- Split View
-
Views
-
Cite
Cite
Hoda Ghodsi, Martin A. Hendry, Mariusz P. Dǎbrowski, Tomasz Denkiewicz, Sudden Future Singularity models as an alternative to dark energy?, Monthly Notices of the Royal Astronomical Society, Volume 414, Issue 2, June 2011, Pages 1517–1525, https://doi.org/10.1111/j.1365-2966.2011.18484.x
- Share Icon Share
Abstract
Current observational evidence does not yet exclude the possibility that dark energy could be in the form of phantom energy. A universe consisting of a phantom constituent will be driven towards a drastic end known as the ‘Big Rip’ singularity where all the matter in the universe will be destroyed. Motivated by this possibility, other evolutionary scenarios have been explored by Barrow, including the phenomena which he called Sudden Future Singularities (SFSs). In such a model it is possible to have a blow up of the pressure occurring at sometime in the future evolution of the universe while the energy density would remain unaffected. The particular evolution of the scale factor of the universe in this model that results in a singular behaviour of the pressure also admits acceleration in the current era. In this paper we will present the results of our confrontation of one example class of SFS models with the available cosmological data from high-redshift SNe, baryon acoustic oscillations (BAOs) and the cosmic microwave background (CMB). We then discuss the viability of the model in question as an alternative to dark energy.
1 INTRODUCTION
The discovery of the acceleration of the universe (Riess et al. 1998; Spergel et al. 2003) resulted in the breaking of the link between geometry and destiny of the universe. It was now the nature of the substance driving this acceleration that would determine the ultimate fate of the universe. Current observations are consistent with a cosmological constant as the origin of this acceleration, but they are not yet able to rule out other possibilities such as phantom energy which will drive the universe towards a ‘Big Rip’ singularity. This situation has encouraged the study of other exotic singularities. One such example is the Sudden Future Singularity (SFS, Type II according to the classification in Nojiri, Odintsov & Tsujikawa 2005), first proposed by Barrow (2004). Other exotic singularities discovered to date include the Finite Scale Factor (FSF) singularity (Type III, Nojiri et al. 2005), the Generalized Sudden Future Singularity (GSFS, Dabrowski, Denkiewicz & Hendry 2007), the Big Separation (Type IV, Dabrowski & Denkiewicz 2010) and the w-singularity (Dabrowski & Denkiewicz 2009). These singularities may serve as alternatives to dark energy (Dabrowski & Denkiewicz 2010). Furthermore, the Big Brake is a special kind of a singularity of the SFS type, which can arise in tachyonic models (Kamenshchik, Kiefer & Sandhofer 2007). In this paper we present the results of our investigation of one example class of models which accommodate an SFS, also proposed by Barrow, by confronting it with the current cosmological observations.
The present work is not the first time that the SFS models have been confronted with cosmological observations. Previously, Dabrowski et al. (2007) performed such a test using the luminosity distance redshift relation applied to SNe data. Their paper showed that these data were consistent with the SFS model over the redshift range probed by the SNe, and moreover could permit an SFS that would occur in a surprisingly short time: less than 10 Myr from now. In this paper we extend the analysis to confront this class of SFS models with some other available cosmological observations and constraints, from the cosmic microwave background (CMB) radiation, baryon acoustic oscillations (BAOs) and the age of the universe.
The structure of this paper is as follows. In Section 2 we introduce some relevant theory underlying SFS models. Section 3 gives an account of the cosmological probes we have used to constrain our SFS model, followed by an explanation of the methods we employed to perform these tests. In Section 4 we present the results of our investigations. Finally in Section 5 we give our conclusions.
2 SFS MODELS
The finiteness of the energy density and the fact that p→∞ at an SFS means that the null (ρc2+p≥ 0), weak (ρc2≥ 0 and ρc2+p≥ 0), strong (ρc2+p≥ 0 and ρc2+ 3p≥ 0) energy conditions are satisfied but the dominant energy condition (ρc2≥ 0, −ρc2≤p≤ρc2) is violated.
We can see that the divergent behaviour of the pressure cannot be linked to the finite energy density or we will not have an SFS. Hence we need to release the assumption of an equation of state which imposes a link between these two quantities. It should be noted that an SFS can occur also in inhomogeneous and anisotropic models (Barrow 2004; Dabrowski 2005).
Similar to sudden singularities are the so called ‘quiescent’ singularities which occur in braneworld models (Shtanov & Sahni 2002). These are a type of sudden singularities whereby the pressure remains finite alongside the density and the Hubble parameter while higher derivatives of the scale factor diverge. Alam & Sahni (2006) confronted braneworld models with quiescent singularities with type Ia supernovae (SNe Ia) and BAO data and concluded that they would not fit observations. Furthermore the exact same SFS with the divergent pressure occurs in non-local cosmology as found by Koivisto (2008). The accelerating universe in this model will lead it towards an SFS rather than a de Sitter type epoch. Koivisto, however, shows that these singularities may be avoided by a slight modification of the non-local model.
Physically SFSs manifest themselves as momentarily infinite peaks of tidal forces, but geodesic completeness is satisfied in our SFS model. Hence SFS are regarded as weak singularities (Fernandez-Jambrina & Lazkoz 2004) which means infalling observers or detectors would not be destroyed by tidal forces (Dabrowski & Denkiewicz 2010). The universe will continue its expanding evolution beyond such weak singularities until for example the occurrence of a more serious singularity that is geodesically incomplete like the Big Rip which can end the universe (Dabrowski et al. 2007).
The other parameters of the model that we should consider are: as, ts and δ. The parameter as, which sets the physical size of the universe at the time of the SFS, will cancel out in the equations of the standard cosmological probes we have used to test the model. Thus we do not need to consider this parameter further.
For constraining the time, ts, when an SFS might occur, we can introduce the dimensionless parameter y0=t0/ts, where t0 is the current age of the universe. Since the singularity is assumed to be in the future, it follows that 0 < y0 < 1.
The current acceleration obtained in the SFS model, as a result of the particular form adopted for the scale factor, leads to a divergence of the pressure at sometime in the future evolution of the model. Dabrowski & Denkiewicz (2010) refer to the cause of this late time acceleration in the SFS model as a ‘pressure-driven dark energy’.
With the parameters n, m and y0 having definite values or ranges, it remained to identify a suitable range of investigation for the non-standardicity parameter δ. In Dabrowski et al. (2007) negative values of δ were associated with acceleration, but we revisited this question in order to check rigorously the range of values of δ which should be considered, taking into account all relevant physical constraints. In so doing we imposed the established observational facts that both the first and second derivatives of the scale factor are currently positive – i.e. we rejected any combination of SFS model parameters for which or . In addition we imposed the physical condition that positive and negative redshifts should correspond to past and future events, respectively. Finally we checked the sign of throughout the evolution of the SFS model and rejected parameter combinations that would predict contraction (i.e. ) at any point in the interval 0 ≤z≤ 1, on the (conservative) grounds that the expansion of the universe is securely observed from e.g. the Hubble diagram of SNe Ia over this range of redshifts. We did not, however, make any further assumptions about the expansion history of the universe outside of this range; in particular parameter combinations that would predict e.g. future contraction of the universe were not excluded from our analysis.
Thus by fixing m = 2/3, as previously discussed, and varying n and y0 over their permitted ranges, we sought to identify those values of δ which were consistent with the above physical constraints. It quickly became apparent that this task was not possible analytically. Therefore we carried out a numerical exploration of the three-dimensional parameter space (n, y0, δ). This then told us that δ should not be positive.
3 COSMOLOGICAL CONSTRAINTS AND ANALYSIS METHODS
3.1 SNe Ia redshift–magnitude relation
Fig. 1 shows the distance modulus as a function of redshift for the SFS model and the concordance model for their respective best-fitting parameters and how they compare with the best SNe data set available. One can see that the SFS model fits the data very well and that it is almost distinguishable from the concordance model over the redshift range probed by the Union2 sample.
3.2 CMB distance priors: R and la
In order to test a model against the CMB data one would ideally calculate the full angular power spectrum of the temperature anisotropies for the model, and compare it with the observed angular power spectrum. However, since this process is rather computationally intensive and complex, a simpler approach is instead to calculate the distance scales to which the power spectrum is very sensitive. The positions of the peaks and troughs of the CMB power spectrum, which can be measured precisely, provide a measure of the distance to the decoupling epoch. The distance ratios measured by the CMB are as follows.
- The angular diametre distance to the last scattering surface (at redshift zCMB, which we take to be 1089), DA(zCMB), divided by the sound horizon size at the decoupling epoch, rs(zCMB) which is quantified by the ‘acoustic scale’ and is defined bywhere the factor (1 +zCMB) accounts for the fact that DA(zCMB) is the proper angular diametre distance and we calculate rs(zCMB) following Wang & Mukherjee (2007).12
- The angular diametre distance to the last scattering surface divided by the Hubble horizon size at the decoupling epoch, which is called the ‘shift parameter’, R and is given by Komatsu et al. (2009):where Ωm is the matter density parameter. It was shown by Wang & Mukherjee (2007) that using these two parameters together is necessary in order to place tight CMB constraints on the parameters of the model of interest. They found that models with the same parameter R but different values for la, and vice versa, in general gave different CMB angular power spectra.13
The observed values for these quantities: R = 1.725 ± 0.018 and la = 302.09 ± 0.76, are obtained from the WMAP7 data and given by Komatsu et al. (2011). However, as pointed out by Elgaroy & Multamaki (2007) the shift parameter is not a directly measurable quantity and it is in fact derived from the CMB data assuming a specific cosmological model. Care therefore needs to be taken when using this quantity as a cosmological constraint.
The value for the shift parameter quoted above, as Komatsu et al. (2009) explains, has been derived assuming a standard FLRW universe with matter, radiation, curvature and dark energy components. The SFS model we are considering is also a standard Friedmann model but it assumes no explicit dark energy component; instead cosmic acceleration is driven by the divergence of pressure resulting from the particular form of scale factor adopted in the model. Matter is also permitted in the SFS and in fact, since we require our model to reduce to the EdS case at early times, we adopt the same matter content as that in the concordance model. Concerning the radiation and curvature components, we follow Komatsu et al. (2009) and ignore these in the shift parameter calculation.
Turning to the dark energy component, here we followed the approach of Elgaroy & Multamaki (2005) and expressed our SFS model in a form equivalent to an evolving dark energy model by computing its effective equation of state, weff(z). (See the appendix for a short derivation of the expression for weff.)
We then looked at the evolution of this effective w index to see how it compared with the observed behaviour. Fig. 2 shows the evolution of the effective equation of state over the redshift range 0 < z < 20, for two representative sets of SFS model parameters. In both cases we see the the same general features: weff≃−1 as z→ 0 and weff→ 0 for large z. These limiting behaviours are in good agreement with current observations. Note, however, that in plot (b) the particular SFS model parameters result in the divergence of weff at certain redshifts, which is caused by the denominator in the expression on the right-hand side of equation (A5) tending to zero at these redshifts. This behaviour is discussed in Shafieloo et al. (2006) where they make the case that in models where dark energy is not treated as an explicit fluid or a field, the equation of state cannot be used as a fundamental quantity and indeed an effective equation of state may display unusual properties like singularities. Very similar divergent behaviour was found e.g. by Sola & Stefancic (2005) in computing an effective equation of state for their evolving Λ model; indeed those authors refer explicitly to each divergent feature as a ‘fake singularity, which is nothing but an artefact of the EOS parametrization’. In our case too the divergence of weff is not seen as an indication of a fundamental physical problem with the model. Nevertheless the general similarity of the limiting behaviour of weff to that of the concordance model gives us confidence, following Elgaroy & Multamaki (2007), that it remains appropriate to use the ‘observed’ value of the shift parameter when investigating our SFS model.
3.3 BAO distance parameter, A
We find that the BAO distance parameter is not immune to the model dependency issue in its derivation either. This issue is considered in detail in e.g. Carneiro et al. (2008) who, following Eisenstein & Hu (1998), identify two (implicitly assumed) conditions which should be valid in order that the BAO distance parameter is applicable. For the model in question firstly the evolution of matter density perturbations during the matter dominated era must be similar to the CC case, and secondly the comoving distance to the horizon at the epoch of matter-radiation equality should scale inversely with ΩmH20. While these conditions are not met for the Carneiro et al. model of a time varying cosmological constant, they are met in our case.
3.4 Age of the universe, t0
3.5 Hubble constant, H0
4 RESULTS
In this section we present posterior distributions for our SFS model parameters inferred from each of the cosmological probes described in Section 3. For ease of presentation we have computed a series of ‘slices’ through the two-dimensional conditional distribution of the parameters n and y0 at given values of the non-standardicity parameter δ. We adopted uniform priors for n and y0 over the intervals 1 < n < 2 and 0 < y0 < 1, respectively, on the theoretical grounds discussed in Section 2. Thus, for these uniform priors, the mode of the posterior distribution function, for each ‘slice’ in δ was coincident with the maximum of the conditional likelihood function.
We computed posterior distributions adopting both a regular grid of (n, y0) values over the two-dimensional parameter space and a simple MCMC-based approach using the Metropolis algorithm. Both methods gave very similar results, although the latter approach is more practical when exploring more general parameter spaces with larger numbers of parameters. We will consider such cases further in a subsequent paper.
Figs 3–5 present contour plots showing the Bayesian credible regions (evaluated at the 68, 95 and 99 per cent level) for n and y0, computed for each of our cosmological probes at a series of fixed values of δ. In each figure we show separately contours for each of the six cosmological probes we have considered, labelled (a)–(f), respectively. These probes are: (a) the SNe Ia redshift-magnitude relation; (b) the CMB shift parameter, R; (c) the CMB acoustic scale, la; (d) the BAO distance parameter, A; (e) the present-day value of the H0; (f) the present age of the universe, t0. Fig. 3 shows results for δ=−0.7. For larger (i.e. less negative) values of δ the posterior distributions for most of the probes were qualitatively similar; significantly, however, the predicted value of the shift parameter was such that no credible region for the SFS model parameters was found at the 99 per cent level. We return to this point below.
One can clearly see from Fig. 3 that there is no part of the (n, y0) plane where the credible regions for the six cosmological probes overlap, for this value of δ. There is substantial overlap between the SNe Ia and BAO contours; this is not surprising given that both probes are sensitive to the E(z) function over a similar range of redshifts.
The age of the universe constraint (f) is rather weak, reflecting the large uncertainty on the observed value, and only a small region of the (n, y0) plane, on the upper left of Fig. 3(f), is excluded at the 95 per cent level; these parameter values are already strongly excluded by all other probes.
In Fig. 3(b) we have magnified the shift parameter contours for greater clarity. Note, however, that the minimum χ2 value for the shift parameter in this plot is already 21.35 (corresponding to a predicted value of R = 1.641) which is unacceptably large – i.e. the likelihood function (and hence the posterior probability) for the SFS model parameters is everywhere vanishingly small for the shift parameter at this value of δ. Similar behaviour was observed for larger values of δ; indeed the minimum χ2 value becomes progressively higher as δ increases. Hence we do not show contour plots for δ > −0.7.
It is interesting to note from panels (b) and (c) in Fig. 3 that the two CMB constraints R and la produce credible regions which do not in fact show any overlap at the 99 per cent level. This illustrates the importance of the point made by Wang and Mukherjee, as noted earlier, that one should use both CMB probes in order to better constrain (or indeed, as is the case here, to reject) a given model, since these probes are sensitive to the model parameters in different ways. Moreover we also see that the credible regions for R and la do not share a common overlap with any of the other four probes either. Hence a fit to the SFS model parameters is strongly excluded for this value of δ.
Figs 4 and 5 present credible regions for two further values of δ, in order to illustrate the pattern of behaviour exhibited by the various probes as δ decreases. First Fig. 4 shows results for δ=−1. We can see that, as before, there appears to be overlap between the credible regions for SNe, BAO distance parameter, H0 and age of the universe; indeed the contours for these probes have shifted only slightly from their position in Fig. 3. However all four show no overlap at the 99 per cent level with either of the CMB probes, which in turn continue to show no overlap with each other.
Finally, Fig. 5 shows the credible regions for δ=−1.5. Here we can see that the contours follow more or less the same shapes as in the previous figures. There is still no overlap between R and la, and neither CMB probe overlaps with any of the other probes, again at the 99 per cent level. Note, moreover, that while the general appearance of the credible regions is similar to Fig. 4, the contours for the acoustic scale, la, in Fig. 5(c) have shifted slightly downwards and appear to be approaching the lower limit of y0. As δ is decreased further this trend continues and hence no fit is obtained. To emphasize that indeed the contours display no overlap with one another, we show in Fig. 6 the most important contours of SNe Ia, CMB shift parameter, R, CMB acoustic scale, la and BAO distance parameter superimposed for values of δ previously considered in Figs 3–5. One can see clearly in these figures that while the SNe Ia data are consistent with an SFS occurring in the very near future, as shown by Dabrowski et al. (2007), the same SFS parameters would not give a fit to the CMB data.
Note that for the purpose of straightforward illustration the contour plots presented in Figs 3–5 have been calculated first without the imposition of the physical constraints discussed in Section 2. A typical example of the application of these conditions, and their impact on the credible regions, is shown in Fig. 7 for the acoustic scale, la. One can see in Fig. 7(b) how the contours are abruptly cut-off, in this case by the inclusion of the condition that the universe currently be accelerating. In other words, in the region enclosed by the contours in the right-hand part of Fig. 7(a), e.g. for y0 < 0.6 and n > 1.5, our SFS model predicts a value of la which is in satisfactory agreement with observations, but for a universe which is not currently accelerating. The other cosmological probes such as the age of the universe and the BAO distance parameter are similarly affected, although much less severely, for certain values of δ. However, since it is already the case that we find no significant overlap between the credible regions over the entire parameter space without applying these additional physical constraints, we will not present any further plots that do include them.
5 SUMMARY AND CONCLUSIONS
In this paper we have investigated one class of SFS models proposed by Barrow, by confronting it with the currently available observational data. After introducing the theory behind the model and explaining its characteristics in Section 1, we reported on the cosmological probes we used in testing our model in Section 2, taking care throughout to consider thoroughly the applicability of each probe to the SFS model under study. We then presented the results of our investigations in Section 4. Specifically we found that the SFS model was not compatible with all of the current cosmological observations considered: while good agreement could be found with e.g. the Union2 SNe data and BAO distance parameter at low redshift, the same model parameters predicted an acoustic scale and a shift parameter for the CMB which were in strong disagreement with WMAP7 observations.
We have presented our results as a series of conditional ‘slices’ through the (n, y0) parameter space at fixed values of δ. We note that the rejection of the SFS model does not depend on our choice of two-dimensional parameter space on which to conduct our analysis. The choice of the (n, y0) space was convenient because of the ranges of these two parameters, but the absence of any region which gave a fit to the model is a robust result that can be extended to the full three-dimensional space of (n, y0, δ). In other words, by showing that there is no fit to the data in conditional two-dimensional spaces we have therefore shown that there is no fit to the data in the full three-dimensional parameter space.
In conclusion we note that our results are for a fixed value of m = 2/3, which ensures that our SFS model reduces to the EdS universe at early times. In a forthcoming paper (Dabrowski et al. in preparation) we will relax this assumption and consider SFS models in which the parameter m is allowed to vary, by confronting these models with the same cosmological observations which we have considered here.
HG would like to thank Dr Arman Shafieloo for his useful comments and suggestions.
REFERENCES