1: \section{Current correlator in the chiral effective theory}\label{sec2}
2: At leading order, the Euclidean Lagrangian of the chiral effective theory is given by \cite{Weinberg:1978kz,Gasser:1983yg}
3: \begin{equation}
4: \mathcal L = \frac{F^2}{4}{\rm Tr}\left\{\partial_{\mu} U^\dagger \partial_{\mu}U \right\}-\frac{\Sigma}{2}{\rm Tr} \left\{e^{i\theta/N_{\rm f}}U\mathcal{M}+\mathcal{M}^\dagger U^\dagger e^{-i\theta/N_{\rm f}} \right\},
5: \end{equation}
6: where $U\in SU(N_{\rm f})$ contains the pseudoscalar degrees of freedom and $\mathcal{M}$ is the mass matrix. For simplicity we consider a mass matrix proportional to the identity, $\mathcal{M}=m\mathbb{I}$.\\
7: $F$ and $\Sigma$ are the pseudoscalar decay constant and the quark condensate in the chiral limit,
8: and $\theta$ represents the vacuum angle.
9:
10: In a finite volume $V=TL^3$ with $L\gg 1/\Lambda_{\rm QCD}$, one can distinguish different chiral regimes.
11: If $M_P$ is the pseudoscalar meson mass, approaching the chiral limit by keeping $M_P L\gg 1$ defines the so-called $p$-regime. In this case the chiral effective theory looks essentially as in the infinite volume case: finite-volume effects are exponentially suppressed by factors $\exp{(-M_P L)}$, while the mass effects are the dominant ones. The power counting in terms of the momentum $p$ and quark mass $m$ is given by
12: \begin{equation}
13: m\sim p^2,\;\;\;\;1/L,1/T\sim p.
14: \end{equation}
15: Alternatively, one can approach the chiral limit while keeping $\mu=m\Sigma V\lesssim 1$; in this case the Compton wavelength associated with the pseudo-Goldstone bosons is much larger than the linear extent $L$ of the box, and volume effects are enhanced. This defines the $\epsilon$-regime, where the power-counting is reorganised such that \cite{Gasser:1986vb,Gasser:1987ah}
16: \begin{equation}
17: m\sim \epsilon^4,\;\;\;\;1/L,1/T\sim \epsilon.
18: \end{equation}
19: One of the most important effects of the reorganisation of the power counting is that, at a given order in the effective theory, fewer Low Energy Couplings (LECs) appear with respect to the $p$-expansion. The fact that the corresponding higher-order counterterms are
20: kinematically suppressed may be convenient for the extraction of LECs by matching the effective theory to lattice QCD.
21:
22: In this work we consider the left-handed current, which at leading order in the effective theory formalism corresponds to
23: \begin{equation}
24: \mathcal{J}_\mu^a=\frac{F^2}{2}{\rm Tr}\left(T^a U \partial_{\mu}U^\dagger \right),
25: \end{equation}
26: where $T^a$ are the traceless generators of SU$(N_{\rm f})$.
27: In particular we are interested in the two-point correlation function
28: \begin{equation}\label{2p_f}
29: \mathcal{C}^{ab}(t)=\int d^3 x \langle \mathcal{J}_0^a(x) \mathcal{J}_0^b(0)\rangle ={\rm Tr}[T^aT^b]\mathcal{C}(t).
30: \end{equation}
31: The chiral effective theory formalism can be extended to the quenched case; in particular, two equivalent methods have been developed to cancel the fermion determinant, namely the supersymmetric formulation and the replica method \cite{Bernard:1992mk,Sharpe:1992ft,Damgaard:2000gh}. An important feature of the quenched setup is that the flavour singlet does not decouple in this case; moreover, its mass parameter $m_0^2/(2N_c)$ is related to the topological susceptibility.\\
32: In the following we summarise the known results for this current correlator from quenched chiral perturbation theory at NLO\footnote{We neglect terms proportional to $\frac{\alpha}{N_c}$, which is the parameter associated to the kinetic term of the singlet field, since it is suppressed for this observable in the simultaneous expansion in momenta and $1/N_c$ \protect\cite{Kaiser:2000gs}.}, in the $\epsilon$- and $p$-regimes, with degenerate quark masses.
33: Current correlators have been recently computed in the effective theory also for non-degenerate quark masses, in the full and partially quenched scenarios, in the case where all quarks are in the $p$- or in the $\epsilon$-regime and in the mixed case, where $m_v\Sigma V\lesssim 1$ for the valence quarks and $m_s\Sigma V\gg 1$ for the sea quarks \cite{Bernardoni:2007hi,Damgaard:2007ep}.
34: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%xs
35: \subsection{$\epsilon$-regime}
36: Topology plays a relevant r\^ole in the $\epsilon$-regime \cite{Leutwyler:1992yt}, such that observables must be considered in sectors of fixed topological charge.
37: In the quenched case, the current correlator in Eq.~(\ref{2p_f}) at NLO and fixed topology $\nu$ is given by \cite{Damgaard:2002qe, Hernandez:2002ds}
38: \begin{equation}\label{eq:ctnlo}
39: \mathcal{C}_{\nu}(t)=\frac{F^2}{2T}\left\{1+\frac{2\mu T^2}{F^2V} \sigma_\nu(\mu)h_1\left(\frac{t}{T}\right) \right\},
40: \end{equation}
41: with
42: \begin{equation}
43: h_1(\tau) =\frac{1}{2}\left[\left(|\tau|-\frac{1}{2}\right)^2-\frac{1}{12} \right],
44: \end{equation}
45: and
46: \begin{equation}\label{eq:cond}
47: \sigma_\nu(\mu)=\mu\left[I_\nu(\mu)K_\nu(\mu)+I_{\nu+1}(\mu)K_{\nu-1}(\mu) \right]+\frac{\nu}{\mu},
48: \end{equation}
49: where $I_\nu$ and $K_\nu$ are modified Bessel functions.
50: The most notable fact is that in the NLO expression only the leading-order LECs $\Sigma$ and $F$ enter, as already anticipated. \\
51: In our analysis we will compare both the time and topology dependence of the QCD correlators computed on the lattice with the expectations of the effective theory.
52: A convenient way to study the topology dependence is to fix $t=T/2$; in the chiral limit one has
53: \begin{equation}
54: \mu\sigma_\nu(\mu)|_{\mu=0}=|\nu|,
55: \end{equation}
56: hence one expects $\mathcal{C}_\nu(T/2)$ to depend linearly on the topological charge $\nu$. Moreover, one obtains the parameter-free prediction
57: \begin{equation}\label{sumrule}
58: 24L^3\left[\mathcal{C}_{\nu_1}(T/2)-\mathcal{C}_{\nu_2}(T/2) \right]|_{\mu=0}=|\nu_2|-|\nu_1|.
59: \end{equation}
60: From this expression and from Eq.~(\ref{eq:ctnlo}) it becomes clear that the sensitivity to topology is quite limited: in order for $\mathcal{C}_{{\nu}}(T/2)$ to be
61: significantly different from $\mathcal{C}_{\nu+\Delta\nu}(T/2)$ one needs the following condition on the relative error:
62: \begin{equation}
63: \frac{\Delta\mathcal{C}_{\nu}(T/2)}{\mathcal{C}_{\nu}(T/2)}\ll\frac{\Delta\nu}{12(FL)^2}\frac{T}{L}.
64: \end{equation}
65: Using the quenched value $F\simeq 100$ MeV from \cite{Giusti:2004yp}, this implies that statistical errors much smaller than $\sim[(14\Delta\nu)T/L]\%$ for $L= 1.5$ fm and $\sim[(8\Delta\nu)T/L]\%$ for $L= 2.0$ fm must be reached.
66: Notice that NLO effects are larger for asymmetric boxes, since they are proportional to $(T/L)^3$;
67: however, if $T\gg L$ one enters in a different kinematic range, called $\delta$-regime, which will not be discussed in this work.\\
68: For $\mu\ll 1$, the leading $\mu$-dependence in the NLO correction is given by
69: \begin{equation}\label{sigma_mass}
70: \mu\sigma_\nu(\mu)=\left\{
71: \begin{array}{ll}
72: |\nu|+\frac{\mu^2}{2|\nu|} +...& (\nu\neq 0)\\
73: \left[\frac{1}{2}-\gamma-\log\left(\frac{\mu}{2} \right) \right]\mu^2+...& (\nu=0),
74: \end{array}
75: \right.
76: \end{equation}
77: where $\gamma$ is the Euler-Mascheroni constant.
78: For $\nu\neq 0$ we then expect a weak sensitivity to quark mass for the current correlator.
79:
80: Matching the left correlator computed in lattice QCD with the chiral effective theory allows to extract the low-energy constant $F$ with control over NLO effects. For this particular correlator the chiral condensate $\Sigma$ appears only at NLO; in particular
81: the expression in Eq.~(\ref{eq:cond}) represents the quenched chiral condensate at finite $\mu$ at leading order in the $\epsilon$-expansion \cite{Osborn:1998qb,Damgaard:1998xy}.
82: At NLO, the condensate retains the same functional form of Eq.~(\ref{eq:cond}) \cite{Osborn:1998qb,Damgaard:2001js}, with $\mu$ replaced by $\mu_{\rm eff}=(m\Sigma_{\rm eff}V)$ and
83: \begin{equation}\label{sigmaeff}
84: \Sigma_{\rm eff}(V)=\Sigma\left[1+w_0\bar{H}(0)\right],
85: \end{equation}
86: where
87: \begin{equation}
88: \bar{H}(x)=\frac{1}{V}\sum_{p\neq 0}\frac{1}{(p^2)^2}\;e^{ipx}.
89: \end{equation}
90: Moreover, we have defined
91: \begin{equation}
92: w_0=\frac{m_0^2}{2N_c F^2},
93: \end{equation}
94: where $m_0^2/(2N_c)$ is the flavor singlet mass parameter;
95: as already anticipated, it is related to the topological susceptibility by the equation\footnote{We adopt the normalisation conventions of \protect\cite{Damgaard:2002qe,Hernandez:2002ds}.}
96: \begin{equation}\label{m0}
97: \frac{\langle \nu^2\rangle}{V}=\frac{m_0^2F^2}{4N_c}.
98: \end{equation}
99: In dimensional regularisation one obtains
100: \begin{equation}
101: \bar{H}(0)=\beta_2+\frac{1}{(4\pi)^2}\left[1+2c_1+\ln\left(\frac{\hat{L}^2}{L_0^2} \right) \right]
102: \end{equation}
103: where $\hat{L}=V^{1/4}$, $1/L_0$ is the ultraviolet subtraction point, and
104: \begin{equation}
105: c_1=\frac{1}{4-d}+\frac{1}{2}\left(-\gamma+\ln(4\pi) \right).
106: \end{equation}
107: $\beta_2$ is a shape coefficient \cite{Hasenfratz:1989pk}, which in the symmetric case $T=L$ takes the value
108: \begin{equation}
109: \beta_2=-0.020305.
110: \end{equation}
111: The infrared ``sickness'' of quenched QCD is reflected here in the fact that $\Sigma_{\rm eff}(V)$ diverges in the limit $\hat{L}\rightarrow \infty$. In the following we define operatively $\Sigma_{\rm eff}$ at a fixed $V$ as
112: \begin{equation}\label{sigmaeff1}
113: \left[\frac{2|\nu|\Sigma}{mV}\left(\sigma_\nu(\mu)-\frac{|\nu|}{\mu}\right)\right]_{m=0}\equiv\Sigma^2_{\rm eff}(L),\;\;\;\;\nu>0.
114: \end{equation}
115: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
116: \subsection{$p$-regime}
117: In the $p$-regime, the NLO finite-volume prediction for the current correlator in the quenched case is given by \cite{Colangelo:1997ch}
118: \begin{equation}\label{corr_que}
119: \mathcal C(t)=\frac{1}{2}M_P^V(F_P^V)^2 \frac{\cosh\left[(T/2-t)M_P^V\right]}{2\sinh\left[TM_P^V/2 \right]}.
120: \end{equation}
121: The pseudoscalar decay constant in this case is volume-independent:
122: \begin{equation}\label{eq:fp}
123: F_P^V=F_P=F\left[1+\frac{M^2}{2(4\pi F)^2}\alpha_5 \right],
124: \end{equation}
125: where $\alpha_i$ are the LECs associated with NLO operators in the (quenched) chiral Lagrangian in the convention of \cite{Heitger:2000ay}, and
126: \begin{equation}\label{eq:msq}
127: M^2=\frac{2m\Sigma}{F^2}.
128: \end{equation}
129: For the finite-volume pseudoscalar meson mass one obtains
130: \begin{eqnarray}
131: \left(M_P^{V}\right)^2 &= & M_P^2\left[1+w_0g_2(M_P,V) \right],\label{mp_fv1}\\
132: M_P^2 & = & M^2\left[1+w_0H(M^2) -\frac{M^2}{(4\pi F)^2}(\alpha_5-2\alpha_8) \right],\label{mp_fv2}
133: \end{eqnarray}
134: where $H(M^2)$ is given, in dimensional regularisation, by
135: \begin{equation}\label{h_dim}
136: H(M^2)=\int \frac{d^dp}{(2\pi)^d}\frac{1}{(p^2+M^2)^2}=\frac{1}{(4\pi)^2}\left[2c_1-\ln\left(\frac{M^2}{\mu^2} \right) \right].
137: \end{equation}
138: The volume-dependent function $g_r$ reads \cite{Hasenfratz:1989pk}
139: \begin{equation}
140: g_r(M_P,V)=\frac{1}{(4\pi)^2}\int_0^\infty \frac{d\lambda}{ \lambda^{3-r}}e^{-\lambda M_P^2}\sum_{n\in\mathbb{Z}^4}\left(1-\delta_{n,0}^{(4)} \right)\times
141: \end{equation}
142: $$
143: \exp\left[-\frac{1}{4\lambda}\left( T^2n_0^2+L^2\sum_{i=1}^3n_i^2 \right) \right].
144: $$
145: In our analysis we will investigate finite-volume effects by comparing lattice results obtained at different volumes $V_1$ and $V_2$. A convenient quantity to consider for this purpose is the ratio $M_P^{V_1}/M_P^{V_2}$, for which one obtains the NLO expression
146: \begin{equation}\label{eq:mp_ratio}
147: \left(\frac{M_P^{V_1}}{M_P^{V_2}}\right)^2=1+w_0\Big[g_2(M_P,V_1)-g_2(M_P,V_2) \Big].
148: \end{equation}
149: If $w_0$ is given as input, this is a parameter-free prediction from the chiral effective theory.\\
150: By reabsorbing the divergences in Eq.~(\ref{h_dim}) in the low-energy constant $\Sigma$ one obtains
151: \begin{equation}
152: \frac{M_P^2}{2m}=\frac{\Sigma(\mu)}{F^2}\left[1-\frac{w_0}{(4\pi)^2}\log\left(\frac{M^2}{\mu^2} \right)- \frac{M^2}{(4\pi F)^2}(\alpha_5-2\alpha_8) \right].
153: \end{equation}
154: $\Sigma(\mu)$ is related at NLO to the $\Sigma_{\rm eff}$ at a given scale $(L_{\rm eff})$ defined in the $\epsilon$-regime in Eq.~(\ref{sigmaeff1}) by
155: \begin{equation}\label{sigma_rel}
156: \Sigma_{\rm eff}(L_{\rm eff})=\Sigma(\mu)\left[1+w_0\left(\beta_2+\frac{1}{(4\pi)^2}\left(1+\log(L_{\rm eff}^2\mu^2)\right)\right) \right].
157: \end{equation}
158:
159:
160:
161:
162: