1: \documentclass[aps,showpacs]{revtex4}
2: \usepackage{fullpage,amsfonts,amssymb,pst-node,epsfig}
3:
4:
5: \setlength{\parindent}{0mm}
6: \setlength{\parskip}{0mm}
7:
8: \begin{document}
9: %\draft
10: \title{Fano theory for hadronic resonances: \\
11: the rho meson and the pionic continuum}
12: \author{N.E. Ligterink\thanks{E-mail: nol1@pitt.edu}}
13: \affiliation{Dept. of Physics and Astronomy, University of
14: Pittsburgh, 3941 O'Hara Street, Pittsburgh, PA 15260, U.S.A.}
15: \date{\today}
16: %\titleabs{
17: \begin{abstract}
18: We develop a model-independent analysis of hadronic scattering data
19: in the resonance region, where the
20: resonance shape follows from the matrix elements of a Hamiltonian.
21: We investigate the rho meson in the tau decay. We demonstrate that
22: the rho meson resonance in the two-pion decay of the tau lepton
23: is described well through the coupling of a bare rho meson
24: to the two-pion and the four-pion continuum.
25: Furthermore, this four-pion
26: continuum corresponds with the data of the four-pion decay channel
27: of the $\tau^-$ at energies up to $1.1$ GeV.
28: \end{abstract}
29: \pacs{25.40.Ny
30: %Resonance reactions,
31: 13.75.Lb
32: %Meson-meson interactions,
33: 14.60.Fg
34: % Taus,
35: 11.10.Ef
36: %Lagrangian and Hamiltonian approach
37: }
38:
39: \maketitle
40:
41: \section{Introduction}
42:
43: The most obvious unstable hadronic states, or resonances, are single
44: absorption peaks in the $\pi N$ scattering data in a particular channel, with a
45: Breit-Wigner, or Lorentz, shape, where the width is related to the
46: coupling to mainly pionic decay channels. However,
47: resonances generally have distorted shapes due to the
48: presence of other resonances or thresholds of competing channels.
49: Their shape might look nothing like the Breit-Wigner shape that
50: is often used to fit the data, and the position of the peak might
51: shift through inter-resonance and resonance-continuum interactions.
52: The extraction of single resonances
53: from complex scattering data is still strongly model dependent~\cite{VDL}.
54:
55: Today, baryons are no longer tied to $\pi N$ scattering,
56: but are also observed in other experiments, such as $\gamma N$
57: scattering. They are particles in their own right, which can be
58: probed with different experiments. However, hadrons are not elementary
59: particles.
60: QCD was discovered as the underlying microscopic field theory
61: of composite hadrons. In principle hadrons are bound states of
62: quarks, but deriving their properties directly from QCD has
63: proven difficult.
64: Therefore, the interest in hadronic resonances
65: has been extended from basic resonance parameters, such as
66: the mass and the
67: width, to microscopic properties that can test the quality of
68: a wide range of microscopic
69: models of hadrons. Although most models reproduce the mass spectrum
70: with a reasonable accuracy, microscopic properties, such as charge radii and
71: magnetic moments, and branching ratios of decays are currently under
72: investigation.
73:
74: Furthermore, quark models allow for states, hybrids or exotics,
75: that do not consist of three quarks, or simple quark-anti-quark pairs.
76: However, to extract a clear signal for these states
77: from experiment is a formidable task. Therefore, it is important to
78: understand the conventional hadronic states and their coupling to
79: possible exotic states better.
80:
81: Since the first analysis of hadronic resonances there have been many
82: theoretical developments. Quantum field theory has replaced quantum
83: mechanics as the fundamental theory for small-scale physics, and renormalization
84: is an unfortunate, but integral part of a modern approach based on local
85: interactions. In hadron physics the issue of handling divergences and
86: renormalization is often by-passed by introducing form factors or
87: vertex functions.
88: Vertex form factors which smear the interactions and make
89: further regularization superfluous fail to have universal applicability;
90: each experiment within a particular energy range
91: is explained with a model which has its own form factors.
92: Global models that fit the world data do a good job in the energy
93: range of the world data, but these models would not necessarily
94: lead to good predictions for new results at higher energies.
95:
96: The local interactions in field theory seem essential to maintain
97: relativistic covariance, and, although models with vertex form factors
98: can reproduce the data well with a limited number of parameters, they
99: are tied to a particular experiment, a
100: particular energy range, and a particular set of channels, and they say
101: very little general about a particular hadronic state. Moreover, a different
102: set of vertex form factors might do just as well.
103:
104: Furthermore, there are bounds and constraints, such as
105: unitarity and analyticity, on the models that describe
106: scattering. They were put on a firm footing by study of the analytical
107: properties of the S-matrix, and are incorporated in quantum field
108: theory. However, since hadrons are not elementary particles, there
109: is no fundamental field theory, with a small number of parameters,
110: that describes all of hadron dynamics. Model field theories of hadron physics
111: serve to bring order in the complexity, to restrict the model space,
112: and to implement constraints from symmetries.
113:
114: Separate from the choice of the model hadronic field theory,
115: the theory of resonances has unfortunately not kept pace with the
116: developments in field theory in general. Consistency requires that
117: the imaginary part of the mass
118: is of an unstable particle follows from a microscopic model that links
119: the particle with its decay channels. Some formal developments have
120: been made~\cite{MS59,Zwa63}; however, in practice the models underlying
121: the resonance
122: shapes that describe actual data, which can be very complicated, are
123: in essence the same as those of thirty years ago~\cite{BW52,tola}.
124: These models aim to locate the
125: resonance poles, but gives very little other information that can
126: constrain models of hadron interactions. Microscopic properties
127: of hadrons remain hidden in the data.
128:
129: Some theoretical progress has been made.
130: For example, resonances have been studied and renormalization is
131: handled in chiral perturbation theory~\cite{KKW}.
132: However, the typical energy of hadronic
133: resonances is too high and lies beyond the region of applicability of
134: chiral perturbation theory. At these energies
135: the chiral symmetry is not important, and other features, such as
136: form factors or a large number of low-energy parameters dominate
137: the fit to the data. The hadronic Lagrangian is not
138: restricted by chiral symmetry at these energies, and many constants
139: need to be fitted to the data.
140:
141:
142: In most cases, attempts to describe medium-energy hadronic physics
143: from an underlying field theory start with a chiral Lagrangian
144: which is augmented with additional constraints. For example,
145: one important, but approximate conservation law in medium-energy
146: hadronic physics, which has been utilized, follows from vector
147: meson dominance. This
148: suggests there is a universal hadronic current by which the
149: photon and the rho meson couple to all hadrons~\cite{HLS}.
150: Furthermore, the large $N_c$ limit has some implications for
151: amplitudes, which can restrict the $SU(3)_f$ chiral Lagrangian
152: so that it has predictive power~\cite{Lut00}.
153: In more traditional chiral Lagrangian approaches unitarity has
154: been used to improve the predictions beyond the original
155: range of applicability~\cite{oller}.
156: Finally, from a more traditional hadronic few-body dynamics point
157: of view, a consistent implementation of covariance can extend
158: the range of validity and restrict the model dependence~\cite{Pic01}.
159: For most of these approaches, scales and form factors still need to be
160: introduced. Furthermore, the models are more designed to test the assumptions
161: made, than to extract universal quantities from a wide range of
162: scattering data. In all these approaches hadrons enter as the
163: elementary fields. Independently, within
164: the Dyson-Schwinger framework, there are investigations on the way to
165: use their rho meson Bethe-Salpeter results in baryon and meson
166: form factors~\cite{tandy}, where the composite nature of hadrons in
167: terms of quarks is resolved.
168:
169: Our aims are much simpler; we would like to determine universal quantities,
170: like the matrix elements of a Hamiltonian, directly and consistently
171: from the data.
172: Consistent tools for the analysis of resonances that satisfy
173: unitarity, analyticity are developed. They meet a microscopic model for
174: hadrons halfway in terms of matrix elements and bare masses.
175: The divergences that arise from the local
176: interactions are renormalized, and the only parameters present are
177: a few coupling constants and bare masses.
178: Simple field-theoretical results, such as the bubble summation for a
179: single channel are reproduced. The method is
180: much more versatile, and this versatility is paramount.
181: The coupled channel data should yield an Hamiltonian with the same
182: asymptotic states first. If there is an important inelastic
183: cross section, the only models that can be trusted are those which
184: explicitly take into account these channels. Possible candidates
185: for resonances can be added and their effects on the data studied. It
186: is not a matter of explaining a particular set of data, but
187: it is important to have independent verifications of resonances
188: in independent experiments.
189:
190: Any other input is considered part of a microscopic model, and should be
191: kept in its simplest form. The approach does not hinge
192: on a particular approximation, but solely on one condition.
193: Any approximation should be made at the level of the Hamiltonian
194: and it should maintain Hermiticity.
195: The Hamiltonian is a good starting point and end point of such tools:
196: formulate a model in terms of a Hamiltonian, and fix the matrix elements
197: with the data. Unitarity, analyticity, and other constraints on the data
198: are automatically fulfilled within this approach.
199:
200: This method for analysis of hadronic
201: resonances is designed for the case of a strong coupling between continua and
202: multiple resonance states. This paper is the first step in developing a set of
203: tools for a fully unitary approach to coupled-channel scattering
204: problems, which is under investigation at the moment. Furthermore,
205: it establishes contact with the local field theories that should describe these
206: systems, and renormalization is handled.
207:
208: In most studies resonances are fitted with Breit-Wigner resonance shapes,
209: however, resonances have often more complicated shapes.
210: In atomic theory there is
211: a wide range of methods developed that describe resonances or
212: auto-ionization profiles. In hadronic physics the interaction is
213: stronger. The resonances have generally a larger width, and
214: therefore more complicated shapes as they interact with other
215: resonances and the asymtotic states.
216: The method used here is an adaption of a method developed by
217: Fano in the 1960's.
218:
219: Fano developed~\cite{Fan61,Cow82,BR97} an analysis
220: for atomic resonances by writing
221: down a general Hamiltonian for such systems. This Hamiltonian
222: can be diagonalised exactly, thereby yielding direct relations
223: between scattering data and the parameters of the Hamiltonian.
224: It is based on the separation of the problem into two parts:
225: the continuum, characterized by the energy of the asymptotic states
226: in a particular channel, and resonances, which in the absence of
227: a continuum are discrete eigenstates, such as, the states
228: that would follow from a constituent quark model, large $N_c$
229: approximation, or a quenched lattice calculation.
230:
231: The continuum should be considered an
232: eigenstate of the decoupled system as well. The wave function could
233: be deformed at the
234: origin with respect to the non-interacting wave function due to
235: direct continuum-continuum interaction as would occur in field theory.
236: In this case the continuum can be composite; with asymptotic states
237: without fixed particle numbers, as we will see in section~\ref{rho}
238: where we discuss the rho meson as an example.
239: By splitting the problem in two and expressing it in the eigenstates of the
240: sub-problems many matrix elements are zero due to orthogonality of the
241: eigenstates of the sub-problems.
242:
243: As a first example we analyze the rho meson resonance in the tau
244: decay. The resonance
245: shape dominates the hadronic decay of the tau lepton into two
246: pions. We choose to describe this problem with a bare rho which
247: couples to a two-pion state.
248: The energy dependence of
249: the coupling can be modeled by a chiral Lagrangian.
250: To extend this simple model, there are two main avenues: introduce
251: a more complicated energy dependence, motivated by some
252: microscopic model, or introduce the next state that is
253: important, which is the four-pion state (with a threshold at
254: $558$ MeV). For the data it is clear that the four-pion decay
255: of the tau lepton dominates the two-pion decay above 1 GeV.
256: The effects of four-pion states are
257: hard to handle in chiral perturbation theory, since this
258: involves at least a three-loop diagram and many intermediate states,
259: such as $\omega\pi$, $\rho\rho$, $\pi\pi\rho$, and $a_0\pi$
260: mesons~\cite{BR69}.
261: However, we will see that a simple four-pion state does precisely
262: what is needed: it corrects the high-energy part of the two-pion
263: decay, and gives a quantitative prediction of the four-pion
264: decay. In the Hamiltonian approach the details of
265: of the rho four-pion coupling cannot be resolved with
266: present-day data. The two effective coupling constants;
267: the two-pion and the
268: four-pion decay of the rho meson are sufficient to describe the
269: hadronic tau lepton
270: below $1.1$ GeV. Above this
271: energy there are many channels which can only be distinguished by
272: looking at invariant masses of subsystems in multi-pion states to which
273: most of these hadronic states decay. At the
274: moment only the three-pion decay of the intermediate omega meson has a
275: clear signal~\cite{Eck}, mainly because of the narrow width of the omega.
276:
277: Many theoretical studies of the rho meson focus on combining vector
278: meson dominance (VMD)~\cite{Sak60} with chiral symmetry. In these models
279: the photon
280: couples to most hadrons through a universal conserved vector current
281: (CVC).
282: The microscopic foundation of this current, and
283: its underlying symmetry, is poorly
284: understood, and a hidden local symmetry (HLS) is
285: postulated~\cite{HLS}. This symmetry mixes the photon with the rho
286: meson. The additional knowledge resulting from this assumption has led
287: to a reasonable understanding of the shape of the rho meson due to
288: coupling with the two-pion continuum.
289:
290: However, since chiral perturbation theory focusses on low-energy behavior,
291: the predicted higher energy behavior does not agree with the data,
292: because the higher order derivative
293: couplings violate unitarity at intermediate energies.
294: Therefore, the high-energy part of the rho resonance fails the
295: data completely. Improved HLS models
296: unitarize previous results, and somewhat improve agreement. We, on the
297: other hand,
298: find the correct high-energy behavior by simply including the four-pion
299: continuum.
300:
301: In the next section, we will develop the general theory of resonances
302: given a Hamiltonian with a number of discrete states and a continuum.
303: In Section \ref{cf} we show that the matrix elements that appear in the
304: Hamiltonian can be related to Feynman diagrams. For a first application
305: to hadronic resonances we apply the method to the rho meson in Section
306: \ref{rho}. In Section \ref{results} we present the results, and the
307: final section we discuss future applications and improvements of this
308: method.
309:
310:
311: \section{Theory}
312:
313: Let us first consider a general Hamiltonian which couples a continuum to a
314: set of discrete states that are orthogonal with respect to each other.
315: Many systems can be cast in this form (it does not allow for
316: continuum-continuum interactions, which is a topic of current
317: investigations). The Hamiltonian has the general form:
318: \begin{eqnarray}
319: H & = & \sum_{i=1}^k |i\rangle \epsilon_i \langle i | + \int
320: d \Delta \ |\Delta \rangle \Delta \langle \Delta | \nonumber \\
321: & & + \sum_{i=1}^k \int W_i(\Delta) d \Delta \
322: \left( |\Delta \rangle e^{-i\phi_i(\Delta)} \langle i |
323: \ + \ |i \rangle e^{i\phi_i(\Delta)} \langle \Delta | \right) \ \ ,
324: \label{ham}
325: \end{eqnarray}
326: where $\epsilon_i = \sqrt{{\bf p}^2+M_i^2}$ with ${\bf p}$ the momentum of the system,
327: and $M_i$ is the mass of the discrete, bare state $i$. The continuum state
328: $|\Delta \rangle$ is labeled with its energy $\Delta$. The coupling is
329: parametrized with the unknown coupling functions $W_i$
330: and phases
331: $\phi_i$. In practice, such Hamiltonian are restricted to a particular
332: channel, with a given partial wave, charge, and other conserved quantum
333: numbers, which we therefore omit from the start.
334:
335: All the integrals are principal value integrals, additional
336: terms that arise from the singularities will be taken in account
337: explicitly, rather than implicitly in a particular pole or
338: $i \epsilon$ prescription, which places the pole at a particular
339: side of the contour integration. The latter is used in scattering theory,
340: where the pole part is associated with open channels, or asymptotic,
341: states. Within an eigenfunction interpretation, we do not necessarily
342: interpret singular parts of the eigenstate with out-states, they
343: could be in-states as well, or, if the interaction is strong, mix with
344: the resonance state. This information is carried in the occupation
345: numbers $\alpha_i$ and $\beta$, which are normalized to unity, even for
346: a singular functional dependence of the spectroscopic strength
347: $\beta$ on the energy.
348:
349:
350: The eigenstates $|\omega\rangle$ with the continuous energy $\omega$ of the
351: Hamiltonian can be expressed as:
352: \begin{equation}
353: |\omega\rangle = \int d \Delta \beta(\omega,\Delta) |\Delta \rangle +
354: \sum_{i=1}^k \alpha_i(\omega) |i\rangle \ \ .
355: \end{equation}
356: where $\alpha_i$'s are the occupation numbers of the discrete state,
357: and $\beta$ is the spectroscopic amplitude, or occupation density, of
358: the continuum states.
359: Substituting this into the Hamiltonian, Eq.~((\ref{ham})), we can express
360: $\beta(\omega,\Delta)$ in terms of the $\alpha$'s:
361: \begin{equation}
362: \beta(\omega,\Delta) = \left( \frac{1}{\omega-\Delta} + z(\omega) \delta(\omega-\Delta)
363: \right) \sum_{i=1}^k \alpha_i(\omega) W_i(\Delta) e^{-i\phi_i(\Delta)} \ \ .
364: \label{beta}
365: \end{equation}
366: where $z(\omega)$ can be any regular function to be determined later.
367: $z(\omega)(\omega-\Delta)\delta(\omega-\Delta)$ vanishes.
368: Substituting $\beta$ back into the second equation of the Hamiltonian,
369: Eq.~(\ref{ham}), yields:
370: \begin{eqnarray}
371: (\omega -\epsilon){\bf \alpha}(\omega) & = & \pi {\cal F}(\omega) \cdot
372: {\bf \alpha}(\omega)+
373: z(\omega) {\bf F}(\omega) \cdot {\bf \alpha}(\omega) \ \ ,
374: \label{mat}
375: \end{eqnarray}
376: where $(\omega -\epsilon)$ is a diagonal matrix with entries $\omega - \epsilon_i$.
377: The hermitian matrix ${\bf F}$, and its Hilbert transform ${\cal F}$
378: necessary for the consistency condition on $z(\omega)$ are defined:
379: \begin{eqnarray}
380: F_{ji}(\xi) & = & W_i(\xi) W_j(\xi) e^{i(\phi_j(\xi) -\phi_i(\xi))}\ \ , \\
381: {\cal F}_{ji}(\eta) & = & \frac{1}{\pi} \int d \xi \frac{F_{ij}(\xi)}{\eta-\xi} \ \ .
382: \label{real_F}
383: \end{eqnarray}
384: Eq.~(\ref{mat}) is multiplied from the left by:
385: \begin{equation}
386: {\bf W}^\dagger(\omega) \cdot\left((\omega -\epsilon)- \pi {\cal F}(\omega) \right)^{-1} \ \ ,
387: \end{equation}
388: where the vector
389: ${\bf W}^\dagger(\omega) = (W_1(\omega) e^{-i\phi_1(\omega)}, W_2(\omega) e^{-i\phi_2(\omega)},
390: \cdots, W_k(\omega) e^{-i\phi_k(\omega)} )$.
391: The whole expression contains a common factor
392: ${\bf W}^\dagger(\omega) \cdot \alpha(\omega)$, which can be
393: divided out, which yields a closed form expression for $z(\omega)$:
394: \begin{equation}
395: z(\omega) = \left( {\bf W}^\dagger(\omega) \cdot\left((\omega -\epsilon)-
396: \pi {\cal F}(\omega) \right)^{-1} \cdot {\bf W}(\omega) \right) ^{-1} \ \ .
397: \label{z}
398: \end{equation}
399: At this point it is important to realize the significance of $z(\omega)$
400: above:
401: it is the projected resolvent $((\omega -\epsilon)-\pi {\cal
402: F}(\omega))^{-1}$ of the discrete spectrum, perturbed through the coupling with the
403: continuum. Given that ${\cal F}$ is hermitian, the matrix $(\epsilon - \pi {\cal F})$ will
404: have a discrete spectrum $\omega_1,\omega_2,\cdots, \omega_k$,
405: \begin{equation}
406: \Omega =
407: \left(\begin{array}{cccc} \omega_1 & & & \cr
408: & \omega_2 & & \cr
409: & & \ddots & \cr
410: & & & \omega_k \end{array} \right) =
411: U^\dagger(\omega) \cdot \left( \epsilon - \pi {\cal F}(\omega) \right) \cdot U(\omega) \ \ .
412: \end{equation}
413: In order to simplify the results, it is convenient to work in the basis of the true spectrum
414: of the discrete system. In this case the inverse, that appears in Eq.~(\ref{z}), is trivial. We
415: perform the substitutions:
416: \begin{eqnarray}
417: \alpha(\omega) & = & U \cdot {\bf a}(\omega) \ \ , \\
418: {\bf W}(\omega) & = & U \cdot {\bf V}(\omega) \ \ ,
419: \end{eqnarray}
420: and likewise for the hermitian conjugates.
421:
422: The normalization of the state $|\omega \rangle $ yields:
423: \begin{eqnarray}
424: \langle \omega' | \omega \rangle & = &
425: \alpha^\dagger(\omega') \cdot \alpha(\omega) \\
426: & & + \alpha^\dagger(\omega')\cdot \frac{z(\omega') {\bf W}(\omega') {\bf W}^\dagger(\omega') -
427: \pi {\cal F}(\omega')}{\omega'-\omega} \cdot \alpha(\omega) \nonumber \\
428: & & - \alpha^\dagger(\omega')\cdot \frac{z(\omega) {\bf W}(\omega) {\bf W}^\dagger(\omega) -
429: \pi {\cal F}(\omega)}{\omega'-\omega} \cdot \alpha(\omega)\nonumber \\
430: & & + (\pi^2 + z^2(\omega))\delta(\omega-\omega')
431: \alpha^\dagger(\omega')\cdot {\bf W}(\omega') {\bf W}^\dagger(\omega)
432: \cdot \alpha(\omega) \ \ , \nonumber
433: \end{eqnarray}
434: where the $\pi^2$ contribution results from the pole term of the product of the two principal
435: value singularities in the spectroscopic densities $\beta$.
436: If we substitute the new variables, {\bf a} and {\bf V},
437: it is easy to see that the first three terms cancel each other.
438: The state $|\omega \rangle $ is an eigenstate by construction and:
439: \begin{equation}
440: \langle \omega' | \omega \rangle = (\pi^2 + z^2(\omega))\delta(\omega-\omega')\alpha^\dagger(\omega')\cdot {\bf W}(\omega') {\bf W}^\dagger(\omega)
441: \cdot \alpha(\omega) \ \ .
442: \end{equation}
443: The probability of finding the system in the discrete state $|i\rangle$, given it was in
444: the discrete state $|j\rangle = |\psi(t=0)\rangle$ is:
445: \begin{equation}
446: \langle i |\psi(t)\rangle = \int d \omega \ \alpha_i (\omega) \alpha_j^\ast(\omega) e^{-i \omega t} \ \ ,
447: \label{decay}
448: \end{equation}
449: where the occupation numbers are given by:
450: \begin{equation}
451: \alpha_i(\omega) =
452: \frac{W_i(\omega) e^{- i \phi_i(\omega)}}
453: {{\bf W}^\dagger {\bf W}\left[z(\omega) + i \pi\right]} \ \ ,
454: \end{equation}
455: and we have chosen a convenient phase. At the position of a
456: resonance $\omega = \omega_j$, $z(\omega_j)$ vanishes, however, this
457: is not necessarily the resonance maximum, as the
458: coupling function depends on the energy. If the coupling
459: function is a constant, a single resonance shape reduces to a Breit-Wigner
460: form:
461: \begin{equation}
462: \alpha_j(\omega) = \frac{W_j}{ (\omega - \omega_j) + i \pi W_j^2 } \ \ .
463: \end{equation}
464:
465: Likewise, the phase shift follows from the inspection of Eq.~(\ref{beta})
466: for the spectral density of the continuum state. The principal value
467: singularity yields the scattering wave function proportional to
468: $-\pi \sin (k(\omega) + \delta_l)$, while the delta function yields
469: the scattering wave function proportional to $\sin (k(\omega) + \delta_l
470: +\frac{\pi}{2})$.
471: Therefore, their relative strength $\pi/z(\omega)$ determines the
472: effect of the coupling to the discrete states to the phase shift
473: $\delta_r$, which yields
474: \begin{equation}
475: \delta_r = - \arctan \frac{\pi}{z(\omega)} \ \ .
476: \label{phase}
477: \end{equation}
478:
479: Clearly, once the eigenstate is known, in terms of the occupation numbers $\alpha_i$ of the
480: discrete states and the occupation number $\beta$ of the continuum, at every energy we can
481: determine any quantity we like. The decay of a state is given by the probability of
482: finding the system in that state, Eq.~(\ref{decay}). In a scattering process,
483: the incoming state
484: must be projected on the continuum part of the eigenstate, and the scattering continuum
485: state can be reached subsequently via one of the discrete states.
486: Transition amplitudes are therefore proportional to $W_j \alpha_j$. For
487: a single channel the $T$-matrix is:
488: \begin{equation}
489: T = \frac{1}{z(\omega) + i \pi} \ \ ,
490: \end{equation}
491: which preserves unitarity.
492:
493:
494:
495: \section{Coupling functions}
496: \label{cf}
497:
498: \begin{figure}
499: \centerline{\includegraphics[width=4cm]{fig1.eps}}
500: \caption{The two diagrams from which the coupling functions $W$ and $W_2$, between the
501: discrete state and the continuum, is derived.}
502: \label{fig2}
503: \end{figure}
504:
505: The masses $M_i$ and the coupling functions $W_i$ are to be determined by other means.
506: We assume that the masses follow from some QCD or constituent-quark
507: based model. Although, the coupling functions; the pion-hadron interaction, could
508: in principle be determined by the same means, in practice that
509: turned out to be a formidable task. Therefore, we turn to Lagrangians with
510: explicit pions and chiral symmetry for
511: the coupling functions~\cite{BR69}. The easiest connection is made
512: midway. Although the meaning of the Green's function, or resolvent, is
513: different in the study of eigenstates, which is stationary, and the
514: study of scattering, which is dynamical, the actual mathematical form
515: is the same. At many levels one can find relations between expressions
516: in scattering theory and stationary eigenvalue problems. We choose the
517: equivalence that is most convenient to us, which is the equivalence
518: of the one-particle self-energy. With
519: chiral Lagrangians
520: we can determine the one-particle irreducible
521: self-energy correction $\Sigma$ of hadrons
522: due to coupling to the pion field,
523: which is given in terms of Feynman diagrams. We can equate this with the
524: real part of energy shift due to the coupling of the $i$-th discrete
525: state with the continuum $\pi {\cal F}_{ii}$
526: in the Fano theory, which also yields the real part of the self-energy
527: due to the coupling with the continuum:
528: \begin{equation}
529: \int_{PV} d \Delta \frac{W_i^2(\Delta)}{\omega - \Delta} \equiv
530: \Sigma_i(\omega) = \frac{1}{2 \omega} {\rm Re} \int d^{4n} k\ I(k) \ \ ,
531: \label{eqW}
532: \end{equation}
533: where $I$ is the integrand of the truncated Feynman self-energy diagram $\Gamma$
534: for the bare hadron $i$.
535: Hence $W_i^2$ is the phase-space of the continuum times the appropriate factors:
536: \begin{equation}
537: W_i^2(\omega) = \frac{\pi}{2 \omega} {\rm Im} \int d^{4n} k\ I(k) \ \ ,
538: \end{equation}
539: where $\omega$ is the sum of the free energies of the particles in the
540: continuum state, which, in a Feynman diagram with energy conservation at
541: the vertices, equals the total incoming energy.
542: Similarly, in the right-hand side of Eq.~(\ref{eqW}) we can
543: recognize the corresponding Goldstone diagram~\cite{BR86}.
544:
545: This connection between the coupling function $W$ and the imaginary part
546: of the Feynman diagram
547: guarantees that the tree-level and bubble-sum results in this approach
548: and with Feynman diagrams are the same.
549: However, the Hamiltonian description
550: does not automatically include the backward, or anti-particle,
551: contributions, since these are related to separate states. These
552: states must be added explicitly. The major advantage is that the
553: resulting linear eigenvalue equation
554: in the energy can be solved exactly, even in the case of
555: multiple resonances and multiple decay channels. In the language of
556: Feynman diagrams this statement corresponds to the simultaneous bubble sum of
557: the different intermediate states.
558:
559: \begin{figure}
560: \centerline{\includegraphics[width=6cm]{fig2.eps}}
561: \caption{The two backward, or Z diagrams. The inclusion of these two
562: states in the coupling functions restores covariance, but does not
563: affect the shape of the resonance, only coupling constants change slightly.}
564: \label{fig3}
565: \end{figure}
566:
567: For the rho meson, analyzed below, we included the anti-particle states,
568: that is, we combined $\rho\to\pi\pi\to\rho$ with the
569: $\rho\to\rho\rho\bar\pi\bar\pi\to\rho$ process and
570: $\rho\to\pi\pi\pi\pi\to\rho$ with
571: $\rho\to\rho\rho\bar\pi\bar\pi\bar\pi\bar\pi\to\rho$. (See Fig.
572: \ref{fig3}.) Although
573: this restores covariance it has little effect on the actual resonance
574: shape, since the threshold of the $\rho\rho\pi\pi$ state
575: is already above $1800$ MeV. The calculation of the self-energy
576: simplifies for the rho meson; the functions depend on $s = \omega^2,
577: \Delta^2$ only, since the forward and the backward contributions combine.
578: Schematically:
579: \begin{equation}
580: \int d \Delta \frac{f(\Delta^2)}{\omega - \Delta} + \int d \Delta
581: \frac{f(\Delta^2)}{\omega - (2 \omega + \Delta)} =
582: \int d \Delta^2 \frac{f(\Delta^2)}{\omega^2 - \Delta^2} \ \ .
583: \label{ap}
584: \end{equation}
585:
586:
587: The self-energy Eq.~(\ref{eqW}) is highly divergent for many pionic problems.
588: We will use on-shell
589: subtraction as our renormalization procedure, which requires implicit
590: counterterms such
591: that the masses $M_i$ and the wave function normalization
592: $\langle i | j \rangle = \delta_{ij}$
593: remain unaltered. In practice the wave-function renormalization means
594: a unit residue of the resonance pole. Below we discuss the rho meson,
595: and show explicitly how these problems are handled.
596:
597: \section{The rho meson}
598: \label{rho}
599:
600: The rho meson plays an important role in photon-hadron interactions at
601: intermediate energies.
602: It seems that the photon couples to hadrons mainly via the isovector-vector
603: rho meson, which in its
604: turn, couples to a universal hadronic current via a coupling constant $g$.
605: We see the rho meson mainly through its decay to two pions. So in order
606: to describe the physical
607: rho meson, we assume a discrete state $|\rho \rangle$ at rest, which couples
608: to the two-pion continuum
609: $|\pi\pi(\Delta) \rangle $ with energy $\Delta$.
610: We have the following matrix elements:
611: \begin{eqnarray}
612: \langle \rho |H| \rho \rangle & = & M \ \ ,\\
613: | \langle \rho | H | \pi\pi (\Delta) \rangle |^2 & = &
614: \frac{g^2 (\Delta^2 - 4 m_\pi^2)^\frac{3}{2} \theta(\Delta-2 m_\pi)}{
615: 48 \pi \Delta} = W^2(\Delta) \ \ ,\\
616: \langle\pi\pi (\Delta') | H | \pi\pi (\Delta) \rangle & = & \Delta \delta(\Delta - \Delta') \ \ ,
617: \end{eqnarray}
618: where the coupling function follows from the interaction term
619: $ g \vec \rho^\mu \cdot [\vec \pi \times \partial_\mu \vec \pi] $ in the chiral
620: Lagrangian, and the vectors are in the isospin space.
621: In practice this
622: particular coupling is only important for the right threshold behavior,
623: for large energies the results are dominated by the
624: phase space.
625: For the moment we will assume that there is no self-interaction in
626: the two-pion continuum, therefore the coupling function $W$ follows the
627: phase space. The explicit unitarity of the approach and the on-shell
628: renormalization lead to the correct large energy behavior.
629:
630: In the future we will investigate the effect of the
631: continuum-continuum interaction on the results.
632: We also include the next
633: contribution to the physical rho, which, considering
634: the possible states in this channel and their energies, must be the
635: four-pion state, with a threshold at $558$ MeV.
636: From an energy of about $1$ GeV there are many other intermediate
637: decay channels
638: important, e.g.: $\pi^6$, $\omega \pi$, $\rho\pi^2$
639: and $a_0\pi$ between $900$ MeV and $1.5$ GeV~\cite{BR69}.
640: They are to be taken in account if one wants to
641: analyze rho decay at energies above $1$ GeV.
642: For example, the $\pi\omega$ system has a threshold
643: ($921$ MeV) close enough to the $\rho$ mass ($770$ MeV) to yield
644: contributions that do not simply follow the four-pion phase space.
645: However, we leave this to future investigations. Although channels open
646: at a slightly lower energy than $1.1$ GeV, their effect is not strong at threshold
647: due to the low-energy behavior manifest in the derivative couplings of
648: the chiral Lagrangian.
649:
650: From the chiral Lagrangian there are many intermediate states that lead to the four-pion
651: state from the discrete rho, however, all these states are highly virtual,
652: so we can approximate
653: the coupling between the discrete rho and the four-pion
654: state with a single derivative coupling. (See Fig.~\ref{fig2}.) This
655: yields a complicated integral, equivalent to a three-loop calculation in
656: Feynman perturbation theory.
657: Ignoring these correlations in the four-pion state, the additional matrix elements are
658: well-approximated by:
659: \begin{eqnarray}
660: | \langle \rho | H |\pi\pi \pi\pi (\Delta)\rangle |^2 & = &
661: \frac{g_2^2 (\Delta^2 - 16 m_\pi^2)^\frac{9}{2}\theta(\Delta-4 m_\pi)}{
662: M^5 \Delta^3} \equiv W_2^2(\Delta) \ \ , \\
663: \langle\pi\pi\pi\pi (\Delta') |H | \pi\pi\pi\pi (\Delta) \rangle & = & \Delta \delta(\Delta - \Delta') \ \ .
664: \end{eqnarray}
665: From semi-analytical three-loop calculations~\cite{Lig00} one can see that
666: the approximation
667: for $W_2$ is reasonable, and the eigenstates of the complete system turn
668: out to be stable under small variations in the coupling function $W_2$.
669: The continuum is composite.
670: It contains a two-pion and a four-pion fraction to the ratio $W/W_2$. We did not allow a
671: direct coupling between them, so they mix with ratios proportional to the coupling
672: strengths to the discrete state $|\rho \rangle$.
673:
674: The real parts, from Eq.~(\ref{real_F}), are divergent integrals, as
675: expected for local interactions. Therefore we need to regularize
676: the results and state how we determine the finite parts.
677: In hadronic physics renormalization has to be pragmatic, since
678: most theories for hadronic physics are not renormalizable; there
679: is no fixed set of parameters, masses and coupling constants that
680: can re-absorb all the divergences by a proper redefinition. Therefore,
681: at any order or scale a number of observables are needed to fix
682: the unknown finite renormalizations of the system. In principle a lot
683: of fine-tuning, to fit theory to experiment, can be done here. However,
684: we take the view that this fine-tuning means fitting the short and
685: medium range physics, which should be done with a proper microscopic
686: theory of hadrons. The bare rho is the result of miscroscopic
687: interaction that should incorporate the short range effects of
688: pions.
689: Therefore all the finite renormalizations are chosen such that the
690: contributions vanishes at the rho mass $M$. This is a
691: choice, but it separates analysis of scattering
692: data from any model assumptions. If some additional knowledge
693: about the structure of the rho meson exists, it should end up in
694: the coupling function. This is important to make the analysis based
695: on the Hamiltonian scale independent, since scale dependence implies model
696: dependence.
697:
698: For this purpose we use on-shell renormalization; subtracting
699: the low-order divergent terms in Taylor expansion in
700: $(\omega^2-M^2)$ of the self-energy ${\cal F}$. Note that
701: since we have included the anti-particle diagrams, Eq.~(\ref{ap}), our
702: self-energy ${\cal F}$ is a function of $s=\omega^2$ rather than
703: $\omega$, hence the expansion is in $(\omega^2-M^2)$ rather than
704: $(\omega - M)$. Odd terms are absent and do not require
705: renormalization.
706:
707: The two-pion coupling function leads to the ordinary mass and
708: wave-function renormalization $c_0$ and $c_1 (\omega^2-M^2)$.
709: The four-pion coupling function requires subtractions
710: $c^{(2)}_0$, $c^{(2)}_1 (\omega^2-M^2)$, $c^{(2)}_2 (\omega^2-M^2)^2$,
711: and $c^{(2)}_3 (\omega^2-M^2)^3$, which are set identical to
712: the same divergent terms in ${\cal F}_{\pi^4}$:
713: \begin{equation}
714: c^{(2)}_n = \frac{1}{n!} \left. \frac{\partial^n {\cal
715: F}_{\pi^4}(\omega^2)}{ \partial
716: (\omega^2)^n} \right|_{\omega = M} \ \ .
717: \end{equation}
718: The effects of the self-energy are small over the whole range of the
719: rho resonance shape, from the two-pion threshold to $1.2$ GeV,
720: which yields effects of the order of a couple of percent.
721:
722: Since the finite renormalization is model independent there are only
723: two parameters to fit the data. These are the two coupling constants $g$ and $g_2$,
724: which yield the strength of the coupling between the bare rho
725: and the two-pion continuum and the four-pion continuum respectively.
726: Since the two decay channels are both open at these energies, There are
727: also have
728: two independent measurements of these two coupling constants;
729: namely, the two-pion decay and the four-pion decay.
730:
731: \begin{figure}
732: \centerline{\includegraphics[width=9cm]{fig3.eps}}
733: \caption{The Fano calculation compared with the $\tau^-$ decay data
734: from the CLEO collaboration~\cite{CLEO}. There are three cases: the two-pion results
735: without four-pion contributions (long dashed), the results for the two-pion (dot-dashed)
736: and four-pion (solid) continua and their sum (dot-dash-dashed), and the change in the
737: two-pion decay of the latter if an artificial barrier term~\cite{CLEO} is introduced, and coupling
738: constants refitted (dot-dot-dashed).}
739: \label{fig4}
740: \end{figure}
741:
742: \section{Results}
743: \label{results}
744:
745: The hadronic tau decay at the CLEO experiment is
746: dominated by the rho meson in the intermediate state.
747: The initial tau lepton decays into a neutrino and a W-boson.
748: Subsequently the W-boson produces a hadronic state via the standard
749: model Lagrangian $Wq\bar q$ vertex. After all the electroweak structure
750: has been
751: resolved, the hadronic part of the decay is dominated by the rho meson
752: resonance at
753: energies between the hadronic decay threshold, which is the two-pion
754: threshold and about $1$ GeV when the $\omega\pi$ channel is open and
755: the first of many other decay channels start to compete.
756:
757: The experimental data from CLEO~\cite{CLEO} is represented in
758: Fig.~\ref{fig4}, where the number of events, freed from electroweak
759: dependences, is plotted as a function
760: of the invariant mass of the two-pion system. The same figure summarizes
761: our analysis. If the bare rho state couples only to a two-pion
762: continuum, the high-energy tail of the resonance
763: peak is grossly overestimated.
764: If the four-pion continuum is included, the two-pion decay channel
765: is suppressed at higher energies, in favor of four-pion decay (they
766: cross just above $1$ GeV). The total pionic decay comes again to the
767: same value as the original two-pion-only result, however, the shape of
768: the inclusive resonance peak at these higher energies is different.
769:
770: For completeness, a fit with a barrier term is also
771: included~\cite{BW52},
772: as was used in the data analysis~\cite{CLEO}.
773: The barrier term has its origin in nuclear quantum mechanics
774: and relates the amplitude of the asymptotic wave function
775: with the inner part in the nuclear potential.
776: These two parts are matched at the Coulomb barrier through which
777: an escaping radiative-decay particle tunnels. The momentum dependence
778: of this matching is reflected in the barrier term. There is no
779: field-theoretical analogue of the barrier term. It reproduces the right
780: low-energy behavior; however, within our analysis, which has only two
781: parameters, we pay for the rather artificial low-energy
782: improvement with a failure at higher energies.
783: The position and the slope of the four-pion decay events compare well with
784: both the OPAL~\cite{OPAL} and the CLEO~\cite{CLEO} plots for four-pion
785: events.
786:
787: Comparing the resonance shape with the results from our Hamiltonian and
788: fitting the two coupling constants $g$ and $g_2$ to hadronic decay of
789: the $\tau^-$ we find a number of results.
790: First, from the fit of the $\pi\pi$ decay of the $\rho$-meson, we find a
791: coupling constant
792: $g_2 \approx 0.2$ to the four-pion state, which agrees well, without any
793: fitting, with the actual four-pion
794: data~\cite{OPAL} between threshold ($0.558$ GeV) and $1.1$ GeV. Beyond
795: $1.1$ GeV we expect other contributions,
796: which decay to four pions, like $\omega \pi$, to be important.
797:
798: Second, if we consider the pion as an elementary particle, the initial state,
799: as generated by the
800: electroweak process, will be the discrete rho state. From the actual
801: two-pion data which falls of as $\omega^{-2}$ for high energies we infer
802: that there is no direct coupling to the two-pion continuum,
803: as this would lead to large amplitudes for large energies.
804: Corrections to this assumption of the pion as an asymptotic but pointlike particle,
805: which follows from combining Fano theory with the chiral Lagrangian,
806: could improve the low-energy fit but would fail to descibe the data at higher
807: energies~\cite{KKW}.
808:
809: Third, the presence of the four-pion continuum changes the two-pion coupling constant.
810: It is interesting to notice that the coupling constants have to be adjusted if additional,
811: non-interacting channels are included. The competition between decay modes
812: generates a simple, but effective final-state interaction.
813:
814: Fourth, the real part, after the on-shell
815: renormalization, yields a negligible effect.
816: Since, with on-shell renormalization, the real part vanishes when the energy equals
817: the rho mass, the value of the real part is small in the neighborhood of this value.
818:
819: Finally, we are unable to get a very good fit to the data near threshold, within the model.
820: There have been several suggestions to improve the
821: results~\cite{KKW,HLS}, based on chiral symmetry.
822: These all use a direct electroweak-pion-pion coupling, which violates
823: unitarity for large energies. Physically there are two explanations for the deviations,
824: within our scheme.
825: Firstly, the coupling between the electroweak state and the bare rho could have a momentum
826: dependence, due to, for example, a pionic, or $q\bar q$ quark, rho wave function.
827: This should be
828: visible in an additional quenching of the electroweak decay modes at the energies
829: just above the
830: two-pion threshold. Secondly, the electroweakly produced initial state could contain
831: a two-pion state, but only at low energies. It is hard to motivate this from a underlying local
832: theory. This is equivalent to unitarized hidden local symmetry (HLS)
833: models~\cite{HLS}.
834:
835: \section{Discussion}
836:
837: For the problem of the rho meson studied here we did not require the full machinery
838: of the Fano theory. For example, there is only one discrete state in the system.
839: In the case of a neutral rho meson the mixing with the $\omega$ meson could be taken into
840: account explicitly and unitarily.
841:
842: For the moment it has been more important to show that the two-pion and the
843: four-pion continua describe the rho resonance well, where the four-pion
844: continuum results in two independent observations: its effect on the two-pion
845: decay, and a direct measurement of the four-pion decay of the rho meson.
846: All this can be discussed at the level of {\it restricted Hamiltonians}.
847: These Hamiltonians contain given states and continua which are expected to
848: dominate the results (in this case a composite continuum consisting of
849: two- and four-pion states in the rho channel, and the bare rho state).
850: The fact that no further approximation is required allows one to draw definite
851: conclusions, whether these states and continua actually dominate the physics
852: in a certain channel and at a certain energy, or other degrees of
853: freedom are
854: necessary. In this case the data between $0.6$ GeV and $1.1$ GeV is well-described
855: by these degrees of freedom.
856:
857: The Fano theory allows one to go further in particle numbers and complexity than
858: other methods. Complex states can be build up hierarchically by admixing states
859: in their expected order, similar to
860: angular momentum coupling schemes in atomic physics. The only aspect not handled in
861: Fano theory is the particle number conserving interaction between continua as occurs
862: in field theory. Instead of these self-interacting states we assume the states to be
863: eigenstates already, with the only noticeable effect a deformation of the phase space
864: in $W$, with respect to the free-particles phase space, due to the changed dispersion
865: relation $\omega({\bf k})\not = \sum_i \sqrt{{\bf k}^2 + m_i^2}$, to be calculated
866: separately or approximated by the free result.
867:
868: In the future we plan to apply this method to baryonic resonances,
869: for which a lot of new data is becoming available from recent JLAB experiments.
870: In these experiments the problems with data analysis are more stringent due
871: to the rich spectra and the higher energies~\cite{VDL}.
872: We will also investigate the way in which to extend Fano theory to allow for
873: continuum-continuum interactions.
874:
875: \section*{Acknowledgments}
876: I would like to thank Wim Ubachs for pointing me in the right direction on
877: atomic physics, Wolfram Weise for suggesting this problem and the
878: encouragement, and Rob Timmermans for useful and stimulating
879: discussions. I would also like to thank Evgeni Kolomeitsev for pointing out
880: useful references and reading material.
881: Finally, I gratefully acknowledge the careful reading and extensive
882: comments by Eric Swanson and Steve Dytman that helped to shape this paper.
883:
884: \begin{thebibliography}{100}
885: \bibitem{VDL} T.~P.~Vrana, S.~A.~Dytman, and T.~S.~H. Lee, Phys.\ Rept.\ {\bf 328},
886: 181 (2000).
887: \bibitem{MS59} P.~T.~Matthews and A.~Salam, Phys.\ Rev.\ {\bf 115}, 1079
888: (1959).
889: \bibitem{Zwa63} D.~Zwanziger, Phys.\ Rev.\ {\bf 131}, 2818 (1963).
890: \bibitem{BW52}J.~M.~Blatt and V.~F.~Weisskopf, {\it Theoretical nuclear
891: physics}, (New York, Wiley, 1952); H.~Feshbach, {\it Theoretical nuclear
892: physics}, (New York, Wiley, 1991).
893: \bibitem{tola} S.~Jadach, J.~H.~K\"uhn, and Z.~Was, Comput.\ Phys.
894: Commun.\ {\bf 64}, 275 (1991); {\bf 70}, 69 (1992); {\bf 76}, 361 (1993).
895: \bibitem{KKW} F.~Klingl, N.~Kaiser, and W.~Weise, Z.~Phys.~{\bf A 356}, 193 (1996).
896: \bibitem{HLS} H.~B.~O'Connell {\it et al.}, Nucl.\ Phys.\ {\bf A 623}, 559 (1997); M.~Benayoun
897: {\it et al.}, Z.\ Phys.\ {\bf C 72}, 221 (1996).
898: \bibitem{Lut00}M.~F.~Lutz and E.~E.~Kolomeitsev,
899: %``Covariant meson baryon scattering with chiral and large N(c)
900: %constraints,''
901: Found.\ Phys.\ {\bf 31}, 1671 (2001).
902: \bibitem{oller}
903: J.~A.~Oller and E.~Oset,
904: %``N/D description of two meson amplitudes and chiral symmetry,''
905: Phys.\ Rev.\ D {\bf 60}, 074023 (1999).
906: \bibitem{Pic01}
907: M.~A.~Pichowsky, A.~Szczepaniak, and J.~T.~Londergan,
908: %``Relativistic unitary description of pi pi scattering,''
909: Phys.\ Rev.\ D {\bf 64}, 036009 (2001).
910: \bibitem{tandy} P.~Tandy (private communication); P.~Maris and P.~C.
911: Tandy, Phys.\ Rev.\ C {\bf 60}, 055214 (1999).
912: \bibitem{Fan61}U.~Fano, Phys.\ Rev.\ {\bf 124}, 1866 (1961).
913: \bibitem{Cow82}R.~D.~Cowan, {\it The Theory of Atomic Structure and Spectra},
914: (University of California Press, 1982).
915: \bibitem{BR97}S.~M.~Barnett and P.~M.~Radmore, {\it Methods in theoretical quantum optics},
916: (Oxford Press, Oxford, 1997).
917: \bibitem{BR69} L.~Banyai and V.~Rittenberg, Phys.\ Rev.\ {\bf 184}, 1903 (1969).
918: \bibitem{Eck}E.~von Toerne (private communication); K.~W.~Edward {\it et
919: al.} (CLEO), Phys.\ Rev.\ D {\bf 61}, 072003 (2000);
920: \bibitem{Sak60} J.~Sakurai, Ann.\ Phys.\ (N.Y.) {\bf 11}, 1 (1960).
921: \bibitem{BR86} J.~P.~Blaizot and G.~Ripka, {\it Quantum Theory
922: of Finite Systems}, {(MIT Press, Cambridge (MA), 1986)}.
923: \bibitem{Lig00}N.~E.~Ligterink, Phys.\ Rev.\ D {\bf 61}, 105010 (2000).
924: \bibitem{CLEO} S.~Anderson {\it et al.} (CLEO), Phys.\ Rev.\ D {\bf 61}, 112002 (2000).
925: \bibitem{OPAL} K.~Ackerstaff {\it et al.} (OPAL), Eur.\ Phys.\ J.\ {\bf C 7} 571 (1999)
926: \end{thebibliography}
927:
928: \end{document}
929: