astro-ph0603613/ms.tex
1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass[preprint2,amsmath,amssymb]{aastex}
3: %\documentclass[preprint,amsmath,amssymb]{aastex}
4: 
5: 
6: \usepackage{graphicx}
7: \usepackage{latexsym}
8: 
9: \begin{document}
10: 
11: 
12: % -----> TITLE PAGE 
13: 
14: \title{Gravitational cooling of self-gravitating Bose-Condensates}
15: \author{F. Siddhartha Guzm\'an}
16: \affil{Instituto de F\'{\i}sica y Matem\'{a}ticas, 
17: Universidad Michoacana de San Nicol\'as de Hidalgo. Edificio C-3, Cd. 
18: Universitaria. C. P. 58040 Morelia, Michoac\'{a}n, M\'{e}xico.}
19: 
20: \and
21: \author{L. Arturo Ure\~{n}a-L\'{o}pez}
22: \affil{Instituto de F\'isica de la Universidad de Guanajuato, 
23: A.P. 150, C.P. 37150, Le\'on, Guanajuato, M\'exico.}
24: 
25: % -----> ABSTRACT
26: 
27: \begin{abstract}
28: Equilibrium configurations for a self-gravitating scalar field with 
29: self-interaction are constructed. The corresponding
30: Schr\"odinger-Poisson (SP) system is solved using finite differences
31: assuming spherical symmetry. It is shown that equilibrium
32: configurations of the SP system are late-time attractor solutions for
33: initially quite arbitrary density profiles, which relax and virialize
34: through the emission of scalar field bursts; a process dubbed
35: gravitational cooling. Among other potential applications, these
36: results indicate that scalar field dark matter models (in its
37: different flavors)  tolerate the introduction of a self-interaction
38: term in the SP equations. This study can be useful in exploring models
39: in which dark matter in galaxies is not point-like.
40: \end{abstract}
41: 
42: \keywords{galaxies: evolution --
43:           galaxies: formation --
44:           galaxies: halos --
45:           dark matter}
46: 
47: 
48: % -----> ARTICLE
49: 
50: % -----> INTRODUCTION
51: \section{Introduction}
52: \label{sec:intro}
53: 
54: One of the most fundamental problems in modern Cosmology is the nature
55: of dark matter. This problem itself has provoked an intense
56: theoretical search for realistic candidates able to drive the
57: formation of the cosmological structure we observe today \cite{sahni2004}. 
58: Among others, there are some candidates that have for long been considered
59: as exotic; a paradigmatic example is the case of scalar fields, which
60: are the simplest particles we can think of \cite{matoslurena2004, 
61: sahni2000}. However, no fundamental scalar field has ever been detected. 
62: 
63: Nevertheless, theoretical research on scalar candidates is a very
64: interesting field in Cosmology nowadays, and many ideas have been
65: proposed. One principal idea is that scalar fields can form Bose
66: condensates, and then are able to form
67: structure \cite{hu, arbeyetal, matoslurena2001}. The quest for the kind
68: of objects that are formed by scalar fields has been investigated in
69: detail using numerical tools. The relativistic regime has been studied
70: in \cite{seidel94, balakrishna98, alcubierre2002, alcubierre2003}, for
71: both complex and real scalar fields.
72: 
73: Scalar fields are an example in which dark matter is not point-like,
74: hence no conclusions can be extracted from the widely studied N-body
75: simulations for dark matter particles. Thus, we need to understand
76: their properties in order to decide whether such properties could be
77: useful to explain what we see in galaxies. On of the most attractive 
78: features of the scalar field model is that the construction of equilibrium 
79: configurations demands a smooth density profile at the origin, which is an 
80: advantage over the point-like dark matter particle models which have shown 
81: to be cuspy.
82: 
83: In general, complex scalar fields can form stable equilibrium configurations
84: called \textit{boson stars}, that are globally regular and whose
85: energy density is time-independent, for a recent review
86: see \cite{schunck}. Real scalar fields also have equilibrium
87: configurations, that were discovered in \cite{seidel91}, and are called
88: \textit{oscillatons}. The latter are also globally regular, but are
89: fully time-dependent. As for their stability, it seems to be 
90: quite robust as far as numerical evolutions is
91: concerned \cite{seidel91,alcubierre2002,alcubierre2003}, but it may be
92: that they are only long-lived \cite{d-page}. 
93: 
94: An important part of any scalar field model of dark matter is
95:   the \textit{scalar potential} $V(\phi)$, which encodes in itself the
96:   self-interactions of the scalar field other than gravitational. The
97:   most popular scalar potential, and also the most studied in
98:   Cosmology, is the quadratic one, $V(\phi)=(m^2/2) \phi^2$ (or
99:   $V(|\phi|)=m^2 |\phi|^2$ if the scalar field is complex), where the
100:   parameter $m$ is identified as the mass of the boson particle.
101: 
102: On the other hand, the inverse of the mass, i.e. the
103:   Compton length of the boson $\lambda_C = m^{-1}$ (in units in which
104:   $\hbar =c=1$), becomes the natural unit of distance of scalar
105:   configurations. Actually, the most compact scalar objects have a
106:   radius of a few Compton lengths. In addition, the boson mass also
107:   establishes the mass scale of the compact object, which is of the
108:   order $\sim m^2_{Pl}/m$, where $m_{Pl}$ is the Planck
109:   mass. Thus, a light scalar mass may provide of very massive
110:   objects\cite{seidel91,harri1,alcubierre2003,schunck,fsglau2004}.
111:   However, scalar potentials more complicated than the quadratic one
112:   can provide of extra parameters which can add new features to
113:   oscillatons and boson stars.
114: 
115: One simple instance of such more general scalar potential is
116:   that containing an extra quartic coupling in the form $V(\phi)=(m^2/2)
117:   \phi^2+\lambda_{int} \phi^4/4$, see
118:   \cite{wasserman,balakrishna98,arbeyetal} and references therein. In
119:   this case we speak of a self-interacting scalar field\footnote{The
120:   quadratic potential is also called \emph{free} potential, as the
121:   scalar field represents in this case a free relativistic particle of
122:   mass $m$ with total energy $E^2 = p^2 + m^2$.}.
123: 
124: The influence a quartic self-interaction can have in the
125:   dynamics of a cosmological \textit{complex} scalar field has been
126:   widely studied in the literature, for recent developments
127:   see\cite{schunck,arbeyetal}. However, there are also some changes
128:   with respect to the free case. The self-interaction term can become
129:   important in situations in which the scalar field has large amplitude, 
130: which
131:   may happen typically in the early universe, and in the interior of
132:   compact scalar objects. For the purposes of this paper, it should be
133:   mentioned that a quartic term makes, for a given scalar field
134:   strength, a boson star larger and more massive than in the free
135:   case\cite{wasserman,balakrishna98,guzman2006}.
136: 
137: On the other hand \textit{oscillatons} with self-interacting terms in
138:   the scalar potential have not been studied yet. The reason for this
139:   may be an intrinsic difficulty in dealing with time-dependent
140:   relativistic configurations.
141: 
142: Nevertheless, a curious point is that the Newtonian regime is the same
143: for boson stars and oscillatons. The relativistic EKG equations in
144: this regime are equivalent to the so-called Schr\"odinger-Poisson (SP)
145: system, see \cite{seidel94,harri1,fsglau2003,fsglau2004} and
146: references therein. 
147: 
148: For a free scalar potential, the Newtonian equilibrium configurations can
149: be as large as a galactic halo if the mass of the boson is very light
150: \cite{arbeyetal,fsglau2003}. That the Newtonian regime is the
151:   astrophysically interesting one can be seen if one calculates the
152:   Compton length associated to a (very light) boson mass of order
153:   $10^{-23}$ eV; this is $\lambda_C \sim 1$
154:   pc\cite{arbeyetal,sahni2000,matoslurena2001,hu}. From the point
155:   of view of particle physics, this boson mass is extremely
156:   small. Therefore, such a light boson will form \emph{relativistic}
157:   objects which are small compared to typical galactic
158:   scales. However, in the \emph{Newtonian} regime, gravity is weaker
159:   and allows the formation of much larger scalar objects.
160: 
161: Within the Newtonian regime, there are also equilibrium solutions for 
162: different number of nodes of the wave function; solutions with nodes are 
163: called excited Newtonian boson stars and are classified according to the 
164: number of nodes the wave function has. Zero-node states are called 
165: \emph{ground} states, and configurations with nodes are generically
166: called \emph{excited states}. Such excited states were also considered in 
167: the past as candidates for dark matter halos in galaxies \cite{koreanos}. 
168: They have the nice property of providing nearly flat rotation curves if 
169: the number of nodes is sufficiently large, see the example shown 
170: in \cite{arbeyetal}. Unfortunately, those excited boson stars are not 
171: stable, and they decay into a ground state in a time much smaller than the 
172: actual age of the universe \cite{fsglau2003,fsglau2004}; the attractor 
173: behavior of the ground state configurations is shown below in this paper 
174: for initially excited state equilibrium configurations. This fact 
175: rules out the possibility of excited boson stars as dark matter halos; 
176: nevertheless, they could still play the role of transient states in the 
177: evolution of scalar field configurations. 
178: 
179: It is interesting to note that the SP system has been widely studied
180: in other fields out of cosmology, see the interesting works
181: in \cite{dale-phd, dale-choi, giovan2001}. Also, they have been proposed to
182: ameliorate the quasar's redshift problem in \cite{svid2004}.
183: 
184: In this paper we want to study the case of scalar configurations with
185: a self-interaction term in the Newtonian limit, that is, we shall
186: study the SP system including a quartic self-interaction term. In
187: doing this, we will learn about the properties that both a boson star
188: and an oscillaton have in the weak field limit.
189: 
190: Scalar field configurations with a quartic
191:   self-interaction have been proposed before to be the dark
192:   matter in galaxies, see \cite{arbeyetal}. But, it is
193:   necessary to understand the dynamical properties of such
194:   configurations. In this sense, this paper extends the research
195:   already done in\cite{fsglau2004} for the (free field) SP system.
196: 
197: Assuming spherical symmetry, the Schr\"odinger-Poisson (SP) 
198: system of equations to be solved reads
199: \label{sp}
200: \begin{eqnarray}
201: i\partial_\tau \psi &=& -\frac{1}{2x} \partial^2_x (x\psi) + U \psi    
202: + \Lambda |\psi|^2\psi
203: \label{schroedinger}\\
204: \partial^2_x (xU) &=& x \psi \psi^\ast . \label{poisson}
205: \end{eqnarray}
206: We have used the dimensionless variables $\tau = mt$, 
207: $x=mr$, where $t$ and $r$ are the physical time and radial coordinates, 
208: respectively, and $m$ is the mass of the boson.
209: 
210: The wave function $\psi$ is coupled to its own gravitational potential 
211: $U$, and the constant $\Lambda$ represents the $s$-wave scattering
212: length in the Gross-Pitaevskii approximation for Bose
213: condensates \cite{gross,pitaevskii}. 
214: 
215: It is known that the SP set of equations without self-interaction has
216: a fancy scaling property \citep{fsglau2004} that simplifies the
217: numerical research. It is possible to establish a similar scaling
218: property when a self-interaction term is included. That is,
219: Eqs.~(\ref{sp}) are invariant under the transformation
220: \begin{equation}
221: \{\tau,x,U,\psi,\Lambda \}
222:  \rightarrow
223: \{\lambda^{-2}\hat{\tau},
224: \lambda^{-1}\hat{x},
225: \lambda^{2}\hat{U},
226: \lambda^{2}\hat{\psi},
227: \lambda^{-2}\hat{\Lambda} \} \label{eq:scaling}
228: \end{equation}
229: 
230: \noindent with $\lambda$ an arbitrary parameter. Unfortunately, there
231: is a main difference with respect to the non-interacting case. In the
232: latter, the solution of the SP assuming $|\psi(0)|=1$ generates the
233: whole and unique branch of possible equilibrium solutions.
234: 
235: For the new SP equations~(\ref{sp}), it is necessary to find first all
236: possible solutions that belong to a \emph{single} value of the self-interaction
237: coefficient $\Lambda$. If we now scale the latter using a certain
238: $\lambda$, we can generate a full \emph{new} branch of solutions for
239: $\hat{\Lambda}$ out of the one belonging to the old $\Lambda$. Though
240: not as easy as in the free case, the scaling
241: transformation~(\ref{eq:scaling}) will still help us to understand the
242: evolution of a scalar field configuration under an arbitrary value of
243: $\Lambda$.
244: 
245: There is also a complete set of physical quantities that are well
246: defined in the Newtonian limit. They obey  the following scaling 
247: transformation
248: 
249: \begin{equation}
250: \{\rho,N,K,W,I \} \rightarrow
251: \{\lambda^{4}\hat{\rho},
252: \lambda \hat{N},
253: \lambda^{3} \hat{K},
254: \lambda^{3} \hat{W},
255: \lambda^{3} \hat{I} \} \, , \label{props}
256: \end{equation}
257: 
258: \noindent where $\rho$ is the density of probability, $N$ is the integral
259: of $\rho$ over the whole space, $K$ and $W$ are the expectation values
260: of the kinetic and gravitational energies, and $I$ is the expectation
261: value of the self-interaction energy. The above quantities will help
262: us to follow the evolution of an arbitrary scalar field configuration and 
263: establish criteria about virialization and equilibrium,
264: 
265: This paper is organized as follows. In Sec.~\ref{sec:ivp}, we present
266: the solution to the initial value problem that will be the equilibrium
267: configurations of the SP system. In Sec.~\ref{sec:numerics}, we
268: describe the numerical methods and the boundary conditions used for
269: the time evolution of the SP system. After that, in 
270: Sec.~\ref{sec:cooling}, we describe what we mean by virialization of 
271: non-equilibrium configurations and the late-time attractor behavior of the 
272: sequence of equilibrium configurations. Finally, in 
273: Sec.~\ref{sec:conclusions}, we draw some comments and conclusions.
274: 
275: % =========== SECTION
276: \section{Equilibrium configurations}
277: \label{sec:ivp}
278: 
279: For equilibrium configurations, we assume that the wave function is of the
280: form $\psi(\tau,x) = e^{i\omega \tau} \phi(x)$, where $\omega$ is a
281: free parameter. This assumption implies that the density of
282: probability $\rho$ and the gravitational potential $U$ are
283: time-independent, whereas the wave function evolves harmonically in
284: time. Eqs.~(\ref{schroedinger}) and~(\ref{poisson}) become
285: \begin{eqnarray}
286: \partial^{2}_{x}(x\phi)&=& 2 x (U-\omega) + 2\Lambda
287: |\phi|^2\phi \, , \label{sch-spherical} \\
288: \partial^{2}_{x}(xU)&=& x\phi^2 \, . \label{poi-spherical}
289: \end{eqnarray}
290: 
291: \noindent The above system has to be solved under the condition of
292: regularity at the origin $\phi(0)=\partial_x \phi(0)=0$, and isolation
293: $\phi(x\rightarrow \infty) = 0$; we will also demand that
294: $U(\infty)=0$. These boundary conditions determine in a unique manner
295: the values of $U(0)$ and $\omega$, which are the only free parameters
296: of the solution.
297: 
298: We made a numerical routine that solves Eqs.~(\ref{schroedinger})
299: and~(\ref{poisson}) using a shooting procedure that tunes the values
300: of $\omega$ and $U(0)$ for a given central value $\phi(0)$. In
301: Fig.~\ref{fig:plotone}, we show the profiles of the initial wave function
302: for some positive and negative values of the self-interaction
303: coefficient $\Lambda$. Also shown in there are the branches of
304: equilibrium configurations for four different values of 
305: $\Lambda$. As expected, for $\Lambda \ge 0$ there is nothing like a 
306: maximum indicating an unstable branch, which is always present (and 
307: typical) in relativistic boson configurations
308: \cite{wasserman,balakrishna98,guzman2004}. 
309: 
310: \begin{figure*}
311: \includegraphics[width=8cm]{f1a.eps}
312: \includegraphics[width=8cm]{f1b.eps}
313: \caption{\label{fig:plotone} (Left) Profiles $\phi(x)$ of equilibrium 
314: solutions of the SP for different values of  the self-interaction 
315: coefficient $\Lambda$. As expected, the larger the coefficient $\Lambda$ 
316: the more massive the equilibrium configuration is. (Right) 
317: Sequences of equilibrium configurations for different values 
318: of $\Lambda$. Each point in the curves represents a solution of the 
319: initial value problem of a total number of particles $N$ and $95$\%
320: radius $r_{95}$ (the radius inside which $95$\% of the total mass is
321: contained in). In this plot it is manifest that the bigger the $\Lambda$ 
322: the less compact a configuration is. The filled circles indicate two 
323: configurations we use as tests for our numerical implementation.} 
324: \end{figure*}
325: 
326: % ========>
327: % NUMERICS
328: % ==========================================================================================>
329: \section{Numerical evolution}
330: \label{sec:numerics}
331: 
332: In order to numerically evolve the SP system, we make a discretization
333: of space and time, and approximate all derivatives using second
334: order accurate finite differencing. The SP system is evolved one time
335: step $\Delta \tau$ using Eq.~(\ref{schroedinger}) to obtain a new wave
336: function, and then we solve Eq.~(\ref{poisson}) to find the
337: corresponding (new) gravitational potential. 
338: 
339: We use an explicit time integrator to solve Eq.~(\ref{schroedinger}), as
340: opposed to the fully implicit method used and described
341: in\cite{fsglau2004}, where the problem of evolution was reduced to a
342: linear system of equations. In the present case, due to the non-linear
343: term in the Hamiltonian of Eq.~(\ref{schroedinger}), the reduction to
344: a linear system of equations is not that simple. 
345: 
346: In any case, independently of the numerical method used to integrate
347: in time, we demand the evolution operator to preserve the number of
348: particles for an \textit{equilibrium configuration} $N=\int |\psi| x^2
349: dx$\footnote{In full units, the total mass is given by
350:   $M=(m^2_\textrm{Pl}/m) N$, where $m_\textrm{Pl}$ is the Planck
351:   mass.}. With a modified version of the iterative Crank-Nicholson
352: evolution method \cite{teukolsky00a}, we were able to reproduce the
353: results found with the implicit method for the linear case, and
354: confirm that the evolution was mass-preserving for a wide range of
355: values of $\Lambda$.
356: 
357: %\begin{figure}[htp]
358: \begin{figure*}
359: \includegraphics[width=8cm]{f2a.eps}
360: \includegraphics[width=8.5cm]{f2b.eps}
361: \includegraphics[width=8cm]{f2c.eps}
362: \includegraphics[width=8.5cm]{f2d.eps}
363: %\includegraphics[width=8cm]{f2e.eps}
364: %\includegraphics[width=8.5cm]{f2f.eps}
365: \caption{\label{fig:plottwo} Evolution of equilibrium configurations 
366: with different values of $\Lambda$. \textit{Figures on the left}. For
367: each case, the central value of $\rho(\tau,0)$ is shown for two
368: resolutions: $\Delta x = 0.08, 0.16$ with the boundary located at
369: $x_B=50$, and a third case with the boundaries located at $x_b=100$
370: and resolution $\Delta =0.08$. The curve corresponding to the coarse
371: resolution has been scaled as $(\rho_{0.16}(0) - 1)/4 + 1$, and the
372: fact that it lies upon the value found for $\rho_{0.08}$ indicates
373: second order convergence of the solutions. Hence, we are running the
374: simulations in the convergence regime of our discretization. The fact
375: that the density found for $\Delta x =0.08$ with the boundaries at two
376: different locations lies upon each other, indicates that the results
377: are independent of the location of the boundary. \textit{Figures on
378:   the right} We show the evolution of the number of particles, and the
379: value of the virialization quantity $2K+3I+W$, which in the continuum
380: case should be zero. We see that its values are four times smaller if
381: the spatial resolution is doubled. This clearly indicates second order
382: convergence of this quantity to zero, which in turn means that we can
383: recover what is expected in the continuum limit.} 
384: \end{figure*}
385: 
386: 
387: % ---> Boundary Conditions
388: {\it Boundary conditions}. At every time step, Eq.~(\ref{poisson}) is
389: integrated inwards, and thus we applied the following boundary
390: condition at the two outermost points of the numerical grid
391: \begin{eqnarray}
392: U(x_{n-1}) &=& -N(x_{n-1})/x_{n-1}  \, , \label{bcU1}\\
393: U(x_n) &=& -N(x_n)/x_n  \, .\label{bcU}
394: \end{eqnarray}
395: 
396: \noindent where $n$ labels the outermost point of the numerical
397: domain. The whole profile of $U(x)$ is then found using a sixth order
398: accurate Numerov algorithm \cite{cphysics}.
399: 
400: As it is discussed in \cite{fsglau2004}, the boundary
401: conditions (\ref{bcU1}) and (\ref{bcU}) are equivalent to impose 
402: the condition $|\psi(\tau, x_n)| \rightarrow 0$ on the wave
403: function; hence, we are forcing the system to remain in the
404: computational domain. Because we want to evolve systems out of
405: equilibrium and allow the flow of particles out of the numerical
406: domain, we implemented a \textit{sponge} over the outermost points of
407: the grid, which consists in adding an imaginary potential $V_j(x)$ to
408: the Schr\"odinger equation (see \cite{fsglau2004,israeli}). The
409: expression we use for the sponge profile is
410: \begin{equation}
411: V_j = -\frac{i}{2} V_0 \left\{ 2 + \tanh \left[(x_j-x_c)/\delta 
412: \right] - \tanh \left( x_c/\delta \right) \right\} \, , \label{imagpot}
413: \end{equation}
414: 
415: \noindent which is a smooth version of a step function with amplitude 
416: $V_0$, centered in $x_c$ and width $\delta$. the minus sign warranties
417: the decay of the number of particles at the outer parts of our
418: integration domain, that is, the imaginary potential behaves as a sink
419: of particles. It is also worth noticing that no term related to the
420: self--interaction appears in this conservation equation since we are
421: assuming $\Lambda$ is real.
422: 
423: {\it Tests.} The obstacles our code must sort out are: i) the
424: evolution of equilibrium configurations, in which the wave function
425: oscillates with the definite frequency $\omega$ found in
426: Sec.~\ref{sec:ivp}, whereas the density of probability and the
427: gravitational potential should remain time-independent; ii) the
428: convergence of physical properties of the system~(\ref{props}).
429: 
430: These two tests are shown at once in Fig.~\ref{fig:plottwo}. On the
431: left hand side, we show the convergence of the central density and the
432: independence of the evolution on the location of the numerical
433: boundary. We know that in the continuum limit, these configurations
434: should evolve keeping $\rho(\tau,0)=1$ for all times. It is shown that
435: our approximation through finite differences converges to that value
436: in the continuum limit. 
437: 
438: The central density shows oscillations whose amplitude converge to zero as the
439: numerical grid is refined. We relate these oscillations to the
440: response of the system to the intrinsic perturbation induced by the
441: truncation error of the finite differencing approximation\footnote{In
442:   \cite{fsglau2004}, we compared this type of oscillations for $\Lambda
443:   =0$ with a linear perturbation analysis and found excellent
444:   coincidence with the Fourier analysis of the numerical evolutions. In the
445:   present case, we deal with a non-linear Schr\"odinger equation, and
446:   such analysis would say little about the cases with $\Lambda \ne 0$.
447:   Nevertheless, we already have evidence that the truncation error could be
448:   responsible for this very regular oscillations.}.
449: On the right hand side, we show how the evolution operator preserves
450: the number of particles all along the simulation. 
451: In a few words,
452: what we show in Figure \ref{fig:plottwo} is how the evolution
453: method, the finite differencing and the boundary conditions work
454: properly all together.
455: 
456: Besides the expected results found for the cases marked with filled 
457: circles in Fig.~\ref{fig:plotone}, we also tried to evolve configurations 
458: located to the left of the maximum for $\Lambda=-0.2$. Such configurations 
459: are very compact, and their evolution resulted in a quick collapse
460: and in the divergence of the central density $\rho(\tau,0)$. In
461: principle, this would be the expected behavior for an unstable
462: equilibrium configuration in the \textit{relativistic} regime. However, we do not account with a clear definition of a collapsed 
463: object in terms of trapped surfaces of horizon formation during the 
464: evolution, so that it is not possible to draw accurate statements
465: within the non-relativistic SP formalism for $\Lambda < 0$. One 
466: possibility would be to continue the evolution of the system with a 
467: relativistic code after certain compactness is achieved, however such 
468: analysis is beyond the scope of this paper. 
469: 
470: % -----> RESULTS
471: \section{Gravitational cooling and virialization}
472: \label{sec:cooling}
473: 
474: The virialization and stabilization of physical configurations 
475: evolving according to Eqs.~(\ref{schroedinger}) and~(\ref{poisson})
476: with $\Lambda=0$ was first proposed in\cite{seidel94}, and revised in
477: detail in \cite{fsglau2004}. The mechanism for relaxation consists in
478: the ejection of scalar field bursts, and the tendency of the system to
479: settle down onto equilibrium configurations. This process was dubbed
480: \textit{gravitational cooling}. 
481: 
482: The criterion to determine whether or not a system was near an
483: equilibrium state involves monitoring the virial theorem relation
484: \begin{equation}
485: 2K + 3I + W = 0 \, , \label{eq:virial}
486: \end{equation}
487: 
488: \noindent where $K,I,W$ are the expectation values of the kinetic,
489: self-interaction, and gravitational potential energies, respectively,
490: for the SP system (see for instance \cite{wang2001}). In
491: Fig.~\ref{fig:plottwo}, we also show the oscillatory behavior of
492: the quantity $2K + 3I + W$, whose amplitude converges to zero in the
493: continuum limit.
494: 
495: \begin{figure*}
496: \includegraphics[width=8cm]{f3a.eps}
497: \includegraphics[width=8cm]{f3b.eps}
498: \includegraphics[width=8cm]{f3c.eps}
499: \includegraphics[width=8cm]{f3d.eps}
500: %\includegraphics[width=8cm]{f3e.eps}
501: %\includegraphics[width=8cm]{f3f.eps}
502: \caption{\label{fig:plotthree} \textit{Top panel} We show the evolution of 
503: the 
504: initial profile $\psi(0,x)=\cos^2(x) e^{-x^2/(3.959)^2}$ with
505: $\Lambda=0$. The system starts with a high kinetic energy, and after a
506: relaxation period the central density $\rho(\tau,0)$ stabilizes around
507: one, whereas the virial relation $2K + 3I + W$ oscillates around
508: zero. The assymptotic behaviors of the the number of particles $N$ and
509: total energy are shown on the right plot. The initial mass 
510: is $M_i=3.617337 \, (m^2_{Pl}/m)$, and during the process around
511: $42\%$ of it was ejected very quickly. \textit{Bottom panel} This is
512: the case for $\Lambda=0.2$, with an initial wave function profile
513: $\psi(0,x) = \cos^2(x) e^{-x^2/(4.085)^2}$. As before, the system
514: also relaxes and stabilizes around an equilibrium configuration. This
515: time, the initial mass is $M_i=3.98 \, (m^2_{Pl}/m)$, and the ejected
516: mass is $\sim$39\%. In both cases the final configurations are nearly 
517: those shown in Figure~\ref{fig:plottwo}.} 
518: \end{figure*}
519: 
520: In Fig.~\ref{fig:plotthree}, we show the virialization process for two
521: different values of $\Lambda$. The initial profile is of the form
522: $\psi(0,x) = \cos^2(x) e^{-x^2/\sigma^2}$, with $\sigma$ a free parameter. The
523: reason why we choose such initial profile is that it provides a high
524: initial kinetic energy, and the process of virialization (if
525: any) will be evident while calculating Eq.~(\ref{eq:virial}); other 
526: initial profiles of garden variety have been evolved, however the 
527: overwarmed initial configurations shown here present in a better way the 
528: relaxation process. The case $\Lambda=0$ is as simple as those found in 
529: the past \cite{seidel94,fsglau2003,fsglau2004}, because it is possible to
530: choose any profile with the confidence that the configuration will
531: approach a rescaled equilibrium configuration (as shown in
532: \cite{fsglau2004}), but we want to be sure that the same happens for
533: $\Lambda \neq 0$.
534: 
535: Our expectations were fulfilled because of the manifest results in
536: Fig.~\ref{fig:plotthree}. All of the studied cases show a relaxation
537: process towards one of the equilibrium configurations shown in
538: Fig.~\ref{fig:plotone}. With these
539: results, we confirm that the equilibrium configurations in
540: Fig.~\ref{fig:plotone}, which are solutions of Eqs.~(\ref{sch-spherical})
541: and~(\ref{poi-spherical}), are indeed late-time attractor solutions.
542: 
543: To finish with, we show some numerical runs of excited states for 
544: $\Lambda=0$ and $\Lambda = 0.2$ in Fig.~\ref{fig:excited}. As expected, 
545: all excited configurations we tried are intrinsically unstable, and 
546: eventually settle down onto an equilibrium configuration. The initial 
547: excited equilibrium configurations start at the far top-right in 
548: the plots, and after a short time the systems approach the branches of 
549: ground state configurations shown in Fig.~\ref{fig:plotone}; the plot 
550: corresponding to the case $\Lambda=0$ is a reproduction of Fig. 16 in 
551: \cite{fsglau2004} with the appropriate redefined radius. The time 
552: of decay is of the order $10^2 m^{-1}$, which is a quite small time scale 
553: compared to the age of the universe, even for small values of the boson
554: mass\footnote{Even for an ultra-light boson mass of $10^{-23}$ eV,
555: the time decay is of order $\sim 100 m^{-1} \sim 10^3$ yr.}.
556: 
557: \begin{figure*}
558: \includegraphics[width=8cm]{f4a.eps}
559: \includegraphics[width=8cm]{f4b.eps}
560: \caption{\label{fig:excited} (Left) The evolution of initial equilibrium 
561: configurations in excited states with $\Lambda=0$. Such configurations 
562: start up with a given mass and radius (top-right in the figure) and after 
563: a while they approach one of the equilibrium solutions in the ground 
564: state through the emission of bursts of scalar field particles; the late 
565: time tendency is the approach to one of the points in the 
566: solid curve that contains all the possible ground state equilibrium 
567: configurations. This is a typical case of gravitational cooling. (Right). 
568: The same attractor behavior for $\Lambda=0.2$. In both plots the solid 
569: lines correspond to those shown in Figure \ref{fig:plotone}. For all 
570: these systems the quantity $2K+3I+W$ was oscillating around zero with an 
571: amplitude converging to zero.} 
572: \end{figure*}
573: 
574: 
575: 
576: % -----> Conclusions
577: \section{Conclusions}
578: \label{sec:conclusions}
579: 
580: We have found stationary solutions to the spherically symmetric SP system 
581: of equations with a self-interaction term of the Gross-Pitaevskii type,
582: according to the mean field approximation of Bose condensates. We
583: found that equilibrium configurations exist that are virialized and stable
584: under the action of small perturbations. 
585: 
586: For initial profiles with a high kinetic energy, we have 
587: shown that there are bursts of scalar field involving a considerable 
588: amount of matter of the order of half the initial mass and for stationary 
589: excited initial configuration the emission of mass is even higher; after 
590: a while, the system relaxes and virializes around an equilibrium
591: configuration. This shows that equilibrium configurations are
592: late-time attractors. We have made numerical experiments with less
593: exotic initial profiles, like gaussians, and found the same tendency
594: to virialize and to accommodate around an equilibrium configuration.
595: 
596: What this shows is that a rather arbitrary initial fluctuation of 
597: self-gravitating scalar field, in the Newtonian regime, \emph{always}
598: evolves towards a virialized configuration, which should
599:     be compared to the relativistic case, in which some
600:     particular initial configurations lead to the formation of black
601:     holes, see\cite{seidel91,balakrishna98,alcubierre2003,guzman2004}. 
602: Hence, a
603: Bose condensate made of ultralight scalar field particles, as those
604: proposed to be the dark matter, {\it tolerates} the introduction of a
605: self-interaction term in the mean field approximation.
606: 
607: We expect the results presented all along this work could be
608: useful for models of scalar dark matter, as some papers have explored
609: the idea of including self-interaction terms 
610: \cite{matoslurena2004,arbeyetal,sahni2004}. We also hope this work
611: will help others to decide whether such objects can be found in the
612: cosmos.
613: 
614: % -----> ACKNOWLEDGMENTS
615: 
616: \acknowledgments
617: This research is partly supported by the bilateral project DFG-CONACyT 
618: 444-113/16/3-1; CONACyT grants 32138-E, 34407-E and 42748; PROMEP grants 
619: UGTO-CA-3 and UMICH-PTC-121; CIC-UMSNH-4.9 and Concyteg 05-16-K117-032. 
620: The runs were carried out in the Ek-bek cluster of the ``Laboratorio de 
621: Superc\'omputo Astrof\'{\i}sico (LASUMA)'' at CINVESTAV-IPN.
622: 
623: % -----> BIBLIOGRAPHY
624: \begin{thebibliography}{}
625: \bibitem[Alcubierre et al. 2002]{alcubierre2002} Alcubierre, M., Guzm\'an, 
626:         F. S., Matos, T., 
627:         N\'u\~nez, D., Ure\~na-L\'opez, L. A. and Wiederhold, P. 2002, 
628:         Class. Quantum Grav. {\bf 19}, 5017.
629: \bibitem[Alcubierre et al. 2003]{alcubierre2003} Alcubierre, M., Becerril, 
630:         R., Guzm\'an, F. S., Matos, T., N\'u\~nez, D. and Ure\~na-L\'opez, 
631:         L. A. 2003, Class. Quantum Grav. {\bf 20}, 2883. 
632: \bibitem[Arbey et al.]{arbeyetal} Arbey, A., Lesgourgues, 
633:         J. and Salati P., 2001 \prd {\bf 64}, 123528; {\it ibid} 
634:         2002, {\bf 65}, 083514; {\it ibid} 2003, {\bf 68}, 023511.
635: \bibitem[Balakrishna et al. 1998]{balakrishna98} Balakrishna, J., Seidel, 
636:         E. and Suen, W-M. 1998, \prd {\bf 58}, 104004
637: \bibitem[Choi 1998]{dale-phd} Choi, D. I. 1998, Ph D Thesis, The 
638:         University of Texas at Austin.
639: \bibitem[Choi 2002]{dale-choi} Choi, D. I. 2002, \pra {\bf 66}, 
640:         063609.
641: \bibitem[Colpi et al. 1986]{wasserman} Colpi, M., Shapiro, S. L. and 
642:         Wasserman, I. 1986, \prl {\bf 57}, 2485.
643: \bibitem[Giovanazzi et al 2001]{giovan2001} Giovanazzi, S. O'Dell, D. 
644:   and Kurizki, G. quant-ph/0010045
645: \bibitem[Gross 1963]{gross}Gross, E. P. 1963, J. Math. Phys. {\bf 4}, 195.
646: \bibitem[Guzm\'an 2004]{guzman2004} Guzm\'an, F. S. 2004, \prd
647:         {\bf 70}, 044033.
648: \bibitem[Guzm\'an 2006]{guzman2006} Guzm\'an, F. S. 2006, \prd
649:         {\bf 73}, 021501(R).
650: %\bibitem[Guzm\'an 2004]{guzman2004} Guzm\'an, F. S. 2004, Phys. Rev. D 
651: %        {\bf 70}, 044033.
652: \bibitem[Guzm\'an \& Ure\~na-L\'opez 2003]{fsglau2003} Guzm\'an, F. S. and 
653:          Ure\~na-L\'opez, L. A. 2003, \prd {\bf 68}, 024023.
654: \bibitem[Guzm\'an \& Ure\~na-L\'opez 2004]{fsglau2004} Guzm\'an, F. S. and 
655:          Ure\~na-L\'opez, L. A. 2004, \prd {\bf 69}, 124033.
656: \bibitem[Harrison et al. 2002]{harri1} Harrison, R., Moroz, I., and
657:   Tod, K. P. 2002, math-ph/0208045; {\it ibid} 2002, math-ph/0208046.
658: \bibitem[Hu et al.2000]{hu} Hu, W, Barkana, R and Gruzinov, A. 2000, \prl 
659:         \textbf{85}, 1158.
660: \bibitem[Israeli 1981]{israeli} Israeli, M and Orszag, S. A. 1981, J. 
661:         Comp. Phys. \textbf{41}, 115.
662: \bibitem[Lee \& Koh 1996]{koreanos} Lee, J-W and Koh I-G, 1996, 
663: 	\prd {\bf 53} 2236.
664: \bibitem[Koonin \& Meredith 1990]{cphysics} Koonin, S. E. and 
665:         Meredith, D. C. 1990, Computational Physics
666:         (Addison-Wesley Publishing Company). 
667: \bibitem[Matos \& Ure\~na-L\'opez 2001]{matoslurena2001} Matos, T. and 
668:         Ure\~na-L\'opez, L. A. 2001, \prd {\bf 72}, 063506.
669: \bibitem[Matos \& Ure\~na-L\'opez 2004]{matoslurena2004} Matos, T. and
670:   Ure\~na-Lopez, L. A. 2004, Int. J. Mod. Phys. {\bf D13}, 2287.
671: \bibitem[Page 2004]{d-page} Page, D. 2004, \prd {\bf 70}, 023002.
672: \bibitem[Pitaevskii 1958]{pitaevskii} Pitaevskii, L. P. 1958, Zh. Eksp. 
673:         Teor. Fiz. {\bf 34}, 1240.
674: \bibitem[Sahni \& Wang 2000]{sahni2000} Sahni, V. and Wang, L. M. 2000,
675:         \prd {\bf 62}, 103517.
676: \bibitem[Sahni 2004]{sahni2004} Sahni, V. 2004, astro-ph/0403324.
677: \bibitem[Schunck \& Mielke 2003]{schunck} Schunck, F. S. and
678:   Mielke, E. W. 2003, Class. Quantum Grav. {\bf 20}, R301.
679: \bibitem[Seidel \& Suen 1994]{seidel94} Seidel, E. and Suen, W-M. 1994, 
680:          \prl {\bf 6}, 1659.
681: \bibitem[Seidel \& Suen 1991]{seidel91} Seidel, E. and Suen, W-M. 1994, 
682:          \prl {\bf 72}, 2516.
683: \bibitem[Svidzinsky 2004]{svid2004} Svidzinsky, A. A. 2004,
684:   astro-ph/0006024.
685: \bibitem[Teukolsky 2000]{teukolsky00a} Teukolsky, S. 2000, \prd 
686:         {\bf 61}, 087501.
687: \bibitem[Wang 2001]{wang2001} Wang, X. Z. 2001, \prd {\bf 64}, 
688:         124009.
689: \end{thebibliography} 
690: 
691: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
692: \end{document}
693: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
694: