A&A 445, 43-49 (2006)
DOI: 10.1051/0004-6361:20053557
R. Gavazzi1,3 - R. Mohayaee2 - B. Fort2
1 - Laboratoire d'Astrophysique, OMP, UMR
5572, 14 Av. Edouard Belin, 31400 Toulouse, France
2 - Institut d'Astrophysique de Paris, UMR7095 CNRS, Université
Pierre & Marie Curie, 98bis Bd. Arago, 75014 Paris, France
3 -
Oxford University, Astrophysics,
Denys Wilkinson Building, Keble Road, Oxford OX1 3RH, UK
Received 2 June 2005 / Accepted 18 August 2005
Abstract
Caustics are high-density structures that form in collisionless media.
Under self-gravity, cold dark matter flows focus onto caustics
which are yet to be resolved in numerical simulations and observed
in the real world. If detected, caustics would provide
strong evidence for dark matter and would rule out alternative models
such as those with modified dynamics. Here, we demonstrate how they
might be observed in weak lensing data. We evaluate the
shear distortion and show that its radial profile is marked by a
characteristic sawtooth pattern due to the caustics in dark matter haloes
that form by selfsimilar accretion. We discuss
the observational complications, mainly due to the poor knowledge
of the virial radii of the haloes and demonstrate that a superposition
of about 600 cluster-size
haloes would give a signal-to-noise ratio which is sufficiently large
for the detection of caustics with ground-based observations. This
number is reduced to 200 for space-based observations.
These bounds can be easily achieved by the ongoing wide field optical
surveys such as CFHTLS and the future space-based projects
SNAP and DUNE which have to be accompanied
by an X-ray follow-up of the selected clusters
for a precise determination of their virial radius.
Key words: cosmology: dark matter - cosmology: gravitational lensing
The nature of dark matter, which constitutes about
of the mass of the
Universe, remains largely unknown. Results from cosmic microwave background
explorers and large-scale galaxy surveys suggest that dark matter is cold with
little velocity dispersion (e.g. Spergel et al. 2003; Tegmark et al. 2004).
If so, then its evolution is mainly governed by its self-gravity
and expressed by Jeans-Vlassov-Poisson equations.
The collisionless nature of dark matter predicts the formation of multi-stream
regions bounded by very high density manifolds known as caustics.
In a cold dark matter Universe, caustics can form on large scales of many megaparsecs, manifesting in the filamentary structure of galaxy distribution (Shandarin & Zeldovich 1989) and also at smaller scales of a few parsecs or kiloparsecs in dark matter haloes. This is well illustrated in some pionering numerical works (Melott & Shandarin 1989). Due to their abundance, the rich observational and numerical data and their high density contrasts, haloes are likely areas for the caustics.
Analytic models for the formation and evolution of dark matter haloes are still rare and most works are based on the selfsimilar accretion model (Gott 1975; Gunn 1977; Fillmore & Goldreich 1984; Bertschinger 1985). In this model, first proposed to explain the rotation curve of galaxies, haloes form by temporally self-similar collapse of dark matter shells onto an initially over-dense perturbation. Dark matter shells initially expand until they reach their turnaround radii where they separate from the background expansion and collapse. After collapse they re-bounce and collapse again and the density profile settles asymptotically into a power-law which is convolved with singular spikes, namely with caustics.
In the following we shall refer to such caustics as "outer caustics''. They are suggestive of the sharp stellar shells observed around giant ellipticals which can arise in the merger of galaxies (Fort et al. 1986; Quinn 1984; Malin & Carter 1980). The main observational difference between the merger and selfsimilar spherical infall is that the former predicts that the caustics are interleaved with the caustic radii alternating on opposite sides of the galaxy and the latter predicts concentric spherical shells.
The spherically symmetric model has been extended to consider infalling matter with angular momentum and calculated the properties of an additional kind of "inner caustics'' with torus-like topology (Sikivie 1998,1999). Such accretion with angular momentum is more relevant for galaxy-size haloes than for clusters of galaxies dominated by radial infall.
Here, we focus on the outer caustics of cluster-size haloes and argue that they can be reasonably approximated by selfsimilar infall models. In the original version of this model, dark matter is absolutely cold, i.e. with zero velocity dispersion, and caustics are infinitely thin concentric spherical shells with diverging densities. However, realistically, dark matter has a small velocity dispersion and these shells have finite thicknesses. The thicknesses of the caustics would however remain very small (due to the coldness of dark matter) and thus they would contain very little mass in spite of their significant density. Various characteristics of the caustics such as their density profile, their thickness and their approximate maximum density for a low velocity dispersion dark matter medium have been recently evaluated (Mohayaee & Shandarin 2005, hereafter MS05).
Detection of dark matter caustics remains a challenging problem both
for 3-dimensional numerical simulations and for observations.
Dark matter annihilation in the caustics has already been studied
(see e.g. Mohayaee & Shandarin 2005; Sikivie et al. 1997; Pieri & Branchini 2005; Hogan 2001). Since
the flux of the annihilation products, e.g. -rays
or antiproton flux, depends on the square of the local
density
, caustics with their
sharply-peaked densities would be the likely places for significant
emissivity. Such kind of observations are promising for the eventual detection
of dark matter in the caustics. However, the major problem with dark matter
detection through gamma-ray emission is the severe background
contamination of the signal and the low signal-to-noise ratio of
present day observations.
It has been shown that rotation curves of galaxies might be sensitive to the presence of inner caustics and claimed a marginal detection based on an ensemble average over 32 rotation curves (Kinney & Sikivie 2000). Inner caustic rings are located in the plane of the disk and are likely to modify rotation curves more efficiently than outer caustics with spherical symmetry. Moreover, rotation curves only probe the dark matter potential of galactic haloes where a large amount of tracers (gas or stars) is available. Therefore, outer caustics which are located well beyond the observable tracers cannot leave a detectable imprint on rotation curves.
Gravitational lensing provides a promising alternative tool. Since lensing probes the projected density profile with no regards to the nature or dynamical state of the deflecting mass, it should be sensitive to the caustics. The lensing properties of dark matter caustics and particularly their efficiency in magnifying and/or producing multiple images of background sources has already been investigated (Charmousis et al. 2003).
Inner caustics may be dense enough to produce substantial magnification and small separation multiple images as in micro-lensing by compact objects. High magnification events due to such caustics may explain the anomaly in flux ratios observed in multiply-imaged quasars which are hardly reproduced by a smooth halo or even subhalos (see e.g. Kochanek & Dalal 2004; Dalal & Kochanek 2002).
Conversely, it has been found that outer caustics are inefficient in magnifying distant sources and would yield at most a few percent net magnification or shear (Charmousis et al. 2003). As we shall detail below, the selfsimilarity of dark matter accretion implies that outer caustics occur at the same radius provided physical radii are properly rescaled by the halo mass. Hence the tiny lensing signal of caustics could show up statistically by averaging over many rescaled haloes. Such a statistical approch is much more complicate for the lensing-based detection of inner caustic rings because of the random orientation of rings (or angular momenta).
In this work, we consider the lensing properties of the outer caustics only and we propose the weak-lensing effect as a potential way to detect caustics. We demonstrate that the caustics will produce sharp variations in the projected surface mass density around haloes. Depending on the height and width of caustics, gravitationally-distorted background galaxies will experience local variations of shear.
If the aforementioned universal property of haloes is fullfilled,
then the rapid progress in X-ray and
lensing observations of cluster of galaxies may offer one possibility to
observe dark matter caustics. Deep and wide surveys such as CFHTLS
will cover fields of view of a few hundreed square degrees and will
provide us with a useful material for the detection of dark matter caustics
through the capability of mass to coherently stretch the image of the
background galaxies. They will provide us with a large enough number
of galaxy clusters to achieve the required level of signal-to-noise
ratio. Wide field spatial surveys will be
even more powerful for the investigation of the lensing properties of
dark matter haloes and their associated caustics. An important
requirement for the detection of caustics is the measurement of virial
radii of the clusters which can be provided by X-rays
(e.g. Arnaud et al. 2005).
Throughout this paper, we assume an Einstein-de Sitter Universe but our
results should be qualitatively similar in a concordance CDM model.
The role of dark energy becomes important at low redshifts (
0.2) which we expect to occur well after the formation of the typical dark
matter haloes we consider here. Furthermore, once a particle turns around
and collapses, it separates from the background expansion and its subsequent
motion should not be affected by the
term.
The paper is organized as follows. We review the three-dimensional and projected properties of self-similar haloes and caustics in Sect. 2. We derive the lensing signal for a single halo, compare it to the noise level of fiducial observations and estimate the number of haloes required to achieve a significant signal-to-noise ratio in Sect. 3, where we also examine the ability of weak lensing to constrain the velocity dispersion of dark matter particles. We summarize, discuss the prospectives for future works and conclude in Sect. 4.
Table 1:
The non-dimensional times (), positions (
),
thicknesses (
), maximum densities (
)
of the first ten caustics, the halo density itself (
)
at the positions of these caustics and other parameters used in expressions (3), (5)-(7).
The present velocity dispersion is that
for neutralinos which is about 0.03 cm s-1. It is worth emphasizing
that the thicknesses of the caustics
do not depend on the halo parameters but only on the velocity
dispersion of dark matter [see expression (6)].
It is instructive to compare the last two columns, which show that for a
cluster-size halo with
Mpc, the caustic density can be many orders of
magnitude higher than the halo density itself.
In the presence of a small velocity dispersion the maximum density and thickness of the caustic shells and their density profiles have been evaluated in the framework of a selfsimilar collapse model (MS05).
The global halo density profile,
asymptotically reached in this process, is well-approximated by
For a perfectly cold dark matter medium, the density profile close
to a caustic at
is (Bertschinger 1985)
![]() |
Figure 1:
Left panel:
in units of the background density,
![]() ![]() ![]() |
Open with DEXTER |
When the temperature of particles is not strictly zero and the velocity of
particles is distributed according to the distribution function f(v),
caustic positions are shifted by a small value
and
the caustic density is modified as (MS05):
![]() |
(4) |
The maximum density at each caustic position can also be approximated
by the expression (see Table 1 and also MS05)
Although the selfsimilar model might seem naive, it provides a good approximation for outer caustics in galaxy-cluster haloes which are not significantly disrupted by merger and substructures and to a good approximation are spherical. Furthermore, fluctuations caused by large scale structure would already be averaged out in the statistical evaluation of the shear.
Since we are concerned with the lensing properties of caustics, we have to
calculate the projected density profile for the caustics and for the halo.
The Abel integral relates the 3-dimensional density ()
and the 2-dimensional density profiles (
) by
![]() |
(8) |
![]() |
(9) |
![]() |
(10) |
Figure 1 shows the 3D (left) and 2D (right) density
profiles for the halo and the caustics. We consider the case of a
perfectly cold dark matter with very peaked caustics and the case in
which they are smoothed out by a finite velocity dispersion
.
This value is very high for most cold dark
matter models but its extremity illustrates well the peak dilution
and shows that caustics survive a large thermal softening. When
considering the projected density, instead of spikes we rather see
stairs that come from the analytic
singularity of
caustics. Consequently even with singular caustics, the projected
density profile is not peaked. The implications for lensing are
discussed in the next section.
The fundamental quantity for gravitational lensing is the lens potential
at the angular position
which is related
to the surface mass density
projected into the lens
plane through:
![]() |
(11) |
For a circularly symmetric lens,
is oriented tangentially
to the lens center and its amplitude at radius r is
.
Since sources are randomly oriented, the tangential component of the observed
galaxies is an unbiased estimator of
.
When averaging the estimate
of
within an aperture of solid angle
,
containing
galaxies (n is the number density of sources), the noise
dispersion of
is
where
is the intrinsic dispersion of source ellipticities
(along one component).
In order to be consistent with our calculations of Sect. 2.2, we define a pseudo-shear:
and the corresponding noise level
.
For an EdS cosmology, and considering an annulus of inner and outer radii
and
respectively, it is straightforward to write
in this useful form:
![]() |
(14) |
Figure 2 shows
as a function of
for the same
values of the thermal velocity dispersions
and
.
Comparing these curves, one can see that the sawtooth patterns due to
caustics survive significantly high temperatures.
Next, we consider the noise level for a fiducial halo at redshift
and a turnaround radius
which is a typical value for clusters
(upper green "binned'' curve). With a single halo the detection of caustics
is impossible. If we are able to stack the signal from a few tens of
clusters (lower blue-binned curve), the noise level will be low enough to
be sensitive to caustics as a whole. However, we would detect a smooth
contribution which is indistinguishable from the halo itself.
Sawtooth patterns cannot be confused with the effect of substructures since the contribution of the latter would be averaged out over the azimuthal angle (most outer caustics have spherical symmetry) and once rescaled, caustics always appear at the same place within haloes. This is not the case for substructures which can appear at any radius inside the host halo.
![]() |
Figure 2:
![]() ![]() ![]() ![]() |
Open with DEXTER |
Consequently, the right way to probe the existence of caustics is to
measure the
signal in excess/default relative to the extreme value
of
,
e.g.
as taken here.
For this purpose, one needs
100 stacked clusters.
With the corresponding noise level, it would be possible to test the
thermal smoothing of caustics and put constraints on
.
However, the sensitivity is poor and only upper limits
can be put on
with a realistic number of haloes.
For instance with N=200 (resp. 500) clusters, we could achieve
a limit
(resp.
)
at a
confidence level.
When considering galaxies instead of clusters with turnaround radii
about 10 times smaller, the number of haloes required to achieve the same
detection level is 104 times higher. So using a few 105 galaxies
between
would yield the same results.
This required level of signal can easily be achieved
with a wide and moderately deep survey like
the ongoing CFHTLS. Typically, a square-degree field of view will contain
a few such cluster-size haloes with
,
a few thousand
elliptical galaxies with
and tens of thousands of
spiral galaxies with
.
The total coverage of the
CFHTLS wide survey is 170 square degree and will contain a large enough
number of clusters/galaxies. Furthermore the wide fields of view are
well-suited to measure shear up to the outermost caustic of clusters
(
at z=0.3 and
at z=0.5).
Spatial observations provide an improvement on shear measurements: (i) the intrinsic dispersion in source ellipticities is lowered
and (ii) the density of sources is increased
arcmin-2. Hence the total number of haloes required to
achieve the same detection level can be lowered by a factor of
three. Future wide spatial surveys like SNAP or DUNE will provide
the required sky coverage.
![]() |
Figure 3:
Difference between the
![]() ![]() ![]() |
Open with DEXTER |
Figure 3 shows the blurring of caustics due to an imperfect
knowledge of the turnaround radius of each stacked halo. We consider a
dispersion around the true value of 3% and 10%. To properly detect
caustics, one needs a precise estimate of
.
Provided dark matter is cold enough,
a few
,
which is a realistic prescription, a reasonable number of stacked
haloes,
600, can overcome the loss of a few percents of relative
precision caused by the errors in the determination of the turnaround radius.
Consequently the number of haloes necessary for detection needs to be
increased in order to achieve the necessary level of signal-to-noise
ratio. Moreover, additional external information for the
determination of
such as X-rays or dynamical observations
in addition to weak lensing data are indispensable for a convincing
detection of caustics. The present-day state-of-the-art X-ray estimates
for the scale radius is
for nearby clusters (z<0.2)
(Arnaud et al. 2005; Pointecouteau et al. 2005) and it seems that a similar precision
can be reached at higher redshifts (0.4<z<0.7) (Kotov & Vikhlinin 2005).
In addition shear, which is used for caustic detection, also provides
constraints on the halo density profile. The virial radius of
some clusters presenting strong and weak lensing features can be measured
with good accuracy (
%)
(Gavazzi 2005; Broadhurst et al. 2005). Consequently, future large
cluster samples with X-rays and lensing data of present-day quality
will provide us with the necessary precision to probe dark matter caustics.
We have shown that the existence of dark matter caustics could be probed
by properly stacking the weak lensing signal of a reasonable
number of haloes. The main observational limitation is perhaps the
precise estimation of the turnaround radius,
,
of superimposed
haloes but we have shown that the loss of a few percent relative accuracy
in the determination of
(or asphericity) can be compensated for
by stacking about 600 haloes.
Although the sensitivity is low, a detection of caustics provides an upper bound for the temperature of dark matter, thus excluding hot dark matter models. The sensitivity is not sufficient to distinguish between various cold dark matter candidates (like axions or neutralinos) since for most of the corresponding velocity dispersions, the shear signal would be similar.
However, a detection of caustics would be a strong argument for the existence of cold dark matter since alternative models like MOND could not explain such density singularities and at most could serve in place of a smooth halo (namely provide an equivalent effective gravitational potential).
Putting constraints on the velocity dispersion of dark matter is a challenging topic in modern cosmology since it offers the possibility to pin-down an actual physical parameter of dark matter. In this paper, we investigated the possibility of such a measurement with the weak lensing effect of dark matter caustics. The implicit observational hypothesis is that we can have a selfsimilar geometrical description of the caustic shell distribution which depends on a single characteristic scaling parameter: i.e. the virial radius.
Wide field surveys such as the ongoing CFHTLS accompanied by X-ray observations can provide the required statistics for a successful detection of caustics. The number of haloes required to be superimposed will be lowered by a further factor of 3 for future space-based experiments like SNAP or DUNE.
Here, we have used the first and most common caustic singularity, that for which the density profile falls with inverse square-root of the distance from the caustic. Caustics of higher-order singularities can also appear in collisionless media and have already been classified (Arnold 1986). It remains a challenging task to generalize our work to haloes with non-vanishing eccentricity where higher-rank caustics would occur and to examine if they can modify the magnification properties of the lensed images and account for anomalous image flux ratios.
Acknowledgements
Special thanks go to Sergei Shandarin for an ongoing collaboration on dark matter caustics. We also thank Francis Bernardeau, Ed Bertschinger, Monique Arnaud, Gary Mamon and Etienne Pointecouteau for many helpful discussions and comments. R.G. is supported at Toulouse from a CNRS postdoc contract #1019 and at Oxford by a grant from the Leverhulme Trust. R.M. is supported by European Gravitational Observatory grant at the school of astronomy, university of Cardiff, UK.