hep-ph0507078/cvc.tex
1: \section{\boldmath COMPARING TAU SPECTRAL FUNCTIONS WITH ELECTRON-POSITRON ANNIHILATION DATA VIA THE CONSERVED VECTOR CURRENT}
2: \label{sec:cvc}
3: 
4: The vacuum polarization functions~(\ref{eq:correlator}) for 
5: the various types of quark currents $ij$ carry the dynamical information
6: for the theoretical evaluation of total cross sections and decay widths.
7: In the limit of isospin invariance the vector current is conserved
8: (CVC), so that the \sf\  of a vector $\tau$ decay mode $X^-\nut$
9: in a given isospin state for the hadronic system is related 
10: to the \ee  annihilation cross section of the corresponding 
11: isovector final state $X^0$
12: \beq
13: \label{eq:cvc}
14:   \sigma_{e^+e^-\to X^0}^{I=1}(s) \:=\:
15:          \frac{4\pi\alpha^2}{s}\,v_{1,\,X^-}(s)~,
16: \eeq
17: where $\alpha$ is the electromagnetic fine structure constant.
18: 
19: \subsection{Isospin breaking and radiative corrections}
20: \label{sec:cvc_isobreak}
21: 
22: Since the breaking of isospin symmetry is expected at some level, 
23: in particular from electromagnetic effects, it is useful to 
24: carefully write down all the factors involved in the 
25: comparison of $\tau$ and \ee  spectral functions in order to make
26: explicit the possible sources causing the breakdown of CVC. For the 
27: dominant $\pi\pi$ spectral functions, we have on the \ee  side
28: \beq
29: \label{eq:ee_ff}
30: 	\sigma (\epem\to \pi^+\pi^-)
31: 	=	\frac{4\pi\alpha^2}{s} v_0(s)~,
32: \eeq
33: where the spectral function $v_0(s)$ is related to the pion form 
34: factor $F^0_\pi(s)$ by
35: \beq
36:     	v_0(s)=	\frac {\beta_0^3(s)} {12} |F^0_\pi(s)|^2~,
37: \eeq
38: and where $\beta_0^3(s)$ is the threshold kinematic factor. Similarly,
39: on the $\tau$ side the relation between spectral function~(\ref{eq:sf}) 
40: and the charged pion form factor reads
41: \beq
42: \label{eq:tau_ff}
43:       	v_-(s) = \frac {\beta_-^3(s)} {12} |F^-_\pi(s)|^2~.
44: \eeq
45: Isospin symmetry implies $v_-(s) =  v_0(s)$. The threshold functions 
46: $\beta_{0,-}$ are defined by
47: \beq
48:  	\beta_{0,-} = \beta(s,m_{\pi^-},m_{\pi^{+,0}})~,
49: \eeq
50: where
51: \beq
52:  \beta(s,m_1,m_2)=\left[\left(1-\frac{(m_1+m_2)^2}{s}\right)
53:                      \left(1-\frac{(m_1-m_2)^2}{s}\right)\right]^{1/2}~.
54: \eeq
55: 
56: The experiments measure $\tau$ decays inclusively with respect to radiative 
57: photons, \ie, for $\taum \to \nu_\tau \pim \piz (\gamma)$. The measured spectral 
58: function is hence $v_-^*(s) =  v_-(s)~G(s)$, where $G(s)$ is a radiative 
59: correction.
60: 
61: Several levels of ${\rm SU}(2)$ breaking can be identified:
62: \bei
63: \item 	{\em Electroweak radiative corrections to $\tau$ decays} 
64: 	are contained in the $\Sew$ 
65: 	factor~\cite{Marciano:1988,braaten}, which is dominated by 
66: 	short-distance effects. It is expected to be weakly 
67: 	dependent on the specific hadronic final state, as verified 
68: 	for the $\taum \to (\pim, \Km) \nut$ decays~\cite{decker-fink}. 
69: 	Detailed calculations have been performed for the 
70: 	$\pim \pi^0$ channel~\cite{ecker,ecker2}, which also confirm the 
71: 	relative smallness of the long-distance contributions. The 
72: 	total correction is
73: 	$\Sew = \Sew^{\rm had} \Sem^{\rm had}/\Sem^{\rm lep}$,
74: 	where $\Sew^{\rm had}$ is the leading-log short-distance electroweak
75: 	factor (which vanishes for leptons) and $\Sem^{\rm had,lep}$ are the
76: 	nonleading electromagnetic corrections. The latter corrections
77: 	have been calculated at the quark level~\cite{braaten}, 
78: 	at the hadron level for the $\pim \pi^0$ decay mode~\cite{ecker},
79: 	and for leptons~\cite{Marciano:1988,braaten}. The 
80: 	total correction amounts to~\cite{Davier:2003a} 
81: 	$\Sew^{\rm incl} = 1.0198 \pm 0.0006$ 
82: 	for the inclusive hadron decay rate and 
83: 	$\Sew^{\pi \pi^0} = (1.0232 \pm 0.0006)\cdot\Gem^{\pi \pi^0}(s)$ 
84: 	for the $\pi \pi^0$ decay mode, where $\Gem^{\pi \pi^0}(s)$ is an
85: 	$s$-dependent long-distance radiative correction~\cite{ecker}.
86: 
87: \item 	{\em The pion mass splitting} breaks isospin symmetry in 
88: 	the spectral functions~\cite{adh,czyz} since $\beta_-(s) \neq \beta_0(s)$.
89: 
90: \item 	Isospin symmetry is also broken in {\em the pion form factor} due
91: 	to the $\pi$ mass splitting~\cite{adh,ecker}.
92: 
93: \item 	A similar effect is expected from {\em the $\rho$ mass splitting}. 
94: 	The theoretical expectation~\cite{bijnens} gives a limit ($<0.7$~MeV), 
95: 	but this is only a rough estimate. Hence the question must be 
96: 	investigated experimentally, the best approach being the explicit 
97: 	comparison of $\tau$ and $\epem$ $2\pi$ spectral functions, after 
98: 	correction for the other isospin-breaking effects. No correction 
99: 	for $\rho$ mass splitting is applied initially.
100: 
101: \item 	Explicit {\em electromagnetic decays} such as $\pi \gamma$, 
102: 	$\eta \gamma$, $\ell^+\ell^-$ and $\pi \pi \gamma$ introduce
103: 	small differences between the widths of the charged and neutral $\rho$'s.
104: 
105: \item 	Isospin violation in the strong amplitude through the {\em mass 
106: 	difference between $u$ and $d$ quarks} is expected to be negligible.
107: 
108: \item 	When comparing $\tau$ with $e^+e^-$ data, an obvious and
109:         locally large correction must be applied to the $\tau$ \sf\
110: 	to introduce the effect of {\em$\rho$--$\omega$ mixing}, 
111: 	only present in the
112: 	neutral channel. This correction is computed using the parameters 
113: 	determined by the \ee experiments in their form factor fits to the 
114: 	$\pi^+ \pi^-$ lineshape modeling $\rho$--$\omega$ 
115: 	interference~\cite{cmd2_new}. 
116: 
117: \eei
118: 
119: \subsection{Tau spectral functions and $e^+e^-$ data}
120: \label{sec:cvceetau}
121: 
122: Data from $\tau$ decays into two- and four-pion final states
123: \tauto\nut\pipiz, \tauto\nut\pitpiz\ and \tauto\nut\tpipiz,
124: are available from ALEPH~\cite{aleph_taubr}, 
125: CLEO~\cite{cleotaurho,cleo_4pi} and OPAL~\cite{opal_vasf}.
126: For the two-pion final states, they are compared in 
127: Fig.~\ref{comp_tau_2pi} in the region around the $\rho$ resonance.
128: For this comparison, each data set is normalized to the world average
129: branching fraction and plotted with respect to their weighted average.
130: The two most precise results from ALEPH and CLEO are in agreement.
131: The statistics are comparable in the two cases, however due to a flat
132: acceptance in ALEPH and a strongly increasing one (with the mass-squared)
133: in CLEO, ALEPH data are more precise below the $\rho$ peak, while CLEO 
134: has the better precision above. In the following we will use the 
135: $\taum\to\pim\piz\nut$ \sf\  averaged over the three experiments.
136: \begin{figure}[t]
137:    \centerline{\epsfxsize8.3cm\epsffile{figures/comp_tau.eps}
138: 	       \epsfxsize8.3cm\epsffile{figures/comp_tau_zoom.eps}}
139:    \vspace{-0.3cm}
140:    \caption[.]{The left figure shows a relative comparison of 
141:         the $\pi^+\pi^-$ \sfs\
142:     	extracted from $\tau$ data from the different experiments, 
143: 	expressed as a ratio to the average $\tau$ \sf. 
144:         The right figure emphasizes 
145:         the $\rho$ region. For CLEO only the statistical errors are shown.} 
146: \label{comp_tau_2pi}
147: \end{figure}
148: 
149: The exclusive low-energy \ee cross sections have been mainly measured by 
150: experiments running at \ee colliders in Novosibirsk and Orsay. 
151: The most precise $\ee\to\pi^+\pi^-$ measurements come from 
152: CMD-2, which are now available in their final form~\cite{cmd2_new}, after 
153: a significant revision where problems related to radiative corrections 
154: have been removed. 
155: %The results are corrected for leptonic and hadronic 
156: %vacuum polarization, and for photon radiation by the pions (final state 
157: %radiation, FSR), so that the measured final state corresponds to 
158: %$\pi^+\pi^-$ including pion-radiated photons and virtual final state QED 
159: %effects. 
160: The overall systematic error of the final data is quoted to be 
161: 0.6\% and is dominated by the uncertainties in the radiative corrections 
162: (0.4\%). 
163:  
164: %The comparison between the cross section results from CMD-2 and from 
165: %previous experiments (corrected for vacuum polarization and FSR,
166: %according to the procedure discussed in Section~\ref{sec:rad}) shows
167: %agreement within the much larger uncertainties (2--10\%) quoted 
168: %by the older experiments. But the new CMD-2 results only cover the mass
169: %range from 0.61 to 0.96\gev, so the older data must still be relied 
170: %upon below and above these values. See Section~\ref{sec:cvc} for 
171: %detailed numerical and graphical comparisons.
172: 
173: Recently, new data on the $\pip\pim$ spectral function
174: in the mass region between $0.60$ and $0.97\gev$ were presented by
175: the KLOE collaboration~\cite{kloe_2pi}, using the---for the purpose
176: of precision measurements---innovative technique 
177: of the radiative return~\cite{isr1,isr2}. The statistical precision 
178: of these data by far outperforms the Novosibirsk sample, but the
179: systematic errors are about twice as large as those assigned by CMD-2.
180: 
181: \subsection{Comparing $\tau$ with $e^+e^-$ data}
182: \label{sec:comp_eetau_2pi}
183: 
184: \subsubsection{The $2\pi$ spectral function}
185: 
186: The $2\pi$ \sfs\  extracted from $\tau$ decays and $\ee$ annihilation data 
187: are compared in Fig.~\ref{fig:tau_ee_2pi}. The \ee  data
188: are taken from~\cite{tof_2pi, olya_2pi, cmd_2pi, cmd2_new, dm1_2pi, dm2_2pi}. 
189: All error bars shown contain statistical and systematic errors.
190: Visually, the agreement appears to be satisfactory, however the large 
191: dynamical range involved does not permit an accurate test. To do so, 
192: the \ee  data are plotted as a point-by-point ratio to the $\tau$  
193: \sf\  in Fig.~\ref{fig:comp_eetau_2pi}. In these latter plots we also show
194: more recent \epem-annihilation results from the SND collaboration~\cite{snd:2005}, 
195: and from the radiative return analysis performed by the KLOE 
196: collaboration~\cite{kloe_2pi}. Several observations can be made:
197: \begin{figure}[t]
198:   \centerline{\epsfxsize14.3cm\epsffile{figures/2pi_eetau.eps}}
199:   \vspace{0.1cm}
200:    \caption[.]{Comparison of the $\pi^+\pi^-$ \sfs\ from 
201:         isospin-breaking corrected $\tau$ data (world average) 
202:         expressed as \ee cross sections, and \ee  annihilation. 
203:         The band indicates 
204: 	the combined $\tau$ and \ee result within $1\sigma$ errors.
205: 	It is given for illustration purpose only. A compendium of 
206: 	references for
207:         the \ee data is given in~\cite{Davier:2003a, Davier:2003b}.}
208: \label{fig:tau_ee_2pi}
209: \end{figure}
210: \bei
211: 
212: \item 	A significant discrepancy, mainly above the $\rho$ peak
213: 	is found between $\tau$ and the \epem data from CMD-2 as well	
214: 	as older data from OLYA.
215: 
216: \item	Overall, the KLOE data seem to confirm the trend exhibited by the 
217: 	other (older) \epem data.
218: 
219: \item	Some disagreement between KLOE and CMD-2
220: 	occurs on the low mass side (KLOE data are large),
221:   	on the $\rho$ peak (KLOE below CMD-2) as well as
222: 	on the high mass side (KLOE data are low).
223: 
224: \item 	A significant discrepancy between the (most recent) SND data and 
225: 	KLOE is observed in the energy domain above the $\rho(770)$ peak.
226: 	The SND data largely dissolve the discrepancy observed
227: 	with the $\tau$ data.
228: 
229: \eei
230: 
231: At this stage, the $\tau$ spectral function has not been corrected 
232: for a possible $\rho^-$--$\rho^0$ mass and width 
233: splitting~\cite{gj,davier_pisa_g-2,aleph_taubr}.
234: In contrast to earlier experimental~\cite{aleph_vsf} and 
235: theoretical results~\cite{bijnens}, a combined pion form 
236: factor fit to the new precise data on $\tau$ spectral 
237: functions and \epem leads to $m_{\rho^-}-m_{\rho^0}=(2.4\pm0.8)\mev$
238: (Section~\ref{sec:eetau_combinedfit}), while no significant width 
239: splitting$^{\,}$ is observed within the fit error of $1.0\mev$.
240: 
241: Considering the mass splitting in the isospin-breaking correction 
242: of the $\tau$ spectral function tends to locally improve though 
243: not restore the agreement between $\tau$ and CMD-2 data, leaving 
244: an overall normalization discrepancy. Increasing the
245: $\Gamma_{\rho^-}-\Gamma_{\rho^0}$ width splitting by $+3\mev$
246: improves the agreement between $\tau$ and KLOE data in the 
247: peak region, while it cannot correct the discrepancies in
248: the tails. Note that a correction of the mass splitting 
249: alone would {\em increase} the discrepancy between the $\tau$ 
250: and \epem-based  results for \amuhadLO (Section~\ref{sec:g-2com}).
251: 
252: %No convincing theoretical solution for a $\tau$-vs-$\ee$ discrepancy
253: %has been suggested to date. A possible $S$-wave contribution to the $\tau$ 
254: %decay amplitude \via\  exchange of a charged Higgs boson~\cite{morse} has 
255: %been found too small to explain the observation~\cite{daonenggao}.
256: %The possibility of contributions from tensor couplings introduced 
257: %by a new force has been outlined in~\cite{chizhov:2003}.
258: \begin{figure}[t]
259:   \centerline{\epsfxsize8.3cm\epsffile{figures/compee_tau_withkloe.eps}
260: 	      \epsfxsize8.3cm\epsffile{figures/compee_tau_zoom_withkloe.eps}}
261:   \vspace{-0.3cm}
262:   \caption[.]{Relative comparison of the isospin-breaking corrected $\tau$
263:         data (world average) and $\pi^+\pi^-$ \sfs\ from \ee annihilation,
264: 	expressed as a ratio to the $\tau$ \sf. The shaded band gives the 
265:         uncertainty on the $\tau$ spectral function. The right hand plot 
266: 	zooms into the $\rho(770)$ peak region where the data density is 
267: 	high. The $\rho$--$\omega$ mixing has been phenomenologically 
268: 	corrected using a fit to the CMD-2 data. The larger mixing observed
269: 	in the SND data, expressed as a $40\%$ increase in the value for the 
270: 	$\omega\to\pip\pim$ \br~\cite{snd:2005}, is responsible for the 
271: 	residual effect after correction seen in the right hand plot.
272: 	}
273: \label{fig:comp_eetau_2pi}
274: \end{figure}
275: \begin{figure}[t]
276:   \vspace{-1.1cm}
277:   \centerline{\epsfxsize8.3cm\epsffile{figures/4pi_eetau.eps}
278:               \epsfxsize8.3cm\epsffile{figures/2pi2pi0_eetau.eps}}
279:   \vspace{-0.3cm}
280:   \caption{Comparison of the $2\pi^+2\pi^-$ (left) and $\pi^+\pi^- 2\pi^0$
281: 	(right) \sfs\  from \ee and isospin-breaking corrected $\tau$ 
282: 	data (ALEPH), expressed as \ee cross sections. 
283: 	References for the \ee data are given in~\cite{Davier:2003a, Davier:2003b}.
284:         The \babar\  results are taken from~\cite{babar_4pi}.}
285: \label{fig:comp_4pi_eetau}
286: \end{figure}
287: 
288: In light of the new SND data~\cite{snd:2005}, it seems appropriate
289: to consider the possibility of a bias in the KLOE results, and one may
290: also question the very small systematic uncertainties claimed by CMD-2.
291: We also point out that the application of the $\tau$ 
292: long-distance correction $\Gem^{\pi \pi^0}(s)$~\cite{ecker} worsens 
293: the agreement with the SND data in the energy domain above the $\rho(770)$ 
294: peak.
295: 
296: \subsubsection{The $4\pi$ spectral functions}
297: \label{sec:4pisf}
298: 
299: The \sf\ measurements of the $\tau$ vector-current final states \pitpiz\  
300: and \tpipiz\  are compared to the cross sections of the corresponding \ee 
301: annihilation into the isovector states $2\pi^- 2\pi^+$ and $\pi^- \pi^+ 2\pi^0$. 
302: Using Eq.~(\ref{eq:spect_fun}) and isospin invariance, the following relations 
303: hold
304: \begin{eqnarray}
305: \label{eq:cvc_4pi}
306:  \sigma_{e^+e^-\rightarrow\pi^+\pi^-\pi^+\pi^-}^{I=1} 
307:         & = &
308:              2\cdot\frac{4\pi\alpha^2}{s}\,
309:              v_{1,\,\pi^-\,3\pi^0\,\nut}~, \\[0.3cm]
310: \label{eq:cvc_2pi2pi0}
311:  \sigma_{e^+e^-\rightarrow\pi^+\pi^-\pi^0\pi^0}^{I=1} 
312:         & = &
313:              \frac{4\pi\alpha^2}{s}\,
314:              \left[v_{1,\,2\pi^-\pi^+\pi^0\,\nut} 
315:                   \:-\:
316:                      v_{1,\,\pi^-\,3\pi^0\,\nut}
317:              \right]~.
318: \end{eqnarray}
319: The comparison of these cross sections is given in 
320: Fig.~\ref{fig:comp_4pi_eetau} for $2\pi^+2\pi^-$ (left hand 
321: plot) and $\pi^+\pi^-2\pi^0$ (right hand plot). The latter mode
322: suffers from discrepancies between the results from the various \ee
323: experiments. The $\tau$ data, combining two measured \sfs\ according to 
324: Eq.~(\ref{eq:cvc_2pi2pi0}), and (coarsely) corrected for isospin breaking 
325: originating from the pion mass splitting~\cite{czyz}, 
326: lie somewhat in between, with large uncertainties above $1.4\gev$ because 
327: of the lack of statistics and a large feed-through background in the 
328: mode with three neutral pions. In spite of these difficulties the  
329: $\pi^-3\pi^0$ \sf\  is in agreement with \ee  data as can be seen in 
330: Fig.~\ref{fig:comp_4pi_eetau}. Due to the inconsistencies among the \ee 
331: experiments a quantitative test of CVC in the $\pi^+\pi^-2\pi^0$ channel 
332: is premature.
333: 
334: \subsubsection{Branching fractions in $\tau$ decays and CVC}
335: \label{sec_brcvc}
336: 
337: It is instructive to compare the $\tau$ and \ee spectral functions 
338: in a more quantitative way by calculating weighted integrals
339: over the mass range of interest up to the $\tau$ mass. 
340: One convenient choice is provided by the $\tau$ branching fractions,
341: which for the spin one part involve as a weight the kinematic factor 
342: $(1-s/m_\tau^2)^2(1+2s/m_\tau^2)$
343: coming from the $V-A$ charged current in $\tau$ decay. It is then 
344: possible to directly compare the measured $\tau$ \brs
345: to their prediction through isospin invariance (CVC), with 
346: the \ee  isovector \sfs\  as input.
347:  
348: Using the universality-improved branching fraction~(\ref{eq:uni_be}), 
349: one finds the results given for the dominant channels 
350: in Table~\ref{tab:brcvc}. The errors quoted for the CVC values are split
351: into uncertainties from ({\it i}) the experimental input (the \ee annihilation 
352: cross sections) and the numerical integration procedure,
353: ({\it ii}) additional radiative corrections applied to some of the \ee 
354: data~\cite{Davier:2003a}, and ({\it iii}) the isospin-breaking corrections 
355: when relating $\tau$ and \ee  \sfs. 
356: 
357: \begin{table}[t]
358: \caption{\label{tab:brcvc}
359: 	Branching fractions of $\tau$ vector decays into 2 and 4 pions in
360:         the final state~\cite{aleph_taubr}. Second column: $\tau$ decay
361: 	measurements. Third column: results inferred from \ee  spectral
362: 	functions using the isospin 
363: 	relations~(\ref{eq:cvc},\ref{eq:tau_ff},\ref{eq:cvc_4pi}) and
364:         correcting for isospin breaking~\cite{Davier:2003a, Davier:2003b}. 
365: 	The 2 pion prediction does not include the recent KLOE and SND data.
366: 	Experimental errors, including uncertainties on the integration 
367:         procedure, and theoretical errors (missing radiative corrections 
368: 	for \ee, and isospin-breaking corrections and $|V_{ud}|$ for 
369: 	$\tau$) are shown separately. Last column: differences between the
370: 	direct measurements in $\tau$ decays and the CVC evaluations,
371:         where the various errors have been added in quadrature.}
372: \begin{center}
373: \setlength{\tabcolsep}{0.0pc}
374: \begin{tabular*}{\textwidth}{@{\extracolsep{\fill}}lrrr} \hline 
375: &&& \\[-0.3cm]
376: 		& \mc{3}{c}{Branching fractions  (in \%)} \\
377: \rs{Mode} 	& \mc{1}{c}{$\tau$ } 	
378: 		& \mc{1}{c}{$e^+e^-$ \via\  CVC} & $\Delta(\tau-e^+e^-)$ 
379: \\[0.15cm]
380: \hline
381: &&& \\[-0.3cm]
382: $~\tau^-\to\nut\pi^-\pi^0$
383: 		& $25.47 \pm 0.13$ 
384: 		& $24.52 \pm \underbrace{0.26_{\rm exp}	
385: 			\pm 0.11_{\rm rad}\pm 0.12_{\rm SU(2)}}_{0.31}$ 
386: 		& $+0.95 \pm 0.33$ 
387: 	\\[0.0cm]
388: $~\tau^-\to\nut\pi^-3\pi^0$
389: 		& $ 0.98 \pm 0.09$ 
390: 		& $ 1.09 \pm \underbrace{0.06_{\rm exp}
391: 		        \pm 0.02_{\rm rad}\pm 0.05_{\rm SU(2)}}_{0.08}$ 
392: 		& $-0.11 \pm 0.12$ 
393: 	\\[0.0cm]
394: $~\tau^-\to\nut2\pi^-\pi^+\pi^0$
395: 		& $ 4.59 \pm 0.09$ 
396: 		& $ 3.63 \pm \underbrace{0.19_{\rm exp}
397: 			\pm 0.04_{\rm rad}\pm 0.09_{\rm SU(2)}}_{0.21}$ 
398: 		& $+0.96 \pm 0.23$ 
399: 	\\[0.5cm]
400: \noalign{\smallskip}\hline
401: \end{tabular*}
402: \end{center}
403: \end{table} 
404: As expected from the preceding discussion, a discrepancy is 
405: observed for the $\taum\to \pim\piz\nut$ \br, with a difference of 
406: $(0.95\pm0.13_\tau\pm0.26_{\rm ee}\pm0.11_{\rm rad}\pm0.12_{\rm SU(2)})\%$, 
407: where the uncertainties are from the $\tau$ \br, 
408: \ee  cross section, \ee  missing radiative corrections and isospin-breaking 
409: corrections (also including the uncertainty on $|V_{ud}|$), respectively. 
410: Adding all errors in quadrature gives a $2.9\sigma$ effect. Including
411: the new SND measurements~\cite{snd:2005} into the $\pip\pim$ data sample 
412: would decrease this discrepancy to approximately $2.5\sigma$. The improvement
413: would have been stronger, if we had discarded older data sets. In effect,
414: using only SND data in the region where these are available 
415: $[0.39$--$0.97\gev]$, and all other data elsewhere, one finds for
416: the predicted \br: $\BR_{\rm CVC}(\tau^-\to\nut\pi^-\pi^0)=(25.12\pm0.36)\%$, 
417: which is in agreement with the $\tau$ result\footnote
418: {
419: 	However, we point out that the scarce SND data points in the tails 
420: 	of the $\rho(770)$ resonance, where the line shape is concave, 
421: 	may lead to an overestimate of the integral computed by the 
422: 	trapezoidal rule as it is done here.
423: }. 
424: More information on this comparison is displayed in 
425: Fig.~\ref{fig:cvc_2pi}, where we also give the results for using
426: only CMD-2 and only KLOE data, respectively, in the available energy 
427: regions (and the average of all other data elsewhere). The discrepancy
428: between the \epem-based prediction using the KLOE data and 
429: the $\tau$ result is at the $3.8\sigma$ level.
430: 
431: The situation in the $4\pi$ channels is different. While there is agreement 
432: for the $\pi^-3\pi^0$ mode within a relative accuracy of $12\%$, 
433: the comparison is not satisfactory for $2\pi^-\pi^+\pi^0$. 
434: In the latter case, the relative difference is very large, 
435: $(23\pm6)\%$, compared to a reasonable level of isospin symmetry 
436: breaking. As such, it rather points to experimental problems that require
437: further investigation. These are emphasized by the scatter observed among 
438: the different \ee  results.
439: \begin{figure}[t]
440:   \centerline{\epsfxsize8.6cm\epsffile{figures/brpipi0.eps}}
441:   \vspace{0.1cm}
442:   \caption[.]{The measured branching fractions for 
443:        	$\taum\to\pim\piz\nut$~\cite{aleph_taubr,cleo_bhpi0,opal_bh} compared 
444: 	to the predictions from the $e^+e^-\to\pi^+\pi^-$ \sfs, applying the 
445: 	isospin-breaking correction factors discussed in 
446: 	Section~\ref{sec:cvc_isobreak}. 
447: 	For the \epem results we have used only the data from the 
448: 	indicated experiments where they are available ($0.61$--$0.96\gev$ 
449: 	for CMD-2, $0.60$--$0.97\gev$ for KLOE, and $0.39$--$0.97\gev$
450: 	for SND), and all data combined in the remaining energy 
451: 	domains below $m_\tau$. Since the compatibility between 
452: 	the \epem data is marginal (\cf\  Fig.~\ref{fig:comp_eetau_2pi}),
453: 	we have refrained from taking their average.
454: 	}
455: \label{fig:cvc_2pi}
456: \end{figure}
457: 
458: \subsection{Fits to the $\pi\pi$ spectral function}
459: 
460: Phenomenological fits to the pion form factor have been performed by the 
461: ALEPH and CLEO collaborations~\cite{aleph_taubr,aleph_vsf,cleotaurho}.
462: Appropriate parameterizations, \eg, the one proposed in~\cite{kuhnsanta},
463: employ relativistic $P$-wave Breit-Wigner propagators, and coherent summation 
464: of the resonance amplitudes using the isobar model.
465: The $\pi\pi$ spectral function is dominated by the broad $\rho$ resonance,
466: parameterized by both collaborations following Gounaris-Sakurai~\cite{gounarissak} 
467: (GS). The GS parameterization takes into account analyticity and unitarity 
468: properties. We will in the following discuss the fits performed in~\cite{aleph_taubr}.
469: 
470: Assuming vector dominance, the pion form factor is given by interfering
471: amplitudes from the known isovector meson resonances $\rho(770)$, 
472: $\rho(1450)$ and $\rho(1700)$ with relative strengths 1, $\beta$, and $\gamma$. 
473: Although one could expect from the quark model that $\beta$ and $\gamma$ are real
474: and respectively negative and positive, the phase of $\beta$, $\phi_\beta$, is left 
475: free in the fits, and only the much smaller $\gamma$ is assumed to be real for lack 
476: of precise experimental information at large masses. Taking into account 
477: $\rho$--$\omega$ mixing, one writes for the $\pi^+\pi^-$ form factor in 
478: \ee annihilation
479: \beq\label{fpi_vdm}
480: F^{I=1,0}_\pi(s) 
481:                  \:=\; 
482:                     \frac{{\rm BW}_{\rho(770)}(s)\,
483:                     \frac{1+\delta\,{\rm BW}_{\omega(783)}(s)}
484:                          {1+\delta} \:+\:                            
485:                              \beta\,{\rm BW}_{\rho(1450)}(s) \:+\: 
486:                              \gamma\,{\rm BW}_{\rho(1700)}(s)}
487:                          {1\:+\:\beta\:+\:\gamma}~,
488: \eeq
489: with the Breit-Wigner propagators
490: \beq\label{eq_BW}
491: {\rm BW}^{\rm GS}_{\rho(m_\rho)}(s) 
492:                = 
493:           \frac{m_\rho^2(\,1+d\cdot\Gamma_\rho/m_\rho)}
494:                {m_\rho^2\:-\:s\:+\:f(s)\:-\:i\sqrt{s}\,\Gamma_\rho(s)}~,
495: \eeq
496: where
497: \beq
498: f(s) = \Gamma_\rho \frac{m_\rho^2}{k^3(m_\rho^2)}\,
499:            \Bigg[ 
500:  \, k^2(s) \left( h(s)-h(m_\rho^2)\right) \:+\:
501:                  (\,m_\rho^2-s)\,k^2(m_\rho^2)\,
502:                    \frac{d h}{d s}\bigg|_{s=m_\rho^2}
503:          \,\Bigg] 
504: \eeq
505: and $\Gamma_\rho=\Gamma_\rho (m_\rho^2)$~.
506: The $P$-wave energy-dependent width is given by
507: \beq\label{width}
508: \Gamma_\rho(s) = \Gamma_\rho \,
509:                     \frac{m_\rho}{\sqrt{s}}
510:                     \left(\frac{k(s)}{k(m_\rho^2)}\right)^{\!\!3}~,
511: \eeq
512: where $k(s)=\frac{1}{2}\,\sqrt{s}\,\beta^-_\pi(s)$ and $k(m_\rho^2)$ are 
513: pion momenta in the $\rho$ rest frame. The function $h(s)$ is defined as
514: \beq
515: h(s) = \frac{2}{\pi}\,\frac{k(s)}{\sqrt{s}}\,
516:            {\rm ln}\frac{\sqrt{s}+2k(s)}{2m_\pi}~,
517: \eeq
518: with $dh/ds|_{m_\rho^2} = 
519: h(m_\rho^2)\left[(8k^2(m_\rho^2))^{-1}-(2m_\rho^2)^{-1}\right]
520: \,+\, (2\pi m_\rho^2)^{-1}$.
521: Since interference with the isospin-violating electromagnetic 
522: $\omega\rightarrow\pi^+\pi^-$ decay occurs only in \ee annihilation
523: $\delta$ is fixed to zero when fitting $\tau$ data.
524: The normalization BW$^{\rm GS}_{\rho(m_\rho)}(0)= 1$ fixes the
525: parameter $d=f(0)/(\Gamma_\rho m_\rho)$, which is found to 
526: be~\cite{gounarissak}
527: \beq
528: d = \frac{3}{\pi}\frac{m_\pi^2}{k^2(m_\rho^2)}\,
529:         {\ln}\frac{m_\rho+2k(m_\rho^2)}{2m_\pi} \:+\:
530:         \frac{m_\rho}{2\pi\,k(m_\rho^2)} \:-\: 
531:         \frac{m_\pi^2 m_\rho}{\pi\,k^3(m_\rho^2)}~.
532: \eeq
533: 
534: \subsubsection{Fit to $\tau$ data}
535: 
536: The fit of the GS model to the isovector $\tau$ data establishes the need
537: for the $\rho(1450)$ contribution to the weak pion form 
538: factor~\cite{aleph_taubr,aleph_vsf}. Only weak evidence is found for a 
539: $\rho(1700)$ contribution lying close to the $\tau$ end-point. 
540: Most of the fit parameters exhibit large correlations, which have
541: to be taken into account when interpreting the results.
542: The $\rho$ mass uncertainty is found to be dominated by systematic effects, 
543: the largest being the knowledge of the $\pi^0$ energy scale (calibration).
544: 
545: \subsubsection{Combined fit to $\tau$ and $e^+e^-$ data}
546: \label{sec:eetau_combinedfit}
547: 
548: The ALEPH collaboration has performed a combined fit using  
549: $\tau$ (ALEPH and CLEO) and \ee data (not yet including KLOE and SND)
550: in order to better constrain the lesser known parameters in the 
551: phenomenological form factor~\cite{aleph_taubr,aleph_vsf}. 
552: For this study, the $\tau$ \sf\  is duly corrected for the 
553: isospin-breaking effects identified in Section~\ref{sec:cvc_isobreak}. Also 
554: photon vacuum polarization contributions are removed in the \ee \sf\ 
555: since they are absent in the $\tau$ data. In this way, the mass and width 
556: of the dominant $\rho(770)$ resonance in the two isospin states can be 
557: determined. For the subleading amplitudes from the higher
558: vector mesons, isospin symmetry is assumed so that common masses 
559: and widths can be used in the fit.
560: 
561: The result of the combined fit is given in Table~\ref{tab:rho_common_fit}. 
562: The differences between the masses and widths of the charged 
563: and neutral $\rho(770)$'s are found to be~\cite{aleph_taubr}
564: \beqn
565: \label{eq:splitting}
566:    m_{\rho^-}-m_{\rho^0}		&=& (2.4 \pm 0.8)\mev~, \\
567:    \Gamma_{\rho^-}-\Gamma_{\rho^0}	&=& (0.2 \pm 1.0)\mev~.
568: \eeqn
569: \begin{figure}[t]
570:    \centerline{\epsfxsize8.3cm\epsffile{figures/2pi_fit_tau.eps}\hspace{0.1cm}
571: 	       \epsfxsize8.3cm\epsffile{figures/2pi_fit_ee.eps}}
572:    \vspace{-0.3cm}
573:    \caption{Phenomenological fit to the ALEPH \pipiz\ spectral function 
574: 	(left) and to \ee  annihilation data (right) using the 
575: 	Gounaris-Sakurai parameterization. The fits are taken 
576: 	from~\cite{aleph_taubr}.}
577: \label{rho_fit_aleph}
578: \end{figure}
579: The mass splitting is somewhat larger than the theoretical prediction 
580: ($<0.7$ MeV)~\cite{bijnens}, but only at the $2\sigma$ level. The expected
581: width splitting, from known isospin breaking, but not taking into account any 
582: $\rho$ mass splitting, is $(0.7\pm0.3)\mev$~\cite{ecker,Davier:2003a}. 
583: However, if the mass difference is taken as an experimental fact, a larger 
584: width difference would be expected. From the chiral model of the $\rho$ 
585: resonance~\cite{pich-portoles1,pich-portoles2,ecker}, one expects
586: \begin{table}[t]
587: \caption{Combined fit to the pion form factor-squared to ALEPH, CLEO 
588:       	$\tau$ and all \ee data, where vacuum polarization and known 
589: 	sources of isospin violation have been 
590: 	excluded from the latter data. The parameterization of the $\rho(770)$, 
591: 	$\rho(1450)$ and $\rho(1700)$ line shapes follows the Gounaris-Sakurai 
592: 	formula. Separate masses and widths are fit for the $\rho(770)$, 
593: 	while common values are kept for the higher vector mesons. All mass 
594: 	and width values are in units of \mev and the phases are 
595:         in degrees. The value of the $\chi^2$ estimator per degree of 
596: 	freedom (DF) is also quoted.}
597: \label{tab:rho_common_fit}
598: \begin{center}
599: \setlength{\tabcolsep}{0.5pc}
600: {\normalsize
601: \begin{tabular*}{\textwidth}{@{\extracolsep{\fill}}lcclcclc}
602: \hline\noalign{\smallskip}
603: \mc{8}{c}{Simultaneous fit to $\tau$ and \ee  \sfs\
604: ($\chi^2/{\rm DF}=383/326$)} \\
605: \noalign{\smallskip}\hline\noalign{\smallskip}
606: $m_{\rho^-(770)}$      	& 775.5 $\pm$ 0.6  	&\vline&
607: $\Gamma_{\rho^-(770)}$ 	& 148.2 $\pm$ 0.8  	&\vline&
608: $\delta$ 		& $ (2.03 \pm 0.10)\times10^{-3}$  \\
609: $m_{\rho^0(770)}$      	& 773.1 $\pm$ 0.5  	&\vline&
610: $\Gamma_{\rho^0(770)}$ 	& 148.0 $\pm$ 0.9  	&\vline&
611: $\phi_\delta$         	& $ (13.0 \pm 2.3)$  	\\
612: $m_{\rho(1450)}$   	& 1409 $\pm$ 12     	&\vline&
613: $\Gamma_{\rho(1450)}$  	& 501 $\pm$ 37       	&\vline&
614: $\beta$   		& 0.166 $\pm$ 0.005 	\\
615: $m_{\rho(1700)}$      	&  1740 $\pm$ 20     	&\vline&
616: $\Gamma_{\rho(1700)}$   &  $\equiv235$       	&\vline&
617: $\phi_\beta$   		& 177.8 $\pm$ 5.2   	\\
618: 			&			&&
619: 			&			&\vline&
620: $\gamma$   		& 0.071 $\pm$ 0.006  	\\
621: 			&			&&
622: 			&			&\vline&
623: $\phi_\gamma$   	& $\equiv0$		\\
624: \noalign{\smallskip}\hline
625: \end{tabular*}
626: }
627: \end{center}
628: \end{table}
629: %\beq
630: %\label{eq:gamma_calc}
631: %\Gamma_{\rho^0} = \Gamma_{\rho^-}\left(\frac {m_{\rho^0}}{m_{\rho^-}}\right)^{\!\!3} 
632: %              \left(\frac {\beta_0}{\beta_-}\right)^{\!\!3}\;+\;\Delta \Gamma_{\rm EM}
633: %\eeq  
634: %where $\Delta \Gamma_{\rm EM}$ is the width difference from electromagnetic decays
635: %(as discussed above). Using~(\ref{eq:splitting}) leads to 
636: a total width difference 
637: of $(2.1 \pm 0.5)\mev$, which is only marginally consistent with the observedvalue.
638: 
639: %Since the $\tau$ results from ALEPH, CLEO and OPAL have been shown to be 
640: %consistent, the question of the correction of the observed $\rho$ mass 
641: %splitting when relating $\tau$ to \epem data is more relevant. However, a 
642: %combined fit using $\tau$ {\em and} \ee data, and requiring for consistency 
643: %the constraint from Eq.~(\ref{eq:gamma_calc}), has a $\chi^2$ probability of only 
644: %$0.6\%$~\cite{aleph_taubr}. In fact, correcting for different masses extracted 
645: %from the fit, and using the corresponding constrained widths, improves the 
646: %agreement between the $\tau$ and \ee line shapes, but at the expense of a 
647: %significant discrepancy in normalization\footnote
648: %{ 
649: %	It should be noted that a correction for this apparent 
650: %	$\rho$ mass splitting increases the present discrepancy for the muon 
651: %	anomalous magnetic moment between the estimates based on $\tau$ and 
652: %	\ee  \sfs~\cite{davier_pisa_g-2}. See the discussion in 
653: %	Section~\ref{sec:vacpol}.
654: %}.
655: As seen in the previous applications, the situation should improve with
656: the inclusion of the new SND data~\cite{snd:2005}. Also the observed $\rho$ 
657: mass difference~(\ref{eq:splitting}) is expected to decrease in this case.
658: