamma^\tau L\, b\right]
39: \;
40: \left[\bar d \gamma_\mu\gamma_\nu\gamma_\rho\gamma_\sigma\gamma_\tau \,L\, u\right]
41: -20 \hat E_2-256 \,\hat Q_2.
42: \label{eq:2loopEvan:CMM}
43: \end{align}
44: We turn to a brief characterization of the considered 2-loop
45: calculation following our recipe from Section~\ref{sec:strategy}.
46:
47: \subsubsection{Step 1: Set-up for loop calculation}
48:
49: In contrast to the calculation of the imaginary part
50: from~Chapter~\ref{ch:ImPart}, we now have to consider the whole set
51: of non-factorizable 2-loop diagrams from
52: Figure~\ref{fig:BtoPiPi:NNLODiags}. Notice that the most complicated
53: diagrams only enter the calculation of the real part. Whereas the
54: diagrams that we considered for the calculation of the imaginary
55: part contained at most one massive ($b$-quark) propagator, we now
56: have to deal with up to three massive propagators.
57:
58: \subsubsection{Step 2: Reduction to Master Integrals}
59:
60: The fact that the diagrams involve between $0$-$3$ massive
61: propagators immediately leads to many distinct topologies and a
62: large number of MIs. In addition to the 14 MIs from
63: Figure~\ref{fig:BtoPiPi:MIs}, we find 22 (real) MIs which are shown
64: in Figure~\ref{fig:BtoPiPi:MIsRe}. We further remark that our {\sc
65: Mathematica} implementation of the reduction algorithm hardly
66: succeeds to reduce the most complicated diagrams of the considered
67: calculation within a reasonable amount of CPU time.
68:
69: \begin{figure}[b!]\vspace{3mm}
70: \centerline{\parbox{15cm}{\centerline{
71: \includegraphics[width=15cm]{FigMIRe.ps}}\vspace{5mm}
72: \caption{\label{fig:BtoPiPi:MIsRe} \small \textit{Additional Master
73: Integrals that have to be considered for the calculation of the real
74: part of the NNLO vertex corrections. For details
75: cf.~Figure~\ref{fig:BtoPiPi:MIs}.}}}}
76: \end{figure}
77:
78: \subsubsection{Step 3: Manipulation of Dirac structures}
79:
80: Concerning the irreducible Dirac structures we find the same set
81: (\ref{eq:BtoPiPi:Dirac}) as in the calculation of the imaginary
82: part. In contrast to our analysis from Chapter~\ref{ch:ImPart}, the
83: last structure now enters the divergent piece of the calculation
84: giving rise to 2-loop evanescent operators according to
85: (\ref{eq:2loopEvan:CMM}).
86:
87: \subsubsection{Step 4: Calculation of Master Integrals}
88:
89: In the last step of the calculation it becomes obvious that the
90: analysis of the real part of the NNLO vertex corrections is much
91: more complex than the one of the imaginary part. We now have to
92: calculate the real parts of the MIs from
93: Figure~\ref{fig:BtoPiPi:MIs} as well as the MIs from
94: Figure~\ref{fig:BtoPiPi:MIsRe}, in general up to five orders in the
95: $\eps$-expansion. In order to tackle this highly challenging task we
96: applied the same techniques as in our calculation from
97: Chapter~\ref{ch:ImPart}. It turns out that almost all MIs can be
98: expressed in terms of the set (\ref{eq:HPLminiIm}) supplemented by
99: the following HPLs of weight $w=4$
100: \begin{align}
101: H(0,0,0,1;u) &= \Lid(u), \no \\
102: H(0,0,1,1;u) &= \Sbb(u), \no \\
103: H(0,1,1,1;u) &= \Sac(u).
104: \end{align}
105: Unfortunately, some MIs with two and three massive propagators do
106: not fit into this pattern. We further have to include HPLs related
107: to the parameter $-1$ and to the argument $u-1$,
108: cf.~(\ref{eq:MI:HPLext1})~--~(\ref{eq:MI:HPLext2}). Still, we find
109: two functions which we could not express in terms of Nielsen
110: polylogarithms: $H(0,-1,0,1;u)$ and ${\cal{A}}(u)$ from
111: (\ref{eq:auxA})\footnote{The appearance of the function
112: ${\cal{A}}(u)$ seems to be an artefact of our calculation, i.e.~of
113: our special choice of the MIs. Though it enters two of our MIs from
114: Figure~\ref{fig:BtoPiPi:MIsRe}, it drops out in their sum in the
115: respective Feynman diagrams and is therefore irrelevant for our
116: purposes.}.
117:
118: \newabs
119: Similar to what we have seen in Chapter~\ref{ch:ImPart}, the
120: situation is more complicated for the MIs which stem from the
121: diagrams with a closed fermion loop. So far we have not yet
122: calculated these MIs with a massive internal quark (wavy lines),
123: i.e.~we work in an approximation which treats all quarks in these
124: loops as massless. We stress that this is certainly inconsistent for
125: an internal $b$-quark with respect to the remainder of the
126: calculation. However, we expect that this will only have a minor
127: impact on our (preliminary) result. A consistent treatment of these
128: diagrams is relegated to~\cite{GBRePart}. The analytical results for
129: all other MIs can be found in Appendix~\ref{app:MIsRe}.
130:
131:
132: \section{Renormalization and IR subtractions}
133:
134: We have shown in detail in Section~\ref{sec:BtoPiPi:UVIR} how to
135: perform appropriate UV and IR subtractions in order to extract the
136: NNLO kernels $T_i^{(2)}$ from the hadronic matrix elements. The
137: essence is summarized in the Master Formula (\ref{eq:Master}). We
138: briefly address the issues that go beyond our analysis of the
139: imaginary part from Chapter~\ref{ch:ImPart}.
140:
141: \subsection{Renormalization}
142:
143: The renormalization of the $b$-quark mass becomes relevant in the
144: considered calculation. We treat the $b$-quark in the on-shell
145: scheme according to
146: \begin{align}
147: Z_m(m) = 1 - \frac{\as C_F}{4\pi} \left( \frac{\mu^2
148: e^{\gamma_E}}{m^2} \right)^\eps \Gamma(\eps) \;
149: \frac{3-2\eps}{1-2\eps} + \calO(\as^2).
150: \end{align}
151: Due to the full NNLO complexity, we require the 2-loop
152: renormalization matrix $\hat{Z}^{(2)}$ which we take from~\cite{GGH}
153: \renewcommand{\arraycolsep}{2mm}
154: \begin{align}
155: \hat{Z}^{(2)} &= \left(
156: \begin{array}{c c c c c c}
157: \rule[-2mm]{0mm}{7mm} \frac{41}{3} & -\frac{58}{9} & -\frac{125}{36} & - \frac{73}{54} & \frac{19}{96} & \frac{5}{108} \\
158: \rule[-2mm]{0mm}{7mm} -29 & 4 & -\frac{73}{12} & 0 & \frac{5}{24} &
159: \frac19
160: \end{array}
161: \right) \, \frac{1}{\eps^2} \no \\
162: & \quad + \left(
163: \begin{array}{c c c c c c}
164: \rule[-2mm]{0mm}{7mm} \frac{317}{36} & -\frac{515}{54} & \frac{4493}{864} & -\frac{49}{648} & \frac{1}{384} & -\frac{35}{864} \\
165: \rule[-2mm]{0mm}{7mm} \frac{349}{12} & 3 &\frac{1031}{144} & \frac89
166: & -\frac{35}{192} & -\frac{7}{72}
167: \end{array}
168: \right) \, \frac{1}{\eps}, \label{eq:Z1}
169: \end{align}
170: where the two lines correspond to the physical operators $\{
171: \hat{Q}_1, \hat{Q}_2\}$ and the columns to the extended operator
172: basis $\{ \hat{Q}_1, \hat{Q}_2, \hat{E}_1, \hat{E}_2, \hat{E}_1',
173: \hat{E}_2' \}$ including evanescent operators.
174:
175: \subsection{IR subtractions}
176:
177: In order to perform the IR subtractions from the right hand side of
178: (\ref{eq:Master}), we must compute the real parts of the NLO kernels
179: $T_i^{(1)}$ to $\calO(\eps^2)$. The kernels vanish in the
180: colour-singlet case $T_2^{(1)}=T_{2,E}^{(1)}=0$, whereas the
181: colour-octet kernels are found to be
182: \begin{align}
183: \Re \;T_1^{(1)}(u) &= \frac{C_F}{2N_c} \bigg\{ \Big(-6L+t_0(u)\Big)
184: \Big( 1+\eps L +\frac12 \eps^2 L^2\Big) \no \\
185: & \hspace{2cm} + \Big(3L^2+t_1(u)\Big) \Big( \eps +\eps^2 L \Big) \no \\
186: & \hspace{2cm} + \Big( -L^3 + t_2(u)\Big) \,\eps^2 + \calO(\eps^3)
187: \bigg\}, \no \\
188: \Re \;T_{1,E}^{(1)}(u) &= -\frac{C_F}{4N_c} \bigg\{
189: \Big(2L+t_{E,0}(u)\Big)\Big( 1+\eps L \Big) \no \\
190: & \hspace{2.2cm} + \Big(-L^2+t_{E,1}(u)\Big)\eps + \calO(\eps^2)
191: \bigg\},
192: \end{align}
193: with $L\equiv\ln \mu^2/m_b^2$ and
194: \begin{align}
195: t_0(u) &= 4 \Lib(u) - \ln^2 u +2 \ln u \ln \ubar + \ln^2 \ubar +(2-3u) \Big( \frac{\ln u}{\ubar} - \frac{\ln \ubar}{u} \Big) -\frac{\pi^2}{3} -22, \no \\
196: t_1(u) &= -2 \Lic(u) -2 \Sab(u) -2 \ln \ubar \, \Lib(u) + \ln^3 u -2 \ln^2 u \ln \ubar + \ln u \ln^2 \ubar - \ln^3 \ubar \no \\
197: & \quad \,+ \frac{2-3u^2}{u\ubar} \Lib(u) - \frac{2-3u}{\ubar} \Big(\ln^2 u - \ln u \ln \ubar \Big) + \frac{6-11u +2\ubar \pi^2}{\ubar} \ln u \no \\
198: & \quad \,+\frac{4-3u}{2u} \ln^2 \ubar - \frac{18-33u+5u
199: \pi^2}{3u}\ln \ubar +\frac{(7-6u)\pi^2}{6\ubar} +2 \zeta_3 -52, \no
200: \\
201: t_2(u) &= 10 \Lid(u) -8 \Sbb(u) +10 \Sac(u) -8 \ln \ubar \, \Lic(u)
202: +10 \ln \ubar \, \Sab(u) - \frac{7}{12} \ln^4 u \no
203: \end{align}
204: \begin{align}
205: & \quad \, +5 \ln^2 \ubar \, \Lib(u) + \frac43 \ln^3 u \ln \ubar -\ln^2 u \ln^2 \ubar +\frac13 \ln u \ln^3 \ubar +\frac{7}{12} \ln^4\ubar \no \\
206: & \quad \, + \frac{2-6u+6u^2}{u \ubar} \Lic(u) -\frac{4-6u+3u^2}{u\ubar} \Big( \Sab(u) +\ln \ubar\,\Lib(u) \Big) -\frac{8-3u}{6u} \ln^3 \ubar \no \\
207: & \quad \, + \frac{2-3u}{6\ubar} \Big( 4 \ln^3 u - 6\ln^2 u \ln\ubar
208: +3 \ln u \ln^2 \ubar \Big) -
209: \frac{60(1-2u)+17\ubar\pi^2}{12\ubar}\ln^2 u \no \\
210: & \quad \, +
211: \frac{3(6-4u-7u^2)+u\ubar \pi^2}{3u\ubar} \Lib(u)
212: +\frac{24-54u+5\ubar \pi^2}{6\ubar} \ln u \ln \ubar
213: +\frac{(29-24u)\pi^2}{6\ubar} \no \\
214: & \quad \, + \frac{6(12-13u)+7u\pi^2}{12u} \ln^2 \ubar +\frac{24(7-13u)+(10-15u)\pi^2}{12\ubar} \ln u - \frac{23\pi^4}{180}\no \\
215: & \quad \, - \frac{24\ubar(7-13u)+(2+23u-27u^2)\pi^2+24u\ubar
216: \zeta_3}{12u\ubar} \ln \ubar + \frac{10-11u}{\ubar} \zeta_3-112, \no
217: \\
218: t_{E,0}(u) &=-\frac{1-2u}{2} \Big( \frac{\ln u}{\ubar} - \frac{\ln
219: \ubar}{u} \Big) + \frac{16}{3}, \no \\
220: t_{E,1}(u) &= -\frac{1-2u}{2u\ubar} \Lib(u) + \frac{1-3u}{4\ubar}
221: \ln^2 u + \frac{u}{2\ubar} \ln u \ln \ubar -\frac{2-3u}{4u} \ln^2
222: \ubar \no \\
223: & \quad \, -\frac{4(1-2u)}{3} \Big( \frac{\ln u}{\ubar} - \frac{\ln
224: \ubar}{u} \Big) - \frac{(6-5u)\pi^2}{12\ubar}+12. \label{eq:T1:Re}
225: \end{align}
226: In combination with the form factor corrections from
227: (\ref{eq:BtoPiPi:FFphys}) and (\ref{eq:BtoPiPi:FFevan}), this
228: determines the first subtraction term in (\ref{eq:Master}). In the
229: second subtraction we require the convolution of the NLO kernels
230: with the wave function corrections from (\ref{eq:Phiphys}) and
231: (\ref{eq:Phievan}). We find
232: \begin{align}
233: F^{(0)} \; \Re \;T_1^{(1)}\; \Phi_\text{amp}^{(1)} & \;=\;
234: \frac{C_F^2}{N_c} \bigg\{ t_3(u) \, \bigg( \frac{1}{\eps} + L \bigg)
235: + t_4(u) +\calO(\eps) \bigg\} \;
236: F^{(0)} \;\Phi^{(0)}, \no \\
237: F_E^{(0)} \; \Re \;T_{1,E}^{(1)}\; \Phi_\text{amp,E}^{(1)} &
238: \;\to\; \frac{C_F^2}{N_c} \bigg\{ 12L+ t_{E,2}(u) +\calO(\eps)
239: \bigg\} \; F^{(0)} \;\Phi^{(0)},
240: \end{align}
241: where
242: \begin{align}
243: t_3(u) &= 4 \Lic(u) + 4 \Sab(u) -4 \ln u \, \Lib(u) +\frac23 \ln^3 u
244: -2 \ln^2 u \ln \ubar -\frac23 \ln^3 \ubar - \frac{\Lib(u)}{u\ubar} \no \\
245: & \quad - \frac{1-3u}{2u\ubar} \Big( u \ln^2u +2 \ubar \ln u \ln
246: \ubar - \ubar \ln^2 \ubar \Big)\! - \frac{3}{2u} \ln \ubar +
247: \frac{(4-3u)\pi^2}{6\ubar}-\frac{15}{2}-4 \zeta_3 ,\no \\
248: t_4(u) &= 12 \Lid(u) -20 \Sbb(u) +12 \Sac(u) -8 \Big(\ln u + \ln
249: \ubar \Big) \Lic(u) +12 \ln u \Sab(u) \no \\
250: & \quad +4 \ln \ubar \, \Sab(u) + \! \Big(4\ln^2 u +4 \ln u \ln
251: \ubar +2 \ln^2 \ubar \Big) \Lib(u) - \frac34 \ln^4 u +\frac73 \ln^3
252: u \ln \ubar \no \\
253: & \quad -\frac12 \ln^2 u \ln^2 \ubar -\frac13 \ln u \ln^3 \ubar
254: +\frac34 \ln^4 \ubar - \frac{4-11u+3u^2}{u\ubar} \Lic(u) +
255: \frac{5-12u}{6\ubar} \ln^3 u\no
256: \end{align}
257: \begin{align}
258: & \quad + \frac{1+u-3u^2}{u\ubar} \Sab(u) +
259: \frac{2-10u+6u^2}{u\ubar} \ln u \, \Lib(u) -
260: \frac{1-5u+3u^2}{u\ubar} \ln \ubar \, \Lib(u) \no \\
261: & \quad +
262: \frac{2-10u+9u^2}{2u\ubar} \ln^2 u \ln \ubar -
263: \frac{1-2u}{2u\ubar} \ln u \ln^2 \ubar - \frac{5-6u}{6u} \ln^3 \ubar \no \\
264: & \quad - \frac{18-24u+15u^2-10u\ubar \pi^2}{3u\ubar} \Lib(u) -
265: \frac{16-27u+4\ubar\pi^2}{4\ubar} \ln^2 u \no \\
266: & \quad - \frac{6-36u+27u^2-4u\ubar\pi^2}{2u\ubar} \ln u \ln \ubar +
267: \frac{3(14-17u) +8u\pi^2}{12u} \ln^2 \ubar \no \\
268: & \quad + \frac{8-15u-4\pi^2-48\ubar \zeta_3}{4\ubar} \ln u + \frac{3(2-3u)}{\ubar} \zeta_3 - \frac{23\pi^4}{60} + \frac{(23-17u)\pi^2}{12\ubar} \no \\
269: & \quad -
270: \frac{81-126u+45u^2-(14-22u+6u^2)\pi^2-192u\ubar\zeta_3}{12u\ubar}
271: \ln \ubar - \frac{137}{4}, \no \\
272: t_{E,2}(u) &= -\frac{6(1-2u)}{u\ubar}\Lib(u) -\frac{6}{u} \ln u \ln
273: \ubar -6\ln u-6\ln \ubar-\frac{\pi^2}{\ubar}+50.
274: \end{align}
275:
276:
277: \newpage
278: \section{Tree amplitudes in NNLO}
279:
280: The NNLO kernels $T_i^{(2)}$ follow from (\ref{eq:Master}) and are
281: indeed found to be free of UV and IR singularities. We emphasize
282: that this provides a very powerful check of our calculation which
283: involves the cancellation of poles up to $1/\eps^4$ ($1/\eps^3$) for
284: the calculation of the real (imaginary) part.
285:
286: \subsection{$\alpha_1$ in CMM basis}
287:
288: We now present preliminary results for the real parts of the NNLO
289: vertex corrections. Our results are still preliminary in the sense
290: that the calculation is not yet complete (massive fermion loops are
291: still missing) and we have not yet performed numerical checks of all
292: MIs. In analogy to (\ref{eq:V12mod}) we write
293: \begin{align}
294: \Re \;\hat{V}^{(1)} &\equiv
295: \int_0^1 du \; \bigg\{ -6 \ln \frac{\mu^2}{m_b^2} +g_2(u) \bigg\} \phi_{M_2}(u), \no\\
296: \Re \;\hat{V}_1^{(2)} &\equiv
297: \int_0^1 du \; \bigg\{ \Big( 36 C_F -29 N_c +2 n_f \Big) \ln^2 \frac{\mu^2}{m_b^2} \no \\
298: & \hspace{3mm} + \Big[ \Big( \frac{29}{3} N_c - \frac23 n_f \Big) g_2(u) - \frac{91}{6} N_c - \frac{10}{3} n_f + C_F \,h_6(u) \Big] \ln \frac{\mu^2}{m_b^2} \no \\
299: & \hspace{3mm} + C_F \, h_7(u) + N_c \, h_8(u) + (n_f-2) \, h_9(u;0) + h_9(u;z) + h_9(u;1) \bigg\} \phi_{M_2}(u), \no \\
300: \Re \;\hat{V}_2^{(2)} &\equiv
301: \int_0^1 du \; \bigg\{ 18 \ln^2 \frac{\mu^2}{m_b^2} +\Big(21-6\, g_2(u)\Big) \ln \frac{\mu^2}{m_b^2} + h_5(u) \bigg\}
302: \phi_{M_2}(u).
303: \end{align}
304: The NLO kernel is found to be
305: \begin{align}
306: g_2(u) &= -22 + \frac{3(1-2u)}{\ubar} \ln u + \bigg[2\Lib(u) - \ln^2
307: u - \frac{1-3u}{\ubar} \ln u -(u\to\ubar) \bigg].
308: \end{align}
309: Concerning the NNLO kernels $h_{5-9}$ we do not quote the
310: expressions for $h_5$, $h_7$ and $h_8$ here, as they are extremely
311: complicated and we have not yet expressed them in terms of a minimal
312: set of HPLs. On the other hand, the expressions for $h_6$ and for
313: $h_9$ (in the massless case) are much simpler and given by
314: \begin{align}
315: h_6(u) &= \bigg[ \frac{327}{2} - \frac{3(1-2u)}{2\ubar} \ln^2 u + \frac{3(1-2u^2)}{2u\ubar} \ln u \ln \ubar - \frac{3(13-24u)}{2\ubar} \ln u \no \\
316: &\hspace{8mm} + \frac{(1-2u^2)\pi^2}{4u\ubar} + (u\leftrightarrow\ubar)\bigg] \no \\
317: &\hspace{5mm} + \bigg[ 8 \Lic(u) -8 \ln u \, \Lib(u) + \frac43 \ln^3 u -4 \ln^2 u \ln \ubar - \frac{13-24u^2}{u\ubar} \Lib(u) \no \\
318: &\hspace{12mm} + \frac{25-24u}{2\ubar} \ln^2 u + \frac{13}{\ubar} \ln u \ln \ubar - \frac{9}{2\ubar} \ln u - \frac{11\pi^2}{6\ubar} - (u\leftrightarrow\ubar) \bigg],\no
319: \end{align}
320: \begin{align}
321: h_9(u;0) &= \bigg[ \frac{125}{12} + \frac{\Lib(u) }{\ubar} + \frac{1-3u}{2\ubar} \ln^2 u + \frac{1+u}{2\ubar} \ln u \ln \ubar -\frac{17(1-2u)}{6\ubar} \ln u \no \\
322: &\hspace{8mm} - \frac{(1+u)\pi^2}{12\ubar} + (u\leftrightarrow\ubar)\bigg] \no \\
323: &\hspace{5mm} + \bigg[ \frac43 \Lic(u) -\frac23 \ln^3 u + \frac43
324: \ln^2 u \ln \ubar -
325: \frac{32-29u}{9\ubar} \Lib(u) + \frac{35-29u}{18\ubar} \ln^2 u \no \\
326: &\hspace{12mm} -
327: \frac{1}{3\ubar} \ln u \ln \ubar - \frac{13+24\ubar
328: \pi^2}{18\ubar} \ln u + \frac{\pi^2}{18\ubar} - (u\leftrightarrow\ubar)
329: \bigg].
330: \end{align}
331:
332: \subsection{Convolution with distribution amplitude}
333:
334: We now perform the convolution integrals by expressing the
335: distribution amplitude of the emitted meson $M_2$ in terms of its
336: Gegenbauer expansion (\ref{eq:Gegenbauer}). We obtain analytical
337: results for the convolutions with the kernels $g_2$, $h_6$ and $h_9$
338: \begin{align}
339: \int_0^1 du \; g_2(u) \; \phi_{M_2}(u) &=
340: -\frac{45}{2} + \frac{11}{2} a_1^{M_2} -\frac{21}{20} a_2^{M_2}\no \\
341: \int_0^1 du \; h_6(u) \; \phi_{M_2}(u) &=
342: 348-\frac{154}{3} a_1^{M_2} + \frac{329}{40} a_2^{M_2}\no \\
343: \int_0^1 du \; h_9(u;0) \; \phi_{M_2}(u) &=
344: \frac{493}{18} - \frac{2\pi^2}{3} - \left( \frac{40}{3} + 2\pi^2
345: \right) a_1^{M_2}+ \left( \frac{8059}{600} - \pi^2 \right) a_2^{M_2},
346: \end{align}
347: and computed the remaining convolution integrals numerically
348: \begin{align}
349: \int_0^1 du \; h_5(u) \; \phi_{M_2}(u) &=
350: 322 - 213 \,a_1^{M_2} + 3.8\,a_2^{M_2} \no \\
351: \int_0^1 du \; h_7(u) \; \phi_{M_2}(u) &=
352: 731 - 348 \,a_1^{M_2} -a_2^{M_2} \no \\
353: \int_0^1 du \; h_8(u) \; \phi_{M_2}(u) &=
354: -409 + 412 \,a_1^{M_2} -32\,a_2^{M_2}.
355: \end{align}
356: The cancellation of all singularities and the finiteness of all
357: convolution integrals completes the explicit factorization proof of
358: the NNLO vertex corrections.
359:
360: \newabs
361: We finally collect all contributions and illustrate the relative
362: importance of the individual vertex corrections setting $\mu=m_b$
363: \begin{align}
364: \Re \;\hat{V}^{(1)}
365: &= -22.5 +5.5 \,a_1^{M_2} - 1.1 \,a_2^{M_2},\no \\
366: \Re \;\hat{V}_1^{(2)}
367: &= -148 +606\,a_1^{M_2}-80\,a_2^{M_2},\no \\
368: \Re \;\hat{V}_2^{(2)}
369: &= 322-213\,a_1^{M_2}+3.8\,a_2^{M_2}.
370: \label{eq:V12:Nums:Re}
371: \end{align}
372:
373:
374: \subsection{Preliminary numerical result}
375:
376: We conclude this chapter with the presentation of a numerical result
377: for the real part of the NNLO vertex corrections. We stress again
378: that this corresponds to a preliminary result which treats the
379: $c$-quark and the $b$-quark in the closed fermion loops as massless
380: quarks. If we reconsider our results for the imaginary part in this
381: approximation, we find deviations of $\sim5\%$ of the individual
382: NNLO contributions. As the NNLO terms are subleading for the real
383: part of the colour-allowed tree amplitude, we expect that this
384: approximation will have only a minor impact here.
385:
386: \newabs
387: Our preliminary result for the real part of the NNLO vertex
388: corrections reads
389: \begin{align}
390: \Re\; \alpha_1(\pi\pi)& =\,~~ 1.01 \big{|}_{V^{(0)}}
391: + 0.03 \big{|}_{V^{(1)}}
392: + 0.03 \big{|}_{V^{(2)}} \no \\
393: & =\,~~ 1.06,
394: \end{align}
395: where we have used Wilson coefficients in NLL approximation for
396: simplicity (they are indeed known to the required NNLL accuracy and
397: can be found in \cite{GorHai}). As expected, the contribution is of
398: minor importance in absolute terms. However, the NNLO corrections
399: are found to be as important as the NLO terms which are numerically
400: suppressed by the small Wilson coefficient. Interestingly, the
401: vertex corrections add again constructively and, as can be seen in
402: comparison with the results from~\cite{BenekeJager}, come again with
403: the opposite of the spectator interactions.
404:
405: \newabs
406: The colour-suppressed tree amplitude $\alpha_2(\pi\pi)$ is
407: phenomenologically more interesting as the respective QCD
408: Factorization prediction is rather low for a satisfactory
409: description of the experimental data. In order to derive the NNLO
410: result for $\alpha_2$ we still have to solve some conceptual aspects
411: concerning a Fierz-symmetric definition of (2-loop) evanescent
412: operators. We therefore relegate the discussion of the
413: colour-suppressed tree amplitude to~\cite{GBRePart}.
414: