0705.3009/ms.tex
1: \documentclass{emulateapj}
2: \usepackage{apjfonts}
3: 
4: \def\etal{et~al.}
5: \def\diff{{\rm d}}
6: \def\teff{T_{\rm eff}}
7: \def\Mo{\rm{M}_\odot}
8: \def\muHz{\mu{\rm Hz}}
9: \def\note #1]{\noindent{\bf #1]}}
10: \def\alphaov{\alpha_{\rm ov}}
11: \def\Rob{R_{\rm ob}}
12: \def\sgnk2{{\rm sgn\left(K^2\right)}}
13: \def\rmd{{\rm d}}
14: \def\rmdd{{\rm d}^2}
15: \def\mF{\mathcal{F}}
16: \def\figdir{./}
17: 
18: \newcommand{\D}{\displaystyle}
19: \newcommand{\mitbf}[1] {\hbox{\mathversion{bold}$#1$}}
20: \newcommand{\vdag}{(v)^\dagger}
21: 
22: \shorttitle{ASTEROSEISMIC SIGNATURES OF CONVECTIVE CORES}
23: \shortauthors{CUNHA \& METCALFE}
24: 
25: \begin{document}
26:  
27: \title{Asteroseismic Signatures of Small Convective Cores}
28: 
29: \author{M.~S. Cunha\altaffilmark{1}}
30: \affil{Centro de Astrof{\'\i}sica da Universidade do Porto, Rua das 
31: Estrelas, 4150-Porto, Portugal}
32: \altaffiltext{1}{Visiting Scientist, High Altitude Observatory, NCAR, PO 
33: Box 3000, Boulder, CO 80307 USA}
34: 
35: \and
36: 
37: \author{T.~S. Metcalfe}
38: \affil{High Altitude Observatory and Scientific Computing Division, NCAR, 
39: PO Box 3000, Boulder, CO 80307 USA}
40: 
41: \email{mcunha@astro.up.pt, travis@hao.ucar.edu}
42: \slugcomment{Astrophysical Journal, accepted for publication}
43: 
44: \begin{abstract}
45: 
46: We present an analytical study of the effect of small convective cores on 
47: the oscillations of solar-like pulsators. Based on an asymptotic analysis 
48: of the wave equation near the center of the star, we derive an expression 
49: for the perturbations to the frequencies of radial modes generated by a 
50: convective core and discuss how these perturbations depend on the 
51: properties of the core. Moreover, we propose a diagnostic tool to isolate 
52: the predicted signature of the core, constructed from a particular 
53: combination of the oscillation frequencies, and we validate this tool with 
54: simulated data. We also show that the proposed tool can be applied to the 
55: pulsation data soon expected from satellite missions such as CoRoT and 
56: Kepler to constrain the amplitude of the discontinuity in the sound speed 
57: at the edge of the convective core, the ratio between the sound speed and 
58: the radius at this same location, and the stellar age.
59: 
60: \end{abstract}
61: 
62: \keywords{convection---methods: analytical---stars: interiors---stars: 
63: oscillations}
64: 
65: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
66: 
67: \section{Introduction}
68: 
69: Space-based asteroseismology began in 1999, when the 5\,cm star tracker 
70: mounted on the Wide-field InfraRed Explorer (WIRE) satellite was used to 
71: detect oscillations in the K giant $\alpha$~UMa \citep{buz00}, and this 
72: instrument has continued to produce a steady stream of asteroseismic data 
73: \citep[e.g., see][]{bru07}. The 15\,cm Micro-variability \& Oscillations 
74: of Stars (MOST) satellite was launched in 2003, and was the first 
75: satellite in orbit that was actually designed for asteroseismology 
76: \citep{wal03}. Although MOST's failure to detect solar-like oscillations 
77: in Procyon \citep{mat04} was later traced to a larger than expected source 
78: of non-white instrumental noise \citep{bed05}, its ability to conduct 
79: nearly uninterrupted time-series photometry for durations of up to two 
80: months has produced exquisite data for pulsating stars across the H-R 
81: diagram \citep[e.g., see][]{ran05,aer06,sai06,row07,cam07}.
82: 
83: The next era of space-based asteroseismology began with the successful 
84: launch of the 27\,cm Convection, Rotation and planetary Transits (CoRoT) 
85: satellite in December 2006 \citep{bag06}, and will continue with the 
86: anticipated launch of the 95\,cm Kepler satellite in 2008 \citep{jcd07}. 
87: These missions promise advances in the study of solar-like oscillations 
88: comparable to what WIRE and MOST have achieved for the larger amplitude 
89: classical pulsators. With the frequency precision expected to approach 
90: $\sim$0.1~$\mu$Hz from this next generation of satellites, we can begin to 
91: consider the possible detection of extremely subtle asteroseismic signals.
92: 
93: There is a long history of constructing combinations of the observed 
94: oscillation frequencies to extract useful asteroseismic information. The 
95: simplest such combination is the average frequency spacing between modes 
96: of consecutive radial order but the same spherical degree---{\it the large 
97: separation}---which provides an estimate of the mean stellar density 
98: \citep{cox80}. Similarly, the average spacing between modes of consecutive 
99: radial order but with spherical degrees that differ by 2---{\it the small 
100: separation}---is sensitive to chemical gradients in the deep interior, 
101: tracing the stellar evolutionary state \citep{dap88}.
102: 
103: The idea that sharp variations in the internal structure of distant stars 
104: generate signatures in the frequencies of low degree modes that can be 
105: recognized and interpreted with the appropriate seismic tools has been 
106: explored in a number of papers. In this context, theoretical derivations 
107: were carried out of the signal expected from sharp variations taking place 
108: at the boundary of convective envelopes \citep{monteiro00} and in the 
109: region of helium ionization \citep{monteiro98,houdek07}. Moreover, the 
110: expected signature of convective cores on the small frequency separations 
111: has also been considered by \citet{roxburgh01}. However, all of these 
112: works assumed that the sharp variation in the structure under 
113: investigation was located far from the turning points of the oscillation 
114: modes, i.e., well within their propagation cavities.  Consequently, the 
115: analyses carried out in these works do not directly apply to studies of 
116: small convective cores, such as those present in main-sequence solar-like 
117: pulsators.
118: 
119: Attempts to identify and interpret in a systematic manner the signatures 
120: of convective cores in solar-like pulsators directly from simulated data 
121: have recently been carried out by \citet{maz06}. Based on an analysis of 
122: the results obtained from simulated data, the authors suggested a 
123: diagnostic tool to estimate the masses of the convective cores and the 
124: stellar ages. While recognizing the importance of studying simulated data, 
125: the chances of constructing the optimal diagnostic tool to detect and 
126: characterize small convective cores may be greater if we know what signal 
127: to expect. Moreover, a theoretical understanding of the expected signal 
128: can provide us with a direct link between the structure under 
129: investigation and the observations---moving us toward our goal of finding 
130: the best way to infer information about that structure. With this in mind, 
131: in this paper we derive the expected signal of a small convective core on 
132: the oscillation frequencies of solar-like pulsators. Unlike 
133: \citet{monteiro00}, \citet{roxburgh01}, and \citet{houdek07}, we do not 
134: assume that the region of sharp structural variation, which in the present 
135: case is the edge of the convective core, is placed well within the 
136: propagation region of the modes. In fact, the analysis is focused on the 
137: opposite case, in which the edge of the convective core is sufficiently 
138: close to the center to affect the radial modes more significantly than the 
139: modes with degree $\ell\ge 1$.
140: 
141: We derive the theoretical expression for the signature of a small 
142: convective core in \S\ref{theory}. In \S\ref{tool} we construct a 
143: diagnostic tool, based on a particular combination of the frequencies, to 
144: isolate this signature. In \S\ref{DATA} we validate the theoretical 
145: predictions and the diagnostic tool using simulated data from a series of 
146: 1.3~$\Mo$ models. Moreover, we discuss the successes and limitations of 
147: the proposed tool, as well as the information that it can reveal. Finally, 
148: in \S\ref{DISC} we outline the potential for future work in this area, and 
149: we discuss possible observational tests.
150: 
151: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
152: 
153: \section{Signature of a convective core}
154: 
155: \subsection{\label{theory}Theoretical expectation}
156: 
157: Let ${\mitbf \xi}=(\xi,0,0)r\, e^{-i\omega t}$ be the displacement, 
158: defined with respect to a spherical coordinate system $(r,\theta,\phi)$, 
159: for radial, adiabatic, acoustic oscillations in a spherically symmetric 
160: star, with $\omega$ the oscillation frequency and $t$ the time. Through 
161: the appropriate mathematical manipulation of the equations of motion and 
162: mass conservation for radial pulsations, under the adiabatic 
163: approximation, the dimensionless displacement $\xi$ is found to obey the 
164: second order differential equation \citep[e.g.][]{gough93},
165: \begin{equation}
166: \frac{{\rm d}}{{\rm d}r}\left(\gamma p r^4\frac{{\rm d}\xi}{{\rm 
167: d}r}\right)+ \left\{r^3\frac{\rm d}{{\rm d}r}\left[\left(3\gamma 
168: -4\right)p\right]+r^4\rho\omega^2\right\}\xi =0,
169: \label{xi}
170: \end{equation}
171: where $\gamma$ is the first adiabatic exponent, $p$ is the pressure and 
172: $\rho$ is the density. By defining new dependent and independent 
173: variables, $\Xi=r_{\rm ref}^{1/2}r^{3/2}\rho^{1/2}c\,\xi$ and 
174: $z=\ln(r/r_{\rm ref})$ respectively, where $c=\sqrt{\gamma p/\rho}$ is the 
175: sound speed and $r_{\rm ref}$ is a fiducial value of $r$, and substituting 
176: into Eq.~(\ref{xi}), the latter reduces to the standard form,
177: \begin{equation}
178: \frac{{\rm d}^2 \Xi}{{\rm d}z^2}+K^2\Xi=0,
179: \label{waveeq}
180: \end{equation}
181: where
182: \begin{equation} 
183: K^2=\frac{\omega^2-\omega_{\rm c}^2}{c^2}\,r^2-\frac{1}{4},
184: \nonumber
185: \end{equation}
186: and $\omega_{\rm c}$ is a critical acoustic frequency \citep[e.g., 
187: see][Eq.~4.8.7]{gough93}. Since our aim is to study the effect of a small 
188: convective core on the eigenfrequencies, we are particularly interested in 
189: the form of $\omega_c$ close to the center of the star. From the general 
190: expression for $\omega_{\rm c}$, we find that when $r\rightarrow 0$, 
191: $\omega_{\rm c}^2\approx 2c^2/r^2$. Using this expression for $\omega_{\rm 
192: c}^2$, we then find the critical frequency $\omega_{\rm c}^*$ at which 
193: $K^2=0$ to be,
194: \begin{equation}
195: \omega_{\rm c}^*\approx\frac{3\,c}{2\,r}.
196: \label{wc}
197: \end{equation}
198: 
199: A model with a small convective core exhibits a sharp structural variation 
200: close to the inner turning point of Eq.~(\ref{waveeq}). This sharp 
201: variation can be seen in the sound speed profiles of main-sequence models 
202: with mass $M=1.3\,\, \Mo$, shown in Figure~\ref{sspeed} for a range of 
203: stellar ages. Our aim is to determine how the eigenfrequencies of an 
204: otherwise similar model with a smooth structure around the edge of the 
205: convective core---hereafter the {\it unperturbed model}---are modified by 
206: the presence of this sharp variation.
207: 
208: % FIGURE 1 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
209: \begin{figure}
210: \plotone{f1.eps}
211: \caption{Sound speed in the inner layers of main-sequence models with 
212: $M=1.3\,\Mo$, and ages ranging from 0.25~Gyr (top curve) to 5.25~Gyr 
213: (bottom curve) in intervals of 0.5~Gyr. The discontinuity in $c$ marks the 
214: edge of the convective core, which expands during the evolution of the 
215: star up to an age of 3.25~Gyr and then begins to contract.}
216: \label{sspeed}
217: \end{figure}
218: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
219: 
220: To achieve this we must find an expression for $\Xi$ that is valid close 
221: to the inner turning point of Eq.~(\ref{waveeq}). Such an expression for 
222: $\Xi$ is derived in Appendix \ref{APA}. In short, we define new dependent 
223: and independent variables and substitute them into Eq.~(\ref{waveeq}) to 
224: obtain an equation that resembles the Airy equation with an additional 
225: term (cf.~Eq.~\ref{airy} in Appendix~\ref{APA}). Applying Olver's 
226: comparison method \citep{olver74} we then find, close to the inner turning 
227: point,
228: \begin{eqnarray}
229: \Xi \approx a \,|x|^{1/4}|K|^{-1/2}\left[A_{\rm 
230: i}\left(0\right)- {A^\prime}_{\rm i}\left(0\right)x\right],
231: \label{xiapprox}
232: \end{eqnarray}
233: where $a$ is a constant related to the normalization of 
234: eigenfunctions, $A_{\rm i}(0)$ and ${A^\prime}_{\rm i}(0)$ are the values 
235: taken at $x=0$ of the first solution to the Airy equation and its first 
236: derivative respectively, and $x$ is the new independent variable, which is 
237: related to $z$ through Eq.~(\ref{defx}) in Appendix~\ref{APA}. In deriving 
238: Eq.~(\ref{xiapprox}) we have neglected terms of $\mathcal{O}(x^3)$. So, 
239: this equation is approximately correct when $x$ is sufficiently close to 
240: 0. Note that unlike $r$, the independent variable $x$ depends on the value 
241: of the eigenfrequency. For a given frequency it goes through 0 at the 
242: inner turning point, and is negative when $K^2 < 0$ and positive when $K^2 
243: > 0$. The relation between $x$ and $r$ for different values of the 
244: eigenfrequency $\omega$, as well as the regions where $K^2 <0$ and 
245: $K^2>0$, are shown in Figure~\ref{pd} for an unperturbed stellar model 
246: with $M=1.3\,\,\Mo$. The unperturbed models were obtained by fitting the 
247:  sound speed profiles of each stellar model with smooth 
248: functions around the edge of the convective core. The density profile was 
249: then derived by assuming hydrostatic equilibrium. 
250: 
251: To study the effect of a small convective core on the oscillation 
252: frequencies, we start from an integral equation for the frequency, which 
253: can be derived from Eq.~(\ref{xi}) after multiplying by $\xi$ and 
254: integrating by parts between $r=0$ and $r=R$, where $R$ is the radius at 
255: the stellar surface. The resulting well known equation is,
256: \begin{equation}
257: \omega^2=\frac{I_2-{\mathcal B}}{I_1},
258: \label{vp}
259: \end{equation}
260: where
261: \begin{eqnarray}
262: I_2 & = &\int_0^R{\left[\gamma pr^4\left(\frac{d\xi}{dr}\right)^2-
263: r^3\frac{d}{dr} \left[\left(3\gamma-4\right)p\right] \xi^2\right]dr}, 
264: \nonumber\\
265: I_1 & = & \int_0^R{\left[\rho r^4\xi^2\right]dr}, 
266: \nonumber
267: \end{eqnarray}
268: and
269: \begin{eqnarray}
270: {\mathcal B} & = & {\mathcal B}(R)-{\mathcal B}(0) = 
271: \left[\gamma pr^4\xi\frac{d\xi}{dr}\right]_{r=0}^{r=R}. 
272: \nonumber
273: \end{eqnarray}
274: When the oscillations are below the acoustic cutoff frequency, the surface 
275: term ${\mathcal B}$ can safely be neglected. In fact, it follows from the 
276: regularity condition at $r=0$ that ${\mathcal B}(0)=0$ and, given the 
277: small value of $p(R)$, if the oscillations are evanescent in the outer 
278: layers ${\mathcal B}(R)$ will also be small.
279: 
280: We now consider the above mentioned unperturbed smooth model and an 
281: otherwise similar model with a sharp structural variation at the edge of a 
282: small convective core---hereafter the {\it perturbed model}. If ${\mathcal 
283: B}$ is neglected in Eq.~(\ref{vp}), the difference between the oscillation 
284: frequencies in the perturbed and unperturbed models in a linear 
285: approximation is given by
286: \begin{equation}
287: \delta\omega=\frac{\delta I_2-\omega^2\delta I_1}{2\omega I_1},
288: \label{vpp}
289: \end{equation}
290: where all unperturbed quantities refer to the smooth model. Equation 
291: (\ref{vp}) with ${\mathcal B}$ neglected constitutes a variational 
292: principle for the frequency $\omega$ 
293: \citep[e.g.][]{chandrasekhar64,gough93}. Thus, to first order the 
294: perturbation to the eigenfunctions does not contribute to the perturbation 
295: to the eigenfrequencies and we can write,
296: \begin{eqnarray}
297: \delta I_1 = \int_0^R \delta\rho r^4\xi^2{\rm dr},
298: \label{di1}
299: \end{eqnarray} 
300: and
301: \begin{eqnarray}
302: \delta I_2 & = & \int_0^R\left\{r^4\delta\left(\gamma p\right)
303: \left(\frac{{\rm d}\xi}{{\rm d}r}\right)^2\right. 
304: \nonumber\\ & & 
305: \left.-r^3\frac{\rm d}{{\rm d}r}\left[3\delta\left(\gamma p\right)
306: -4\delta p\right]\xi^2\right\}{\rm d}r,
307: \label{di2}
308: \end{eqnarray}
309: where $\delta\rho$, $\delta p$ and $\delta\gamma$ denote perturbations at 
310: fixed radius.
311: To obtain an expression for $\delta\omega$ as a function of $\omega$, we 
312: manipulate Eqs.~(\ref{vpp}-\ref{di2}) in a way very similar to that 
313: presented in \citet{monteiro94}. In particular, Eqs.~(\ref{di1}) and 
314: (\ref{di2}) are written in terms of the perturbations $\delta (\gamma p)$, 
315: $\delta c^2$ and $\delta g$, with $g=-\frac{1}{\rho} \frac{{\rm d}p}{{\rm 
316: d}r}$ and then integrated by parts (see Appendix~\ref{APB} for details). 
317: As discussed in Appendix~\ref{APB}, the perturbation to the 
318: eigenfrequencies is determined predominantly by the terms associated with 
319: $\delta c^2$.  Describing this perturbation with a modified step function, 
320: we find (cf.~Eq.~\ref{final})
321: \begin{eqnarray}
322: \delta\omega \approx -\frac{A_\delta}{2\omega I_1}\left[\frac{|K|r^3\rho 
323: c^2\xi^2}{|x|^{1/2}}\right]_{x=0}\left[\frac{r\left(4g+\omega^2r\right)}
324: {c^2|K|^2}\right]_{x=x_{\rm d}} 
325: \nonumber\\ 
326: \times \left[|x|x-\frac{A^\prime_{\rm i}(0)}{A_{\rm i}(0)}|x|x^2+\frac{1}{3}
327: \frac{A^\prime_{\rm i}(0)^2}{A_{\rm i}(0)^2}|x|x^3\right]_{x=x_{\rm d}},
328: \label{dw2}
329: \end{eqnarray}
330: where $A_\delta$ is a positive constant related to the sound speed 
331: increase at the edge of the convective core and $x_{\rm d}$ is the value 
332: taken by the independent variable $x$ at the location of the edge of the 
333: convective core. Both $x$ and $x_{\rm d}$ are functions of frequency. 
334: Through this dependence, Eq.~(\ref{dw2}) expresses how the frequency 
335: perturbation is modulated with frequency. According to this equation the 
336: frequency perturbation goes through zero when $x_{\rm d}=0$, i.e.~when the 
337: frequency is such that the turning point and the edge of the convective 
338: core are located at the same depth.
339: 
340: % FIGURE 2 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
341: \begin{figure} 
342: \plotone{f2.eps} 
343: \caption{The full curve shows the
344: critical frequency obtained by imposing $K^2=0$ for the inner layers
345: of an unperturbed main-sequence model with $M=1.3\, \Mo$ at the age
346: when its convective core reaches maximum size (3.25~Gyr). For each
347: frequency, this curve gives the maximum propagation depth of a radial
348: mode when the eigenfunctions are the solutions to
349: Eq.~(\ref{waveeq}). The vertical line shows the radial position of the
350: edge of the convective core. The horizontal lines are axes for the
351: independent variable $x$ in the range $[-0.5,0.5]$. These are shown for 
352: eigenmode frequencies of the corresponding perturbed model, with radial 
353: orders ranging from $n=11$ (bottom line) to $n=20$ (top line).}
354: \label{pd} 
355: \end{figure}
356: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
357: 
358: If the signal anticipated by Eq.~(\ref{dw2}) could be detected in the 
359: oscillation frequencies of real stars, then information about the size and 
360: position of the sound speed discontinuity at the edge of the convective 
361: core could be inferred from the amplitude $A_\delta$ and the frequency at 
362: which $\delta\omega=0$, respectively. With this in mind, in the next 
363: section we present a diagnostic tool that will allow us to isolate the 
364: signal derived above.
365: 
366: \subsection{\label{tool}Diagnostic tool}
367: 
368: When deriving Eq.~(\ref{dw2}) we assumed that our model differs from a 
369: smooth model only at the edge of the convective core. However, sharp 
370: variations in the internal structure of a star are known to occur also at 
371: other locations, such as the base of the outer convective region and the 
372: region of helium ionization. These, in turn, will generate additional 
373: perturbations to the eigenfrequencies, as compared with those of a smooth 
374: model \citep[e.g.,][]{monteiro94,basu97,monteiro05,houdek07}. Thus, to 
375: isolate the signal produced by the edge of a convective core, we must 
376: combine the oscillation frequencies of real stars in such a way as to 
377: cancel out, as much as possible, the signal coming from all other sources. 
378: With this in mind, we write the frequency 
379: $\nu_{n,\ell}=\omega_{n,\ell}/2\pi$ of a mode with radial order $n$ and 
380: degree $\ell$ in a given star as the sum of three components
381: \begin{equation}
382: \nu_{n,\ell} = \nu_{n,\ell}^s+\delta\nu_{n,\ell}^p+\delta\nu_{n,\ell}^c,
383: \label{mu}
384: \end{equation}
385: where $\nu_{n,\ell}^s$ is the frequency of the mode in the smooth model, 
386: $\delta\nu_{n,\ell}^p$ is the perturbation to that frequency arising from 
387: all sharp variations in the structure of the star taking place well within 
388: the propagation region of the same mode (which we shall consider to be of 
389: low degree, with $\ell\le 3$), and $\delta\nu_{n,\ell}^c$ is the frequency 
390: perturbation produced by the edge of the small convective core.
391: 
392: While sharp variations in the envelope of a star will affect the 
393: frequencies of all modes of low degree, the discontinuity in the sound 
394: speed at the edge of a small convective core for a star slightly more 
395: massive than the Sun might be expected to affect primarily the frequencies 
396: of radial modes. This is not to say that the frequencies of modes of 
397: degree $\ell=1$, which also propagate to very deep layers, will not be 
398: affected at all by the convective core. In fact, to correctly derive the 
399: expected signal for a dipolar mode, the Cowling approximation should be 
400: avoided. The derivation should then be started from the formalism recently 
401: presented by \citet{takata06}, which shows that the inner turning point of 
402: dipolar modes is modified when the equations are treated without 
403: neglecting the perturbation to the gravitational potential. However, such 
404: an analysis is beyond the scope of the present paper. Hence, based on the 
405: idea that radial modes are most sensitive to the deepest layers of a star, 
406: we will proceed by assuming $\delta\nu_{n,\ell}^c\approx 0$ for modes with 
407: $\ell\ge 1$. This assumption will be tested in the following section. 
408: Moreover, given that we are interested in modes with degree $\ell\leq 3$, 
409: we neglect the weak dependence of $\delta\nu_{n,\ell}^p$ on $\ell$ 
410: \citep[e.g.,][]{monteiro94,houdek07} when subtracting the frequencies of 
411: nearly degenerate modes. With these assumptions, the {\it scaled small 
412: separations} 
413: $D_{\ell,\ell+2}\equiv\left(\nu_{n,\ell}-\nu_{n-1,\ell+2}\right)/ 
414: \left(4\ell+6\right)$ for pairs of modes with degree $\ell\leq 3$ divided 
415: by the large separations 
416: $\Delta\nu_{n,\ell}\equiv\nu_{n+1,\ell}-\nu_{n,\ell}$ take the form
417: \begin{eqnarray}
418: \frac{D_{02}}{\Delta\nu_{n-1,1}}& \approx &
419: \frac{\nu_{n,0}^s-\nu_{n-1,2}^s}{6\,{\Delta\nu_{n-1,1}}}+
420: \frac{\delta\nu_{n,0}^c}{6\,{\Delta\nu_{n-1,1}}},
421: \label{ss02}\\
422: \frac{D_{13}}{\Delta\nu_{n,0}}& \approx &
423: \frac{\nu_{n,1}^s-\nu_{n-1,3}^s}{10\,{\Delta\nu_{n,0}}}
424: \label{ss13}.
425: \end{eqnarray}
426: In the high-frequency asymptotic regime, the quantity 
427: \begin{equation}
428: D_\delta\equiv\frac{\pi}{4\ell+6}\frac{\nu_{n,\ell}^s-\nu_{n-1,\ell+2}^s}
429: {{\Delta\nu_{n,\ell}}}
430: \label{bigd}
431: \end{equation}
432: in the unperturbed model is essentially independent of the pair of modes 
433: considered \citep[$\ell=0,2$ or $\ell=1,3$; 
434: see][]{roxburgh00b,roxburgh00a,roxburgh03}. Thus, we proceed by 
435: subtracting Eq.~(\ref{ss13}) from Eq.~(\ref{ss02}) to find the relation
436: \begin{equation}
437: \frac{D_{02}}{\Delta\nu_{n-1,1}}-\frac{D_{13}}{\Delta\nu_{n,0}}\approx
438: \frac{\delta\nu_{n,0}^c}{6\,{\Delta\nu_{n-1,1}}}.
439: \label{signal}
440: \end{equation}
441: Note that for the large separations appearing in the denominators, we used 
442: pairs of modes that encompass the frequency range of the modes used in the 
443: scaled small separations, which appear in the numerators. When defined in 
444: this way, the frequency differences presented in Eq.~(\ref{signal}) are a 
445: smoother function of frequency than when $\Delta\nu_{n,0}$ and 
446: $\Delta\nu_{n,1}$ are used in the denominators of the first and second 
447: term, respectively. In practice, this means that when subtracting 
448: Eq.~(\ref{ss13}) from Eq.~(\ref{ss02}) we also had to assume that 
449: $\frac{\Delta\nu_{n,0}}{\Delta\nu_{n-1,1}}\approx 
450: \frac{\Delta\nu_{n,1}}{\Delta\nu_{n,0}}$.
451: 
452: The frequency perturbation $\delta\nu_{n,0}^c$ is expected to be modulated 
453: with frequency according to Eq.~(\ref{dw2}). Thus, if the assumptions 
454: leading to Eq.~(\ref{signal}) hold, we would expect to see a corresponding 
455: modulation when the oscillation frequencies observed in a given solar-like 
456: pulsator are combined in the manner prescribed. In the next section we 
457: test this prediction on simulated data and we discuss what might be 
458: inferred about solar-like stars from this diagnostic tool, once 
459: observations with the necessary precision become available.
460: 
461: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
462: 
463: \section{\label{DATA}Tests with simulated data}
464: 
465: \subsection{\label{models}Models}
466: 
467: We used the Aarhus stellar evolution code \citep[ASTEC;][]{jcd82} coupled 
468: with the adiabatic pulsation code (ADIPLS) to calculate the theoretical 
469: oscillation frequencies for 1.3~$\Mo$ models with ages from 0.25 to 5.25 
470: Gyr, sampling the full range of convective core sizes along the evolution 
471: track. The ASTEC code used the equation of state (EOS) of \cite{eff73} 
472: without Coulomb corrections, and opacities from the OPAL tables 
473: \citep{ir96}, supplemented by Kurucz opacities at low temperatures. The 
474: nuclear reaction rates came from \cite{bp92}, convection was described by 
475: the mixing-length theory of \cite{bv58}, and we did not include the 
476: effects of diffusion.
477: 
478: We set the initial metallicity to $Z_0=0.02$, and the initial hydrogen 
479: mass fraction to $X_0=0.74$. We fixed the mixing-length parameter at 
480: $\alpha=1.9$ and included core overshoot using $\alpha_{\rm ov}=0.25$ with 
481: complete mixing in the overshoot region and considering the associated 
482: changes to the adiabatic temperature gradient. For each selected model, we 
483: calculated the radial and non-radial p-mode frequencies for oscillations 
484: with spherical degree $\ell$=0-3 and radial order $n$=1$\sim$30. We used 
485: these frequencies to construct the combinations shown in 
486: Eq.~(\ref{signal}), and compared them to the theoretical predictions.
487: 
488: % FIGURE 3 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
489: \begin{figure}
490: \plotone{f3.eps}
491: \caption{Frequency differences, as defined by the left hand side of 
492: Eq.~(\ref{signal}), for a model with $M=1.3$~M$_\odot$ at an age of 0.25 
493: Gyr when its convective core is very small and there is no discontinuity 
494: in the sound speed (see the top curve in Figure~\ref{sspeed}).}
495: \label{difnocc}
496: \end{figure}
497: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
498: 
499: \subsection{\label{tests}Testing the diagnostic tool}
500: 
501: A number of approximations were made when deriving the diagnostic tool 
502: proposed in Eq.~(\ref{signal}). Thus, it is essential that we evaluate the 
503: ability of this seismic tool when it comes to isolating the signal from 
504: the convective core.
505: 
506: % FIGURE 4 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
507: \begin{figure*}
508: \plottwo{f4a.eps}{f4b.eps}
509: \caption{Left panel: the same as Figure~\ref{difnocc}, but for models with 
510: ages ranging from 0.25 Gyr (bottom left) to 5.25 Gyr (top left) in 
511: intervals of 0.5 Gyr. Error bars are shown for the model with maximum 
512: convective core size. These correspond to $1 \sigma$ errors in the 
513: frequency differences when the relative error in the individual 
514: frequencies is assumed to be $10^{-4}$. For this model, the radial order 
515: of the modes spanning the frequency range considered varies between $n=9$ 
516: and $n=27$. Right panel: comparison between the theoretical signal 
517: expected from the convective core (dashed lines) and that contained in the 
518: frequency differences (full curves) for models with ages ranging from 2.25 
519: Gyr (bottom left) to 3.75 Gyr (top left) in intervals of 0.5 Gyr. To 
520: determine the expected signal, we computed $\delta\nu^{\rm c}_{n,0}$ using 
521: Eq.~(\ref{dw2}) and then normalized it by the large separations to 
522: reproduce the right hand side of Eq.~(\ref{signal}).}
523: \label{comparison}
524: \end{figure*}
525: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
526:    
527: With this in mind, we start by testing the last assumption made when 
528: deriving Eq.~(\ref{signal})---namely, the weak dependence of the value of 
529: $D_\delta$ in the unperturbed model on the pair of modes considered. To 
530: this end, we compute eigenfrequencies for our youngest model. At an age of 
531: 0.25~Gyr the convective core is extremely small, and there is no apparent 
532: discontinuity in the sound speed (see the top curve in 
533: Figure~\ref{sspeed}). Under these conditions, we expect 
534: $\delta\nu_{n,0}^{c}\approx 0$. Consequently, if the quantity $D_\delta$ 
535: were strictly independent of the pair of modes considered, the frequency 
536: differences defined by the expression on the left hand side (LHS) of 
537: Eq.~(\ref{signal}) would be approximately zero for all frequencies.
538: 
539: Figure~\ref{difnocc} shows the frequency differences obtained for our 
540: youngest model. Clearly, in the lower frequency domain the frequency 
541: differences deviate significantly from zero. This reflects the fact that 
542: $D_\delta$ is nearly independent of the pair of modes considered only in 
543: the high-frequency asymptotic regime. At lower frequencies this 
544: assumption breaks down, and consequently so does the approximation made 
545: when deriving Eq.~(\ref{signal}) from Eqs.~(\ref{ss02}) and (\ref{ss13}). 
546: Nevertheless, at high frequencies ($\ga 2$~mHz in Figure~\ref{difnocc}) 
547: the frequency differences converge to a constant small value, which we 
548: shall designate $\nu_{\rm asymp}$, confirming our assumption of the near 
549: degeneracy of $D_\delta$ with respect to the pairs of modes $\ell=0,2$ and 
550: $\ell=1,3$.
551: 
552: To compare these results with those obtained for older models, we 
553: determine the frequency differences defined by the LHS of 
554: Eq.~(\ref{signal}) for our sequence of models, with ages ranging from 
555: 0.25~Gyr to 5.25~Gyr. The results are shown in the left panel of 
556: Figure~\ref{comparison}. It is clear from this figure that in the 
557: high-frequency domain, where Eq.~(\ref{signal}) is expected to be valid, 
558: the absolute value of the slope of the frequency differences increases 
559: with age. This can be understood as a direct consequence of the increase 
560: with age of the sound speed discontinuity at the edge of the convective 
561: core. According to Eq.~(\ref{dw2}), such an increase manifests itself 
562: through the amplitude $A_\delta$ by an increase in the absolute value of 
563: the frequency perturbation at fixed frequency. This increase is 
564: consequently reflected in the frequency differences defined by the LHS of 
565: Eq.~(\ref{signal}), and thus in the slope of the curves plotted in 
566: Figure~\ref{comparison}.
567: 
568: To determine whether the signature of the convective core on the frequency 
569: differences might be detected in future observations of solar-like 
570: pulsators, we have included for one of our models the expected 1$\sigma$ 
571: error bars for the frequency differences, calculated with the assumption 
572: that the relative error on the individual frequencies is 10$^{-4}$ 
573: \citep[e.g., 0.2~$\mu$Hz at 2~mHz as expected from CoRoT;][]{bag06}. It is 
574: evident from the figure that if individual frequencies are indeed 
575: determined with such precision, we will be able to not only detect the 
576: effect of the convective core on the frequencies, but also distinguish 
577: between different possible models based on this signature. Moreover, we 
578: see that our ability to detect this signature in real data will depend on 
579: the number of radial modes detected, as well as on their radial order.
580: 
581: We made additional assumptions, other than the one referred to at the 
582: beginning of this section, while deriving Eq.~(\ref{signal}). In 
583: particular, we assumed that the effect of the convective core on the 
584: frequencies of modes with degree $\ell \ge 1$ is negligible, when compared 
585: to the effect on the frequencies of radial modes. To check whether this 
586: assumption is valid, we compare the frequency differences calculated for a 
587: subset of our models with the theoretical expectations derived from 
588: Eq.~(\ref{dw2}). The results are shown in the right panel of 
589: Figure~\ref{comparison}. We see from the figure that in the high-frequency 
590: asymptotic regime the theoretical signal derived from the analysis of the 
591: wave equation near the inner turning point is indeed isolated when the 
592: oscillation frequencies are combined in the way suggested by 
593: Eq.~(\ref{signal}). We note that there is no model dependent free 
594: parameter in the theoretical expression for the frequency perturbations. 
595: The difference in the slopes of the theoretical curves at high-frequency, 
596: seen also in the frequency differences calculated from simulated data, is 
597: thus a direct consequence of the dependence of the amplitude $A_\delta$ on 
598: the increase in the sound speed at the edge of the convective core.
599: 
600: % FIGURE 5 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
601: \begin{figure*}
602: \plottwo{f5a.eps}{f5b.eps}
603: \caption{Left panel: measurement of the slope of the frequency differences 
604: in the high-frequency domain as a function of the relative increase in the 
605: squared sound speed at the edge of the convective core (see text for the 
606: mathematical definition of $\Delta$). The ages of the models shown range 
607: from 2.25 Gyr (top left) to 5.25 Gyr (bottom right) in intervals of 
608: 0.25~Gyr. Right panel: for models with ages ranging from 2.25 Gyr to 5.25 
609: Gyr, we show the frequency $\nu_{\rm max}$ at which the frequency 
610: differences suggested by the LHS of Eq.~(\ref{signal}) reach their maximum 
611: (crosses), and for the same models the value of $\,\,(2\pi)^{-1}\,3c/2r$ (circles) at the edge of the convective core. 
612: Since $c$ is discontinuous at the edge of the convective core, we use the 
613: mean of the values of $c$ on the left and right side of the discontinuity 
614: when calculating this expression. For models with ages greater than 4 Gyr, 
615: we show the frequency $\nu_{\rm cross}$ at which the frequency differences 
616: cross the asymptotic value for the youngest model (diamonds).}
617: \label{slope}
618: \end{figure*}
619: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
620: 
621: In contrast to these results, Figure~\ref{comparison} shows that in the 
622: low-frequency domain the combination of frequencies suggested by 
623: Eq.~(\ref{signal}) does not properly isolate the theoretical signature 
624: expected from Eq.~(\ref{dw2}). Just as in the case discussed for 
625: Figure~\ref{difnocc}, this is likely a consequence of the breakdown at low 
626: frequencies of the assumption that $D_\delta$ is independent of the pair 
627: of modes considered. According to Figure~\ref{difnocc}, when the frequency 
628: perturbation associated with the convective core is negligible, the 
629: frequency differences become negative in the low-frequency domain, 
630: decreasing as the frequency decreases. On the other hand, the frequency 
631: perturbation associated with the convective core becomes positive for 
632: frequencies below a particular value---namely, the frequency at which the 
633: position of the discontinuity in the sound speed coincides with the inner 
634: turning point of Eq.~(\ref{waveeq}). In the low-frequency domain, we thus 
635: expect our diagnostic tool to show a superposition of these two behaviors. 
636: Indeed, this is seen for frequencies smaller than $\sim$1.5~mHz in the 
637: left panel of Figure~\ref{comparison}. The competition between these two 
638: behaviors results in the presence of a maximum in the frequency 
639: differences calculated for each model.
640: 
641: \subsection{Characterizing the convective core}
642: 
643: The validation of the diagnostic tool proposed in Eq.~(\ref{signal}) in 
644: the high-frequency domain makes us confident that we understand the origin 
645: of the signature observed in the simulated data, and that we can use this 
646: signature to extract information about the convective core. As mentioned 
647: in \S\ref{theory} and \S\ref{tests}, information about the size of the 
648: sound speed discontinuity at the edge of the convective core can be 
649: extracted from the amplitude of the signal, and in principle information 
650: about the ratio $[c/r]_{\rm edge}$ at the edge of the convective core can 
651: be extracted from the value of the frequency at which the frequency 
652: perturbation goes through zero. Moreover, given the age dependence of the 
653: size of the sound speed discontinuity at the edge of the core, the 
654: signature discussed above might also be used to infer the evolutionary 
655: status of pulsating stars within the region of the H-R diagram considered 
656: here.
657: 
658: These ideas are illustrated in Figure~\ref{slope}. In the left panel the 
659: quantity $\Delta$, defined as the difference between the frequency 
660: differences $\left(\frac{D_{02}}{\Delta\nu_{n-1,1}}- 
661: \frac{D_{13}}{\Delta\nu_{n,0}}\right)$ of modes with radial orders $n=k+2$ 
662: and $n=k-2$, with $k$ such that $\nu_k \sim 1750$~$\mu$Hz for each model, is 
663: shown as a function of the relative increase in the squared sound speed at 
664: the edge of the convective core. The quantity $\Delta$ provides a measure 
665: of the slope of the frequency difference curves, such as those plotted in 
666: the left panel of Figure~\ref{comparison}, in the high-frequency domain. 
667: It is clear from Figure~\ref{slope} that a linear relation exists between 
668: $\Delta$ and the squared sound speed increase at the edge of the 
669: convective core. We also considered alternative quantities indicative of 
670: the slope of the same curves taken at constant radial order rather than at 
671: constant frequency. In all cases the results indicate a linear relation 
672: with the squared sound speed increase at the edge of the convective core. 
673: Note that when studying data with noise, the use of the quantity $\Delta$ 
674: defined by only two data points is not advisable. Instead, it would be 
675: more convenient to extract the slope of the frequency difference curves in 
676: the high-frequency domain through a least-squares fit to the appropriate 
677: part of the curve under consideration.
678: 
679: Since the proposed diagnostic tool fails to reproduce the theoretical 
680: signal in the low-frequency domain, the idea of using this tool to 
681: determine the value of the frequency at which the frequency perturbation 
682: associated with the convective core goes through zero---and thus 
683: extracting information about the physical size of the core---becomes 
684: compromised. Nevertheless, since the maxima that characterize the curves 
685: defined by the frequency differences are a consequence of the positive 
686: frequency perturbations in the low-frequency domain, we might expect the 
687: position of these maxima to contain information about the size of the 
688: convective core. With this in mind, we have determined for each model the 
689: frequency $\nu_{\rm max}$ at which the frequency differences defined by 
690: the LHS of Eq.~(\ref{signal}) reach their maximum value. These are plotted 
691: as a function of age in the right panel of Figure~\ref{slope}, and are 
692: compared with the frequencies $\left[(2\pi)^{-1}\,3c/2r\right]_{\rm 
693: edge}$ at which the frequency perturbation associated with the convective 
694: core goes through zero.
695: 
696: Clearly, the frequency at which the frequency differences reach their 
697: maxima, indicated by the crosses in Figure~\ref{slope}, is strongly 
698: correlated with the age of the model. This is not surprising since the 
699: magnitudes of the frequency perturbations, which influence the position of 
700: the maxima, are also correlated with age. Nevertheless, a change in the 
701: slope is noticeable around an age of $\sim$4~Gyr. This is likely a 
702: consequence of the contracting core. As seen in Figure~\ref{sspeed}, when 
703: the core begins to contract, the sound speed variation near the edge of 
704: the convective core begins to spread over a larger radial extent. Our 
705: theoretical analysis, which assumes that the sound speed variation at the 
706: edge of the core is well represented by a modified step function, becomes 
707: less adequate in this case. Nevertheless, an inspection of the frequency 
708: differences shown in the left panel of Figure~\ref{comparison} indicates 
709: that in the older models where this effect becomes evident, the amplitude 
710: of the frequency differences in the low-frequency domain is indeed 
711: substantially larger than in all other models. This may result from a 
712: sudden increase in the amplitude of the frequency perturbation in the 
713: low-frequency domain, or from an improvement in the degeneracy of 
714: $D_\delta$ calculated for pairs of modes with degrees $\ell=0,2$ and 
715: $\ell=1,3$. Whatever the case, the relative importance of the latter in 
716: determining the functional form of the frequency differences appears to be 
717: smaller for models with ages greater than $\sim$4~Gyr. Thus, one might 
718: expect that for these models the frequency 
719: $\left[(2\pi)^{-1}\,3c/2r\right]_{\rm edge}$ will be better estimated 
720: by the frequency $\nu_{\rm cross}$ at which the frequency differences 
721: cross the asymptotic value for the youngest model in Figure~\ref{difnocc}. 
722: Indeed this seems to be the case, as shown by the diamonds in the right 
723: panel of Figure~\ref{slope}.
724: 
725: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
726: 
727: \section{\label{DISC}Discussion}
728: 
729: We have derived the expected signature of small convective cores on the 
730: oscillation frequencies of solar-like pulsators, we have suggested a 
731: diagnostic tool to isolate this signature, and finally we have discussed 
732: what information the signature may reveal about the convective cores 
733: themselves and about the evolutionary status of the corresponding stars.
734: 
735: The results demonstrate that our ability to isolate the convective core 
736: signature with the suggested diagnostic tool is significantly better in 
737: the high-frequency asymptotic domain than at lower frequencies. 
738: Consequently, the information inferred from this diagnostic tool at high 
739: frequencies is expected to be more robust. In particular, this is the case 
740: for the relation found between the slope of the frequency difference 
741: curves and the relative increase in the squared sound speed at the edge of 
742: the convective core. Based on the theoretical signal, which in the 
743: high-frequency domain is well represented by the proposed diagnostic tool, 
744: we expect this relation to hold also in a more general context when 
745: models of different mass, chemical composition, or with different amounts 
746: of core overshoot are considered.
747: 
748: By contrast, information inferred from the behaviour of the diagnostic 
749: tool at lower frequencies, such as the estimation of the ratio $[c/r]_{\rm 
750: edge}$ from the position of the maximum in the frequency differences, or 
751: from the value of the frequency at which these differences cross $\nu_{\rm 
752: asymp}$, is likely to show a stronger model dependence. As emphasized
753: earlier, at lower frequencies the functional form of the diagnostic tool 
754: is influenced by the deviation from degeneracy of the quantity $D_\delta$. 
755: Consequently, this tool does not capture, as well as it does at higher 
756: frequencies, the effect of the convective core on the oscillations, even 
757: though it is still influenced by it. Therefore, it is important to extend 
758: the present study to different models, varying quantities such as mass, 
759: chemical composition, and core overshoot, to determine the extent to which 
760: the quantities studied---particularly in the low-frequency domain---depend 
761: on the properties of the models.
762: 
763: One aspect of major importance that has been considered by most works 
764: involving diagnostic tools to study the inner regions of stars is the 
765: ability of these tools to determine the stellar age. The correlations 
766: between the quantities studied in \S\ref{DATA} and the age of the models 
767: are clear. Hence, in principle the former might be used to infer the 
768: evolutionary status of pulsating stars. 
769: In fact, this correlation is responsible for the one found by \cite{maz06} 
770: between their quantity $\theta$ and the central hydrogen abundance $X_c$, 
771: (cf. figure 7 of their paper). Their $\theta$ is proportional to the 
772: difference $(D_{13}-D_{02})$ averaged over several radial orders $n$, and 
773: through $X_c$ it reflects the change in the slope seen in the left panel 
774: of our Figure~\ref{comparison}. Thus, our analysis also provides the 
775: theoretical basis for the correlation found by \cite{maz06}.
776:  Nevertheless, it is important to investigate how these correlations
777:  depend on other stellar parameters, and to ascertain whether there
778:  are degeneracies in the age determination when the appropriate
779:  parameter space is explored.
780: Also, as mentioned in \S~\ref{models}, the models used to test our 
781: diagnostic tool did not include diffusion. The presence of diffusion will 
782: likely result in a smoother transition in $c^2$. We hope to tackle the 
783: consequences of including diffusion in future work.
784: 
785: Asteroseismic data with the precision necessary to test our diagnostic 
786: tool observationally are soon expected from the CoRoT satellite 
787: \citep{bag06}. The second scheduled 5-month run will include the star 
788: HD~49933, which exhibits solar-like oscillations \citep{mos05} and has 
789: approximately the required mass and age to reveal the expected signal. 
790: Future observations of the slightly more massive spectroscopic binary star 
791: 12~Bo{\"o}tis \citep{mig07} could also provide an interesting test of our 
792: predictions.
793: 
794: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
795: 
796: \acknowledgments 
797: 
798: The authors are very grateful to J{\o}rgen Christensen-Dalsgaard and 
799: Douglas Gough for helpful comments and suggestions. This work was 
800: supported in part by the EC's FP6, FCT and FEDER (POCI2010) through the 
801: HELAS international collaboration and through the project 
802: POCI/CTE-AST/57610/2004, by a Fulbright grant under the Mutual Educational 
803: Exchange Program, and by NCAR through the ECSA and HAO Visiting Scientist 
804: Programs. The National Center for Atmospheric Research is a federally 
805: funded research and development center sponsored by the U.S. National 
806: Science Foundation.
807: 
808: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
809: 
810: \appendix
811: 
812: \section{\label{APA}A.~Solution to the wave equation near the inner turning 
813: point}
814: 
815: In this appendix we derive an approximate solution to Eq.~(\ref{waveeq}), 
816: valid near the inner turning point $z_1$ at which $K^2=0$. We closely 
817: follow the derivation presented in \S4.8.5 of \citet{gough93}, though the 
818: dependent and independent variables in our Eq.~(\ref{waveeq}) differ from 
819: those in \citeauthor{gough93}'s Eq.~(4.8.5).
820: 
821: We start by defining new independent and dependent variables, $x$ and 
822: $\psi$ respectively
823: \begin{eqnarray}
824: x & = &\sgnk2\left[\sgnk2\frac{3}{2}
825: \int_{z_1}^z|K|\rmd z\right]^{\frac{2}{3}},
826: \label{defx} \\
827: \psi & = & |x|^{-\frac{1}{4}}|K|^{\frac{1}{2}}\Xi
828: \equiv s^{-\frac{1}{2}}\Xi.
829: \label{defpsi}
830: \end{eqnarray}
831: With these definitions, $x=0$ at the inner turning point. Substituting 
832: these into Eq.~(\ref{waveeq}) we find
833: \begin{equation}
834: \frac{\rmdd\psi}{\rmd x^2}+x\psi=-s^{\frac{3}{2}}
835: \frac{\rmdd s^{\frac{1}{2}}}{\rmd z^2}\psi,
836: \label{airy}
837: \end{equation}
838: where we keep the independent variable $z$ in the right hand side (RHS) to 
839: keep the equation in a condensed form.
840: 
841: To find an asymptotic solution to Eq.~(\ref{airy}) we regard the term on 
842: the RHS as a small perturbation to the Airy equation \citep{olver74}. We 
843: note that our original transformation of the independent variable $z={\rm 
844: ln}(r/r_{\rm ref})$, was motivated precisely by the need to keep the RHS 
845: bounded when $r$ tends to zero \citep{langer37}. In fact, a 
846: straightforward analysis of this term shows that it will tend to zero as 
847: $r$ tends to zero.
848: 
849: To leading order, we thus have that the solution to Eq.~(\ref{airy}) 
850: around the inner turning point can be obtained from
851: \begin{equation}
852: \psi\approx aA_{i}\left(-x\right)+bB_{i}\left(-x\right),
853: \end{equation}
854: where $A_i$ and $B_i$ are the solutions to the Airy equation, and $a$ and 
855: $b$ are constants. Since the solution must tend to zero as $x$ tends to 
856: $-\infty$, we must have $b=0$. Using the ascending series for $A_i$ 
857: \citep[e.g.][\S10.4, p.446]{abramowitz72}, we thus find for $|x|\ll 1$
858: \begin{equation}
859: \psi\approx a\left[A_i\left(0\right)-A_i^\prime\left(0\right)x 
860: +\mathcal{O}\left(x^3\right)\right],
861: \end{equation}
862: where 
863: \begin{eqnarray}
864: A_i\left(0\right)=\frac{1}{3^{\frac{2}{3}}\Gamma\left(\frac{2}{3}\right)} 
865: &\,\,\, {\rm and}\,\,\,& 
866: A_i^\prime\left(0\right)=
867: -\frac{1}{3^{\frac{1}{3}}\Gamma\left(\frac{1}{3}\right)}
868: \nonumber,
869: \end{eqnarray}
870: and $\Gamma$ is the Gamma function. 
871: Transforming back, we then find that
872: \begin{eqnarray}
873: \Xi \approx  a\, s^{\frac{1}{2}}\left[A_{\rm i}\left(0\right)- 
874: {A^\prime}_{\rm i}\left(0\right)x\right],
875: \end{eqnarray}
876: with $s$ defined by Eq.~(\ref{defpsi}).
877: 
878: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
879: 
880: \section{\label{APB}B.~Variational analysis}
881: 
882: In this appendix we derive the expression for the frequency perturbations 
883: resulting from the presence of a small convective core. Our starting point 
884: is Eq.~(\ref{vpp}), with $\delta I_1$ and $\delta I_2$ defined by 
885: Eqs.~(\ref{di1}) and (\ref{di2}). Introducing the new variables $x$ and 
886: $\psi$ into Eqs.~(\ref{di1}) and (\ref{di2}) and noting that
887: \begin{eqnarray}
888: \left(\frac{\rmd\psi}{\rmd x}\right)^2=
889: \frac{1}{2}\frac{\rmdd\psi^2}{\rmd x^2}-\psi\frac{\rmdd\psi}{\rmd x^2} 
890: & \,\,\, {\rm and} \,\,\, & 
891: \frac{\rmdd\psi}{\rmd x^2}\approx -x\psi,
892: \end{eqnarray}
893: we find, after some algebra,
894: \begin{eqnarray}
895: 2I_1\omega\delta\omega \approx 
896: \int_{X_0}^{X_R}\left[f_1\psi^2+f_2\frac{\rmd\psi^2}{\rmd x}
897: +f_3\frac{\rmdd \psi^2}{\rmd x^2}\right]\rmd x,
898: \label{dw1}
899: \end{eqnarray}
900: where $X_0$ and $X_R$ are the limits of the original integral in terms of 
901: the new variable $x$ and
902: \begin{eqnarray}
903: f_1&=&r^3q^2\left\{s^{-1}\left[x+\left(\frac{1}{q} \frac{\rmd q}{\rmd 
904: x}\right)^2\right]\delta\left(\gamma p\right)- \frac{\rmd}{\rmd 
905: x}\left[3\delta\left(\gamma p\right)-4\delta p\right]- 
906: \omega^2r^2s\delta\rho\right\},
907: \nonumber\\
908: f_2&=&\frac{1}{2}r^3s^{-1}\frac{\rmd q^2}{\rmd x}\delta\left(\gamma 
909: p\right),
910: \nonumber \\
911: f_3&=&\frac{1}{2}r^3s^{-1}q^2\delta\left(\gamma p\right),
912: \nonumber 
913: \end{eqnarray}
914: with $q=s^{1/2}r^{-3/2}{r_{\rm ref}}^{-1/2}\rho^{-1/2}c^{-1}$.
915: 
916: Next, we express Eq.~(\ref{dw1}) in terms of the perturbations 
917: $\delta(\gamma p)/\gamma p$ and $\delta c^2/c^2$. Using the relations
918: \begin{eqnarray}
919: \frac{\delta\rho}{\rho}=\frac{\delta\left(\gamma p\right)}{\gamma 
920: p}-\frac{\delta c^2}{c^2} & \,\,\, {\rm and}\,\,\, &\frac{\rmd\delta 
921: p}{\rmd x}=-\rho grs\left(\frac{\delta g}{g}+\frac{\delta\left(\gamma 
922: p\right)}{\gamma p}-\frac{\delta c^2}{c^2}\right),
923: \end{eqnarray}
924: we find
925: \begin{eqnarray}
926: 2I_1\omega\delta\omega \approx 
927: \int_{X_0}^{X_R}\left[\left(\delta\mF_1+\frac{\rmd}{\rmd 
928: x}\left(\delta\mF_0\right)\right)\psi^2+\delta\mF_2\frac{\rmd\psi^2}{\rmd 
929: x}+\delta\mF_3\frac{\rmdd \psi^2}{\rmd x^2}\right]\rmd x,
930: \label{dw3}
931: \end{eqnarray}
932: where
933: \begin{eqnarray}
934: \delta\mF_0&=&-\left(3r^3q^2\gamma p\right)\frac{\delta\left(\gamma 
935: p\right)}{\gamma p}, 
936: \nonumber \\
937: \delta\mF_1&=&\left\{r^3q^2\left[s^{-1}\left(x+\left(\frac{1}{q}\frac{\rmd 
938: q}{\rmd x}\right)^2\right)\gamma p-\rho 
939: rs\left(4g+\omega^2r\right)\right]+3\gamma 
940: p\frac{\rmd\left(r^3q^2\right)}{\rmd x}\right\}\frac{\delta\left(\gamma 
941: p\right)}{\gamma p} \nonumber \\ &&+\left[r^4q^2\rho 
942: s\left(4g+\omega^2r\right)\right]\frac{\delta c^2}{c^2} \nonumber \\ & 
943: &-\left(4r^4q^2\rho sg\right)\frac{\delta g}{g}, 
944: \nonumber \\
945: \delta\mF_2&=&\left(\frac{1}{2}r^3s^{-1}\frac{\rmd q^2}{\rmd x}\gamma 
946: p\right)\frac{\delta\left(\gamma p\right)}{\gamma p}, 
947: \nonumber \\
948: \delta\mF_3&=&\left(\frac{1}{2}r^3s^{-1}q^2\gamma 
949: p\right)\frac{\delta\left(\gamma p\right)}{\gamma p}.
950: \nonumber
951: \end{eqnarray}
952:  
953: The perturbation associated with the discontinuity at the edge of the 
954: convective core is rather localized. Thus, we consider an interval 
955: $[X_a,X_b]$ in which the perturbation differs from zero and assume it to 
956: be equal to zero outside that interval. 
957: By requiring both that the sound speed is everywhere the same in the 
958: perturbed and unperturbed models except near the discontinuity, and that 
959: both models are in hydrostatic equilibrium, the small difference in their 
960: density profiles extends beyond the region of the discontinuity, towards 
961: the center of the star. Nevertheless, because we are interested in 
962: isolating the frequency perturbation associated with the rapid variation 
963: at the edge of the core, we do not consider the perturbation arising from 
964: this additional difference between the two models.
965: 
966: Integrating by parts all terms of 
967: the integral on the RHS of Eq.~(\ref{dw3}) except for that involving 
968: $\delta\mF_0$, we find
969: \begin{eqnarray}
970: 2I_1\omega\delta\omega \approx 
971: -\int_{X_a}^{X_b}\left[\left(\frac{\rmd\delta\mF_1}{\rmd 
972: x}\right)\overline{\psi^2}+\left(-\frac{\rmd\delta\mF_0}{\rmd 
973: x}+\frac{\rmd\delta\mF_2}{\rmd x}\right)\psi^2+\frac{\rmd\delta\mF_3}{\rmd 
974: x}\frac{\rmd\psi^2}{\rmd x}\right]\rmd x,
975: \label{dw4}
976: \end{eqnarray}
977: where $\overline{\psi^2}=\int\psi^2\rmd x$. 
978: 
979: At the edge of the convective core, $\delta c^2 /c^2$ is discontinuous 
980: while all other perturbations in Eq.~(\ref{dw3}) are not. Hence, the main 
981: contribution to the frequency perturbations will come from the term 
982: involving the perturbation to the sound speed. Accordingly, we neglect all 
983: other terms and find
984: \begin{eqnarray}
985: 2I_1\omega\delta\omega \approx -\int_{X_a}^{X_b}\frac{\rmd}{\rmd 
986: x}\left[r^4q^2\rho s\left(4g+\omega^2r\right)\frac{\delta 
987: c^2}{c^2}\right]\overline{\psi^2}\rmd x.
988: \label{dw5}
989: \end{eqnarray}
990: 
991: We model the sharp variation $\delta c^2 /c^2$ at the edge of the 
992: convective core, $x_{\rm d}$, with a function of the type $\delta c^2 /c^2 
993: \propto [H(x-x_{\rmd})-1/2]$, where $H$ is the Heaviside Step Function and 
994: the proportionality constant is determined by the amplitude of the sharp 
995: variation. Under this assumption, and keeping in mind that the term 
996: multiplying $\delta c^2 /c^2$ in Eq.~(\ref{dw5}) is a slowly varying 
997: function of $x$, we have
998: \begin{equation}
999: \frac{\rmd}{\rmd x}\left[r^4q^2\rho s\left(4g+\omega^2r\right)\frac{\delta 
1000: c^2}{c^2}\right]\approx A_\delta \left[r^4q^2\rho 
1001: s\left(4g+\omega^2r\right)\right]\delta\left(x-x_\rmd\right),
1002: \label{deriv}
1003: \end{equation}
1004: where $A_\delta$ is the amplitude associated with the $\delta$-function. 
1005: We thus find, after some algebra,
1006: \begin{eqnarray}
1007: 2I_1\omega\delta\omega \approx -A_\delta 
1008: a^2A_i\left(0\right)^2\left[\frac{r\left(4g+\omega^2r\right)} {r_{\rm 
1009: ref}c^2|K|^2}\left(|x|x-\frac{A^\prime_{\rm i}(0)}{A_{\rm 
1010: i}(0)}|x|x^2+\frac{1}{3} \frac{A^\prime_{\rm i}(0)^2}{A_{\rm 
1011: i}(0)^2}|x|x^3\right)\right]_{x=x_{\rm d}}.
1012: \label{dw6}
1013: \end{eqnarray}
1014: In practice, if the approximations introduced above are reasonably 
1015: satisfied, $A_\delta$ is expected to be proportional to the amplitude of 
1016: the sharp variation $\delta c^2 /c^2$ at $x_\rmd$, with the 
1017: proportionality constant independent of the model considered. Indeed we 
1018: found this to be the case in all of our models except for those older than 
1019: $\sim$4~Gyr. For these models the discontinuity in $c^2$ is not as sharp 
1020: as for younger models. Even so, the functional form of the perturbations 
1021: was still found to be relatively well represented by Eq.~(\ref{dw6}) in 
1022: this case.
1023: 
1024: Finally, the constant $a$ can be expressed in terms of the amplitude of 
1025: the eigenfunction $\xi$ at the inner turning point. Combining the 
1026: definitions of $\psi$ and $\Xi$, we find
1027: \begin{eqnarray}
1028: \left[\psi\right]_{x=0}=\left[q^{-1}\xi\right]_{x=0}=a A_i\left(0\right).
1029: \end{eqnarray} 
1030: Substituting this expression for $a$ into Eq.~(\ref{dw6}), we finally 
1031: have
1032: \begin{eqnarray}
1033: \delta\omega \approx -\frac{A_\delta}{2\omega I_1}\left[\frac{|K|r^3\rho 
1034: c^2\xi^2}{|x|^{1/2}}\right]_{x=0}\left[\frac{r\left(4g+\omega^2r\right)} 
1035: {c^2|K|^2}\right]_{x=x_{\rm d}}\left[|x|x-\frac{A^\prime_{\rm 
1036: i}(0)}{A_{\rm i}(0)}|x|x^2+\frac{1}{3} \frac{A^\prime_{\rm i}(0)^2}{A_{\rm 
1037: i}(0)^2}|x|x^3\right]_{x=x_{\rm d}}.
1038: \label{final}
1039: \end{eqnarray}
1040: For practical reasons, when calculating the theoretical frequency 
1041: perturbations from Eq.~(\ref{final}), we use the quantities $\omega$, 
1042: $I_1$ and $\xi_{x=0}$ calculated for the model with the discontinuity. To 
1043: first order, this should have no impact on the theoretical signal.
1044: 
1045: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1046: 
1047: \begin{thebibliography}{}
1048: 
1049: \bibitem[Abramowitz \& Stegun(1972)]{abramowitz72} Abramowitz, M. \& 
1050: Stegun, I. A.\ 1972, Handbook of Mathematical Functions (9th printing, New 
1051: York: Dover)
1052: 
1053: \bibitem[Aerts et al.(2006)]{aer06} Aerts, C., et al.\ 2006, \apjl, 642, 
1054: L165
1055: 
1056: \bibitem[Baglin et al.(2006)]{bag06} Baglin, A., et al.\ 2006, ESA SP-624: 
1057: Proceedings of SOHO 18/GONG 2006/HELAS I, Beyond the spherical Sun, 18
1058: 
1059: \bibitem[Bahcall \& Pinsonneault(1992)]{bp92} Bahcall, J.~N., \& 
1060: Pinsonneault, M.~H.\ 1992, \apjl, 395, L119
1061: 
1062: \bibitem[Basu(1997)]{basu97} Basu, S.\ 1997, \mnras, 288, 572
1063: 
1064: \bibitem[Bedding et al.(2005)]{bed05} Bedding, T.~R., et al.\ 2005, \aap, 
1065: 432, L43
1066: 
1067: \bibitem[B{\"o}hm-Vitense(1958)]{bv58} B{\"o}hm-Vitense, E.\ 1958, 
1068: Zeitschrift fur Astrophysik, 46, 108
1069: 
1070: \bibitem[Bruntt et al.(2007)]{bru07} Bruntt, H., et al.\ 2007, \aap, 461, 
1071: 619
1072: 
1073: \bibitem[Buzasi et al.(2000)]{buz00} Buzasi, D., Catanzarite, J., Laher, 
1074: R., Conrow, T., Shupe, D., Gautier, T.~N., III, Kreidl, T., \& Everett, 
1075: D.\ 2000, \apjl, 532, L133
1076: 
1077: \bibitem[Cameron et al.(2007)]{cam07} Cameron, C., et al.\ 2007, 
1078: Communications in Asteroseismology, in press (astro-ph/0610822)
1079: 
1080: \bibitem[Chandrasekhar(1964)]{chandrasekhar64} Chandrasekhar, S.\ 1964, 
1081: \apj, 139, 664
1082: 
1083: \bibitem[Christensen-Dalsgaard(1982)]{jcd82} Christensen-Dalsgaard, J.\ 
1084: 1982, \mnras, 199, 735
1085: 
1086: \bibitem[Christensen-Dalsgaard et al.(2007)]{jcd07} Christensen-Dalsgaard, 
1087: J., Arentoft, T., Brown, T.~M., Gilliland, R.~L., Kjeldsen, H., Borucki, 
1088: W.~J., \& Koch, D.\ 2007, Communications in Asteroseismology, in press 
1089: (astro-ph/0701323)
1090: 
1091: \bibitem[Cox(1980)]{cox80} Cox, J.~P.\ 1980, Theory of stellar pulsation 
1092: (Princeton, NJ: Princeton University Press)
1093: 
1094: \bibitem[Dappen et al.(1988)]{dap88} Dappen, W., Dziembowski, W.~A., \& 
1095: Sienkiewicz, R.\ 1988, IAU Symp.~123: Advances in Helio- and 
1096: Asteroseismology, 123, 233
1097: 
1098: \bibitem[Eggleton et al.(1973)]{eff73} Eggleton, P.~P., Faulkner, J., \& 
1099: Flannery, B.~P.\ 1973, \aap, 23, 325
1100: 
1101: \bibitem[Gough(1993)]{gough93} Gough, D.~O.\ 1993, in Astrophysical fluid 
1102: dynamics, eds. J-P. Zahn \& J. Zinn-Justin (Amsterdam: Elsevier), 399
1103: 
1104: \bibitem[Houdek \& Gough(2007)]{houdek07} Houdek, G. \& Gough, D.O.\ 2007, 
1105: \mnras, 375, 861 
1106: 
1107: \bibitem[Iglesias \& Rogers(1996)]{ir96} Iglesias, C.~A., \& Rogers, 
1108: F.~J.\ 1996, \apj, 464, 943
1109: 
1110: \bibitem[Langer(1937)]{langer37} Langer, R.E.\ 1937, Phys.~Rev., 51, 
1111: 669
1112: 
1113: \bibitem[Matthews et al.(2004)]{mat04} Matthews, J.~M., Kusching, R., 
1114: Guenther, D.~B., Walker, G.~A.~H., Moffat, A.~F.~J., Rucinski, S.~M., 
1115: Sasselov, D., \& Weiss, W.~W.\ 2004, \nat, 430, 51
1116: 
1117: \bibitem[Mazumdar et al.(2006)]{maz06} Mazumdar, A., Basu, S., Collier, 
1118: B.~L., \& Demarque, P.\ 2006, \mnras, 372, 949
1119: 
1120: \bibitem[Miglio et al.(2007)]{mig07} Miglio, A., Montalban, J., \& 
1121: Maceroni, C.\ 2007, \mnras, 377, 373
1122: 
1123: \bibitem[Monteiro et al.(1994)]{monteiro94} Monteiro, M.~J.~P.~F.~G., 
1124: Christensen-Dalsgaard, J., \& Thompson, M.~J.\ 1994, \aap, 283, 247
1125: 
1126: \bibitem[Monteiro et al.(2000)]{monteiro00} Monteiro, M.~J.~P.~F.~G., 
1127: Christensen-Dalsgaard, J., \& Thompson, M.~J.\ 2000, \mnras, 316, 165
1128: 
1129: \bibitem[Monteiro \& Thompson(1998)]{monteiro98} Monteiro, M.~J.~P.~F.~G., 
1130: \& Thompson, M.~J.\ 1998, IAU Symp.~185: New Eyes to See Inside the Sun 
1131: and Stars, 185, 317
1132: 
1133: \bibitem[Monteiro \& Thompson(2005)]{monteiro05} Monteiro, M.~J.~P.~F.~G., 
1134: \& Thompson, M.~J.\ 2005, \mnras, 361, 1187
1135: 
1136: \bibitem[Mosser et al.(2005)]{mos05} Mosser, B., et al.\ 2005, \aap, 431, 
1137: L13
1138: 
1139: \bibitem[Olver(1974)]{olver74} Olver, F.~W.~J.\ 1974, Asymptotics and 
1140: special functions (New York: Academic Press)
1141: 
1142: \bibitem[Randall et al.(2005)]{ran05} Randall, S.~K., et al.\ 2005, \apj, 
1143: 633, 460
1144: 
1145: \bibitem[Rowe et al.(2006)]{row07} Rowe, J.~F., et al.\ 2007, 
1146: Communications in Asteroseismology, in press (astro-ph/0610640)
1147: 
1148: \bibitem[Roxburgh \& Vorontsov(2000a)]{roxburgh00a} Roxburgh, I.~W., \& 
1149: Vorontsov, S.~V.\ 2000a, \mnras, 317, 141
1150: 
1151: \bibitem[Roxburgh \& Vorontsov(2000b)]{roxburgh00b} Roxburgh, I.~W., \& 
1152: Vorontsov, S.~V.\ 2000b, \mnras, 317, 151
1153: 
1154: \bibitem[Roxburgh \& Vorontsov(2001)]{roxburgh01} Roxburgh, I.~W., \& 
1155: Vorontsov, S.~V.\ 2001, \mnras, 322, 85
1156: 
1157: \bibitem[Roxburgh \& Vorontsov(2003)]{roxburgh03} Roxburgh, I.~W., \& 
1158: Vorontsov, S.~V.\ 2003, \aap, 411, 215
1159: 
1160: \bibitem[Saio et al.(2006)]{sai06} Saio, H., et al.\ 2006, \apj, 650, 1111
1161: 
1162: \bibitem[Takata(2006)]{takata06} Takata, M., 2006, \pasj, 58, 893
1163: 
1164: \bibitem[Walker et al.(2003)]{wal03} Walker, G., et al.\ 2003, \pasp, 115, 
1165: 1023
1166: 
1167: \end{thebibliography}
1168: 
1169: \end{document}
1170: