WorksPhysics and Astronomy
The Full-Shape Analysis of the Fading Dark Sector Model
Abstract
We analyze the Fading Dark Sector (FDS) model, which is motivated by string theory and in which the fading of the dark sector is driven by a scalar field. The model is specified by two numbers that parametrize the logarithmic derivatives of the dark energy and dark matter densities with respect to the field in Planck units. In previous work, we showed that this model provides a good fit to the DESI BAO, CMB, and SN datasets. Here, we extend that analysis to include the DESI DR1 full-shape galaxy power spectrum, which adds information about the time evolution of cosmic structure and provides an independent consistency check of the FDS model. We find that, when the DESI DR1 full-shape data are combined with CMB+DESI DR2 BAO+DES-Dovekie, the FDS model outperforms the Chevallier-Polarski-Linder (CPL) - parameterization by at equal parameter count and is preferred over CDM at . The inferred values of with the full-shape data are unchanged from our previous analysis, so this improvement indicates that FDS makes a prediction for the growth of structure that captures the observed growth history better than the - parameterization.
Introduction
Recent measurements of the baryon acoustic oscillation (BAO) scale by DESI are in tension with predictions of the standard CDM model inferred from observations of the cosmic microwave background (CMB) [1, 2]. This tension is commonly described as a preference by DESI for a smaller matter fraction than the one obtained from CMB measurements. Additionally, supernova samples [3, 4, 5, 6] have their own different preference for a value of that is larger than the one preferred by the CMB. This means that excluding any single dataset still leaves a tension within CDM. Various modifications to the standard cosmological model have been proposed to explain these discrepancies (see for example [7] and references therein). These include phenomenological and astrophysical approaches such as the possibility of having a larger optical depth or replacing the cosmological constant with a dark energy fluid, parametrized by its equation of state (such as or more general variations such as in [8]). Other attempts include modifications to the physics such as spatial curvature, non-minimal couplings, modified gravity, modified neutrino sector or interacting dark sectors [9, 10, 11]. The Fading Dark Sector model (FDS) studied here, which was motivated from string theory (rather than DESI data) and proposed in 2019 [9], belongs to the latter category.
Several of these models are degenerate at the level of the background cosmology, which makes them indistinguishable using distance measurements alone, such as the BAO scale. For example, distance measurements can be used to infer an effective for a dark energy fluid but several underlying physical theories can give rise to the same expansion history and hence the same effective despite having important differences in their microphysics (see for example [12]). The full shape (FS) of the galaxy power spectrum contains additional information that can break this degeneracy between models [13]. This information arises from redshift space distortions (which give sensitivity to the growth rate), the Alcock–Paczynski effect (which gives geometric information on the expansion
sion) and the broadband shape of the power spectrum (which can determine the scale of equality). Altogether, full-shape (FS) analyses can discriminate between models that look equivalent at the background level.
In physical models that arise from a concrete modification of the underlying physics, the growth rate of structure, along with the other FS observables mentioned above, are predictions of the model. This is because the same parameters that modify the background expansion also fix the evolution of perturbations. This is unlike phenomenological parameterizations where the behavior of perturbations is an independent assumption that is not tied to . In the typical – parameterization, for instance, one assumes that the dark energy fluid is smooth and gravity is general relativity and only after making these assumptions can one calculate modifications to the growth rate of structure. A model that is defined microscopically does not have room for these additional assumptions. FS data can therefore be used as a test which can falsify or disfavor some models relative to others. In particular, there could be models, where the background evolution alleviates the tension between geometric observables, which fail the FS test, e.g., by producing too much structure. In this paper, we perform this test on the fading dark sector (FDS) model.
The FDS model fits well with the cosmology of the Dark Dimension Scenario [14] which is motivated by string-theoretic considerations and leads to a unification of the dark sector (see [15] for a review). The FDS model, which we discuss in section 2, is a two parameter extension of CDM. Relative to CDM, it includes an additional scalar field which controls both the dark energy and the dark matter mass. Locally in field space the potential is approximated by an exponential where we are measuring in Planck units. The mass of the dark matter is likewise taken to be locally . The two exponents which denote the logarithmic slopes of dark energy and dark matter with respect to , are the additional parameters of the model and are inputs from microscopic physics that determine the cosmological evolution. The most non-trivial aspect of the FDS model, in addition to the fact that it couples the dark energy and dark matter through , is that based on expectations from string theory the parameters should take values. This feature of the FDS model distinguishes it from a generic interacting dark matter/dark energy model. For example, with , one expects in Planck units with no tuning beyond that of the vacuum energy itself. Indeed fitting FDS to recent cosmological data, such as the DESI DR2 BAO and DES-Dovekie supernova measurements, leads to parameters for both [16] as predicted in [9].
The evolution of is dictated by the potential during dark energy domination and rolls in the direction that reduces the dark energy density. On the other hand, during matter domination, the evolution of is dictated by the dark matter density and rolls during that era in the direction that decreases the dark matter density. The evolution of therefore naturally causes the dominant dark component, be it the dark matter or dark energy, to decrease and gives rise to the fading dark sector (FDS).
Our analysis shows that the inclusion of FS data leaves the inferred mean values of unchanged from those preferred by geometric distance measures (such as DESI DR2 BAO) and it also tightens the constraints on in a way consistent with our previous findings [16]. Importantly, not only is the fit not degraded by adding information from the FS power spectrum, but the significance of the model improves relative to both CDM as well as the phenomenological CPL parametrization. We find that, when the DESI DR1 full-shape data are combined with CMB+DESI DR2 BAO+DES-Dovekie, the FDS model outperforms the CPL parameterization by at equal parameter count and is preferred over CDM at ! This pattern persists even without any SN datasets, i.e., only with DESI and CMB data, which yields a significance over CDM and over CPL by .
The rest of the paper is organized as follows. In section 2, we review the cosmology of the FDS model and the structure of the one-loop power spectrum. In section 3, we introduce the data used in this paper. In section 4, we present results from the Markov Chain Monte Carlo analysis of the FDS model and the physics of the best-fit model. We also analyze the growth of structure in the FDS model. In section 5, we end with concluding remarks. We relegate the derivation of the equations of motion governing the one-loop contribution to the multipole moments of the galaxy power spectrum to appendix A and discuss the mapping of parameters to the DESI FS data in appendix B.
Cosmological dynamics
In the fading dark sector, the dark energy and dark matter are coupled through a scalar field
where is a dimensionless scalar field. As we have discussed in the introduction, the scalar field evolves with the effective potential which leads to fading of dark matter (in matter dominated era) and dark energy (in the dark energy dominated era). The scalar in the FDS is light, so that it is able to roll on cosmological scales, and its coupling to dark matter means that it can mediate a long-range fifth force in the dark sector. On subhorizon scales the scalar perturbation can be treated in the quasi-static approximation. Its effect is an additional long-range fifth force (with strength ) to gravitational force acting only on the dark matter:
where denotes the coupled dark matter component and
with the effective mass of the scalar perturbation. Thus, on short scales, , the force between dark matter particles is enhanced by a factor as was calculated in [16]. The controlling the strength of the fifth force is also the same parameter that sets the fading of dark matter in the matter dominated era. This is an example the lack of freedom in physical models to modify cosmological dynamics separately from the microphysics.
The coupling also changes the momentum dilution of the dark matter fluid. It is useful to characterize this effect by the dimensionless quantity
The two modifications have physically distinct origins: describes the additional scalar force originating from the fluctuation of the scalar field, while describes the modified drag associated with the evolving dark matter mass. The potential parameter affects structure formation indirectly through the background evolution of , , and the matter fractions, and therefore also through .
The background and linear perturbation equations for the FDS model have been derived and implemented previously in [16]. Here, to probe the physics of the FDS model in the mildly non-linear scale, we adapt the framework of the effective field theory of large-scale structure (EFTofLSS) for the FDS model. The EFTofLSS for the FDS model is considerably more involved compared to e.g., CDM. In particular, we cannot rely on the derived Einstein–de Sitter kernels of the EFTofLSS for CDM due to 1) the baryons and fading dark matter have different equations of motion in the regime of interest and 2) the additional terms arising from the fifth force enhancement and the scalar induced friction appearing in the modified Euler equations for perturbations of dark matter are time dependent. Hence, we compute the FDS kernels and evolve the system with the exact time dependence. Importantly, for the EFTofLSS for the FDS model, we do not assume that the density and velocity perturbations of the baryon and fading dark matter fluid can be factorized into time-independent kernels, where the time-dependence can be entirely absorbed into the growth factors. Hence, the full system of kernels, including the recursion relation defining higher-order corrections, are all time dependent.1 galaxy survey measurements, we perform an Alcock–Paczynski mapping given by eq. (82) which brings galaxy power spectrum into its appropriate form for eq. (6).
Galaxy power spectrum
In redshift space, the galaxy power spectrum can be organized as
where is the cosine of the angle between the Fourier mode and the line of sight . The first term contains the linear density and redshift-space-distortion contribution, contains the standard and one-loop terms evaluated with our model-specific kernels, contains the EFT counterterms that absorb unresolved short-distance dynamics, and describes stochastic contributions associated with the discrete galaxy tracers. The observable multipole moments of the redshift space galaxy power spectrum can then be deduced from as
where is the Legendre polynomial of order . We compute the monopole, quadrupole, and hexadecapole, corresponding to .
The detailed derivation of for the FDS model is given in appendix A. Here, we outline this procedure for computing and consequently . We start with deriving the higher-order corrections to the continuity and Euler equations for baryons and the fading dark matter, which are given by eqs. (25) to (27). To solve these equations, we adapt the kernel expansion for the density and velocity perturbation for baryons and dark matter given in eqs. (33) and (34). Then, we derive the exact-time dependent recursion relation for the kernels in eq. (39). With the higher-order corrections computed for the matter power spectrum, we use the two-fluid Eulerian basis for the galaxy bias shown in eq. (46) and use its relation to total and relative matter perturbations (along with higher-order corrections) to derive the relevant kernels appearing in the galaxy density and velocity expansions. Upon mapping the power spectrum to redshift space via eq. (56) to account for the redshift-space distortions, adding the appropriate counterterms (A57) and stochastic corrections (A58), and lastly performing an IR resummation, we arrive at the final form of the one-loop corrected galaxy power spectrum eq. (78). As a last step to make contact with
Data
We have implemented the fading dark sector in CLASS-PT [18] and we use Cobaya [19] to run the Markov Chain Monte Carlo (MCMC) simulations.
DESI full-shape likelihood
We test our EFTofLSS analysis against the public DESI DR1 full-shape likelihood [13].2 The measured data vectors, covariance matrices, effective redshifts, scale cuts, and survey window functions are kept identical to those of the DESI analysis; only the theoretical galaxy power spectrum is replaced by the coupled dark sector description described above. The DESI DR1 galaxy full-shape measurements are carried out for all the BGS, LRG, ELG, and QSO samples. To specify the power spectrum only, we use the option spectrum-pole-rotated to separate the DR1 full-shape likelihood from the DR1 BAO likelihood.
For each galaxy sample, the three nonlinear bias parameters sampled explicitly are
where is the Lagrangian linear bias while and are the quadratic local and tidal Lagrangian bias parameters, respectively, and is the rms amplitude of the linear, zero-momentum matter fluctuations in spheres of radius at redshift .
The remaining EFT directions are chosen such that they enter the likelihood linearly. In particular, adopting the DESI notation, we analytically marginalize over two counterterm amplitudes, and , and two stochastic amplitudes, and . Their priors, together with those of the sampled bias parameters, are summarized in tbl:IV. The nuisance parameters are independent between the six DESI full-shape tracer/redshift samples.
Additional datasets
In addition to the DESI DR1 galaxy full-shape likelihoods, we also include CMB temperature, polarization, and lensing measurements, DESI DR1 and DR2 BAO measurements, and the DES-Dovekie Type-Ia supernova compilation.
For the CMB, we combine the Planck low- TT and EE likelihoods [20] with the Planck PR4/ NPIPE CamSpec high- TT, TE, and EE spectra [21, 22]. We additionally include CMB lensing information from the joint ACT DR6 [23, 24] and Planck lensing likelihood [25]. This choice is consistent with [2].
When BAO information is included together with full shape, we use the DESI DR1 FS [26] and DESI DR2 BAO [2] likelihood, which we refer to as DESI1.5. In joint analyses of DESI DR1 full-shape galaxy power spectrum and DESI DR2 BAO, [27, 28] have found that the cross correlation between the two datasets have little impact on the determination of the cosmological parameters. Hence, for our present simulations, we assume the independence of full-shape and BAO as in [27] by imposing the following likelihoods
Lastly, we also make use of the Type Ia supernova (SN) luminosity distances. In the present paper, we use the recently recalibrated DES-Dovekie sample [6]. Motivated by our previous findings for the FDS model [16], we expect little deviation for the MCMC and best-fit results among the choice of recalibrated supernova datasets (e.g., recalibrated Pantheon+ or Union3.1 [29, 30]).
Results
Since the analysis pipeline we have developed involves new kernels and modifications of standard codes, we start by checking the outputs of these codes against known results. In particular, we validate the CDM limit (i.e. and ) against results published by the DESI collaboration. As mentioned in the appendix, our new kernels also reduce to the standard CDM ones [31] in the appropriate limit. Our implementation is therefore under control in this limit.3
Having checked the CDM limit, we now allow the parameters to vary in an MCMC simulation to determine the preference for departures from CDM. These parameters are added to the standard six CDM parameters so our model has two extra parameters and we take that into account when calculating improved significances. In addition to MCMCs, we run minimizers to find the best fit parameter values. The result of these MCMC and minimizer runs are shown in Table I where we compare our FDS model to the benchmark – parametrization. The constraints on cosmological parameters in the FDS model from using various dataset combinations are shown in Table III. In order to better understand where constraints come from, we study different combinations of datasets to diagnose what information is added by each dataset.
First, we study a baseline data variation where we include only CMB data as discussed in section 3.2. The CMB alone shows no preference for departures from CDM. In particular, our model modifies late-time cosmology and leaves early-time CMB physics unaltered. For example, the scalar field has not changed appreciably from its initial value before recombination. This leaves the characteristic sound horizon fixed so that the CMB is mainly sensitive to integrated quantities along the line of sight. While there is additional information in the CMB dataset about late-time cosmology from late ISW and lensing, it seems like these effects are comparatively weak and do not show a preference for departures from CDM. The mild improvement of fit, for – is consistent with fitting noise with two extra parameters. In particular, the CPL parametrization has the freedom to wander into unphysical parameter space where secondary effects (such as the CMB lensing anomaly) fit slightly better.
Once the DESI BAO data is included, we break the degeneracy and get sensitivity to more than just integrated quantities to the CMB. In particular, DESI BAO provides information about distances to various intermediate redshifts between today and the CMB allowing different expansion histories that give the same distance to the CMB to be distinguished. With the inclusion of BAO data, we see a preference for – and FDS over CDM that is ultimately rooted in the tension described in [2]. Since the included BAO data is based on background distances, this preference is driven by departures of the distances measured by DESI from those predicted by the CMB within CDM. This raises the question of whether measurements of the growth of structure (without the BAO peak) also agree with this conclusion.
We answer this question by studying the full-shape (FS) power spectrum of DESI DR1 in combination with the CMB. The FS data contains weaker information about the BAO scale which DESI provides by a reconstruction procedure that undoes the effects of bulk flows. On the other hand, FS data contains other information that can be obtained from redshift-space distortions (RSD), the broadband shape of the spectrum and the Alcock–Paczynski effect. As mentioned previously, these are sensitive to the growth of structure and also provide information on cosmological parameters affecting the expansion history of the universe. As seen in Table I, we find that the FS data has the same preference as DESI DR2 BAO with deviations at less than . The FDS model provides a marginally better fit to CMB+FS compared to – although the difference in is only about 0.8 which could be due to noise. This is because FDS can match the low preferred by DESI data without reducing the amplitude of structure while in – there is no independent boost to the amplitude of structure. Moreover, FS data provides a better measurement of the parameter , reducing the error by a factor of 1.3. From the constraints in figure 1 and in Table I, we see that the FS data prefers a larger value of compared to DESI BAO. However, late-time effects depend on the product since the field rolls due to its potential with logarithmic slope and the DM fading is controlled by . In order to prevent effects associated to the fading of DM and modifications to the background cosmology from becoming too big, the fit prefers smaller values of , thereby also tightening the error bars.

Figure 1. Contours showing the 1- and 2- constraints on and for the FDS model across various combination of datasets.
| Datasets | FDS | |||||
| CMB | ||||||
| CMB + DESI DR2 BAO | ||||||
| CMB + DESI DR1 FS | ||||||
| CMB + DESI1.5 | ||||||
| CMB + DESI1.5 + DES-Dovekie |
Table 1. Statistical significance comparing the CPL parametrization and the fading dark sector model across various combinations of datasets, to CDM. Additionally, the best fit, the mean, and the values for , are provided for the fading dark sector model.
| Datasets | FDS | |||||
| CMB + DESI DR2 BAO + DES-Dovekie | ||||||
| CMB + DESI1.5 + DES-Dovekie |
Table 2. The effect of adding DESI DR1 FS data to the dataset combination involving CMB + DESI DR2 BAO + DES-Dovekie on the statistical significance of the best-fit for the CPL parametrization and the fading dark sector model, compared to CDM.
| Datasets | ||||
| CMB | ||||
| CMB + DESI DR2 BAO | ||||
| CMB + DESI DR1 FS | ||||
| CMB + DESI1.5 | ||||
| CMB + DESI1.5 + DES-Dovekie |
Table 3. Constraints on cosmological parameter of the FDS model.
Since the modified – and FDS are consistent under the BAO and FS datasets separately, we study the combination of both datasets which we call DESI1.5. With this combination, we see a remarkable reversal of preference for FDS over by about . The three data combinations suggest that provides a slightly better fit to background distances measured by DESI but the preference tips in favor of FDS once the FS power spectrum is included. This indicates that BAO and FS data favor different regions of the – parameter space, whereas for the FDS model the two datasets reinforce each other. With DESI1.5, we again see a tightening of the errors on by a factor of 1.7 compared to CMB+DESI DR2 BAO.
Finally, we include the supernovae (SNe) sample from DES-Dovekie which is a measure of the expansion history at low redshift. When combined with CMB+DESI DR2 BAO, the SNe sample doesn’t distinguish between FDS and – [16]. However, with FS data the FDS model reaches a 4.1 preference over CDM and is favored over – by . This is further highlighted in Table II. This is a similar preference to the one mentioned above without SNe. We see that the SNe do not disfavor the region of parameter space selected by CMB+DESI1.5 and in this sense the datasets all reinforce each other. The inferred values of also remain consistent with the ones from CMB+DESI1.5. The parameter shifts to slightly more negative values reflecting the preference of the SNe data for larger . This is an illustration of the role played by the potential in the FDS model. As observed in [34], a model with a dark fifth force would have difficulty accommodating the SNe data but we see here that our model is capable of doing that precisely because the mediator of the fifth force also plays the role of dark energy. Future surveys such as Euclid [35] or SNe samples from ZTF [36], Rubin [37] and Roman [38] will provide more constraining power and may be able to distinguish between models such as FDS and – more decisively.
Growth of structure
The FS data favor the FDS model because of its altered growth of structure compared to –. In this section, we examine this preference in more detail. We start by recalling the definition of the linear growth factor and the linear growth rate :
These curves are shown in Figure 2 for the models we consider in this analysis. In principle, and depend on the scale through the range of the fifth force but on scales relevant for the DESI data, we have and the dependence is very weak. This is illustrated in Figure 3. Because of this we choose to evaluate and at which is the pivot for the ShapeFit analysis [26]. The differences in the growth functions between models are due to slightly different expansion histories as well as evolution of perturbations.

Figure 2. Figure 2: Comparison among the best-fit growth factor and growth rate of CDM, the CPL parameterization, and the FDS model at from fitting to CMB+DESI1.5+DES-Dovekie. The growth factor is normalized to its value today.

Figure 3. The behavior of the strength of the fifth force in the dark sector as a function of redshift and wavenumber for the best-fit FDS model from fitting to CMB + DESI1.5 + DES-Dovekie. The sampled redshifts are relevant to DESI DR1 while the gray dashed vertical lines indicate the range of to which DESI DR1 full-shape data measures. The black horizontal dashed line indicates the value of for this best-fit FDS model.
In the context of FDS, The growth of structure is different because of competing physical effects. On the one hand, there is a fifth force in the dark sector whose strength is proportional to and this adds to the gravitational attraction and speeds up matter in-fall. Additionally, as dark matter loses mass, momentum dilution due to the cosmological expansion becomes slower and further enhances the growth of structure. On the other hand, the fading of dark matter decreases its mass and slows down gravitational collapse thereby reducing the growth. The importance of these effects varies during cosmic evolution and in some eras they may even be reversed (e.g. the dark matter gets heavier during the dark energy era in our model). The net effect of these new phenomena is conveniently summarized by the effective growth index.
The effective growth index expresses the linear growth rate in terms of the fractional matter density [39]:
where is the actual matter fraction including the effects of fading. In our model we find for compared to the standard value in CDM. Since is smaller than 1 at late times, the larger value of means that structure grows more slowly in our model than CDM. DESI’s measured growth index gives [40] is consistent with both values although the central value although the central value slightly prefers a suppressed growth rate, as in our model.
Observations, such as redshift-space distortions, provide a more direct measurement of the quantity , i.e. the product of the linear growth rate and amplitude of matter fluctuations at computed from the linear power spectrum. We show a plot of from DESI’s ShapeFit analysis [26] in Figure 2 along with theoretical curves for the models studied in this paper. The data points are merely illustrative because they have been obtained with templates assuming a CDM cosmology. They are not used directly in the likelihood. The theoretical curves show that FDS has slightly more structure than CPL and CDM. This is to be contrasted with the slightly slower growth rate mentioned previously. These two facts imply that the extra physical effects (such as fifth forces) in FDS acting through the early history of the universe build a larger amplitude even though the growth rate today is slightly slower. Observations seem to prefer this combination of larger amplitude and slower growth rate which our FDS model is able to accommodate.
Discussion
In this paper, we studied the Fading Dark Sector (FDS) including data from the full shape (FS) of the galaxy power spectrum measured by DESI in their first data release. We found that adding the FS data (without including any SN dataset), raises the significance of our model from 2.5 to 3.4 while keeping the inferred value of stable at around . The FS data also tightens our measurement of , reducing the error bar by a factor of 1.3–1.7. This improvement comes from the sensitivity of the FS analysis to the growth of structure which proceeds differently in our model due to the presence of fifth forces and varying dark matter masses.
The importance of these additional features of the FDS model is most visible when the model is compared to the standard cosmological alternative, i.e. the – parameterization. While the DESI BAO data has a slight preference for – at the level of , the situation is reversed once the FS data is included. This is an indication that – lacks the features necessary to provide a good fit to the growth of structure, even though it has more flexibility to fit the background cosmology by venturing into unphysical regions of parameter space. For the FDS model, the altered growth of structure is a prediction since it is determined by the same parameters that fix the background evolution. From this perspective, FDS passes a non-trivial check since remain unchanged with additional data. Moreover, the improvement in significance of the FDS model with additional data indicates that these features are important for cosmology. This is unlike the – parameterization which is disfavored compared to FDS by the inclusion of additional FS data.
Other models in the literature such as [17, 34] also have dark sector fifth forces. Analyses of these models have provided hints for a dark fifth force consistent in strength with what we find. However, our FDS model includes additional dynamics in the dark sector because of the potential and these dynamics are favored by data at the level of . When including DES-Dovekie data (we expect the same for other SNe data [16] but have not run those with FS analysis), the error bar on is tightened while remaining consistent with previous results. Without the dynamics of dark energy dictated by , it is unclear whether a model would be able to accommodate the SNe data [34]. The value of remains consistent under all cosmological probes predicting a dark fifth force the strength of gravity in the dark sector. Galactic dynamics can offer an independent test of this force.
Recently, studies of U-shaped galactic disk warps have placed stringent constraints on the strength of a dark fifth force bounding [41]. This result is in tension with coupled dark sector resolutions to the DESI-CMB-SNe tension and would rule out our value of . However that paper assumes a core density profile which may be different for our model because of the dark matter decay and the resulting kick velocity which partially flattens the core as we show in an upcoming work [42]. In addition the baryonic activity at the core may contribute to further flattening of the density profiles at the core. The quoted limit is conditional on the halo-profile prescription and in particular its cuspy core and it would be important to check the robustness of this limit to observationally allowed broad cores. This, as well as other warp-modeling systematics, could weaken this bound making it compatible with our model prediction of a long range dark fifth force around of the gravitational force. Resolving this tension would require further scrutiny of the observational and theoretical methods involved.
Our analysis could be improved in various ways. First, we did not make use of the available galaxy bispectrum data but earlier results (in the context of CDM and CPL, see e.g., [27, 43, 44, 45]), show only small deviations when the bispectrum data is included. We believe this will be the case for our model as well so we leave the inclusion of the bispectrum data for the future. Moreover, models of coupled dark sectors have not been validated to the same extent as CDM in the context of the EFTofLSS. That is to say, one could run N-body simulations with a coupled dark sector to test the robustness of the inference process and check whether the cosmology could be reproduced faithfully. Since our model is a small deviation from CDM, we believe that qualitative results along this direction in the context of CDM should carry over to FDS.
In the near future, several datasets will further test the FDS hypothesis. The most immediate is DESI DR2 FS power spectrum which will be a more precise measurement of the growth of structure. Beyond DESI, Euclid [35] and Roman [46] will map galaxy clustering and also measure the growth of structure, providing further tests of the FDS model. Supernova observations will also improve the constraints on the expansion history. These include samples from ZTF [36] and the Vera Rubin Observatory [37] which will extend the number and redshift of observed supernovae. Since our potential parameter is constrained mainly by the late-time expansion, these data will play a decisive role in our understanding of dynamical dark energy.
In summary, cosmological distance measurements provided hints that something beyond CDM is needed, but several models are degenerate under these measurements alone. The full shape of the galaxy power spectrum begins to distinguish between these models. FDS is a model that contains two additional parameters which can be determined from distance measurements and these same inferred values turn out to account for the growth of structure outperforming the – parameterization by for the same number of parameters. When SNe data is included, the model reaches a preference over CDM and is, to our knowledge, the only model in the literature with this high significance, in addition to being motivated from string theoretic considerations. If upcoming data confirm this departure from CDM, the dynamics of the dark sector would imply the presence of a fifth force that cosmology and galactic surveys can measure.
Acknowledgements
We would like to thank Mustapha Ishak-Boushaki for encouraging us to do the full-shape analysis and for helpful conversations. We would also like to thank the SCGP for the 2026 summer workshop in physics where this work was initiated.
The simulations in this paper were performed on the FASRC Cannon cluster supported by the FAS Division of Science Research Computing Group at Harvard University. The implementation of the FDS model is available at dhwu14/CLASS-PT-exp. We used OpenAI ChatGPT-6 to assist with coding and interfacing CLASS-PT with DESI DR1 FS data, which have all been checked and verified by the authors. We take full responsibility for the results presented in this paper.
The work of CV and DW is supported in part by a grant from the Simons Foundation (602883, CV) and the DellaPietra Foundation.
References
Effective field theory for the fading dark sector
In this appendix, we derive the perturbative equations associated to the fading dark sector. The background and linear perturbation equations of motion have been derived for coupled dark sectors in [47, 48]. The fading dark sector has been first implemented in CLASS in [9] and subsequently studied using recent cosmological data [16]. Here, we focus on the higher-order corrections coming from the one-loop corrections computed using the EFTofLSS for the fading dark sector model. The computations are implemented as a modification of CLASS-PT [18]. We denote the fading dark matter species by in order to avoid confusion with the potential parameter (appearing as ). We denote baryons by . Additionally, we denote , , primes as derivatives with respect to conformal time, and introduce for convenience. We also work in Planck units with . We also adopt the Fourier normalization as . The perturbative expansion is done in the pressureless regime with metric perturbations treated at the sub-horizon Newtonian order. The scalar response in the nonlinear kernels is quasi-static and linear in the scalar fluctuation.
We start with the scalar field background equation, which for the fading dark sector takes on the following form
The equation of motion for the linear perturbation of the scalar field is then
where with , denotes the metric perturbation in synchronous gauge and the effective scalar mass is 4
The dark matter density behaves as .
For and , we work in the sub-horizon, pressureless, quasi-static (QS) regime where the scalar perturbation reduces to
The scalar field therefore produces an attractive force acting on dark matter, with the physical fifth-force enhancement being , as denoted in the main text in eq. (3).
Let us next derive the perturbation equations for the fading dark matter and baryon fluids in Newtonian gauge. The number current for each conserved species obeys where represents the species index. In the non-relativistic limit, with and , we arrive at the continuity equation [49]
where . The standard linear relativistic equations for baryons and dark matter moving under the Newtonian potential can be found in [49]. The Newtonian gravitational potential obeys and, to first order in , the spacetime metric is
For the coupled dark matter, differentiating gives an additional velocity drag term and the scalar acceleration. Hence, the Euler equations in the limit and deep in the matter domination era can be approximated as
where the sub-horizon gravitational potential obeys
Further taking the divergence of the Euler equations, we obtain
where . Notice that (A10) contains no pressure and no tight coupling between photons and baryons appearing in the usual Newtonian gauge [49]. For the pressure term, we can compare with the gravitational potential which upon a Fourier transformation is
The (baryonic) Jeans scale can be deduced when we balance the pressure and the gravitational term leading to .5 Hence, in the linear and quasi-linear regime of interest to the EFTofLSS with , the pressure term becomes negligible. On the other hand, within the photon baryon coupling term, the denotes the Thomson scattering rate per unit conformal time while denotes the conversion of photon momentum exchange into acceleration of baryons. In the tight-coupling era before recombination (), the term is important as the coupling is large , even though . However, shortly after recombination, the free electron abundance decreases sharply, leading to the Thomson scattering rate to decrease and photons no longer exert a strong drag on the baryon fluid. We can compare this effect to the Hubble friction term
where denotes the ionized fraction. Deep in the matter domination era, such that the above ratio becomes which implies that the coupling term becomes negligible compared to its amplitude near recombination.
When the large-scale flow is irrotational , we consequently have . Setting as the velocity momentum, as the density momentum, and , the Fourier transform of the perturbation equations become
where we have introduced
The time dependence and the -dependence of are shown in Figure 3 where in the regime relevant to the DESI DR1 FS measurements, remains a good approximation.
We can further transform these perturbative equations by changing to and . The continuity equation becomes
To convert the Euler equations, we have
These lead to
where we have made use of .
Perturbation kernels and matter loops
To solve the above equations of motion, we can expand both baryon and dark matter by a common Gaussian initial field (see e.g., [31] and references therein)
where denotes the Dirac delta function. At first order, and are the density and normalized velocity-divergence growth functions and satisfy the following relations
where
This is the exact linear operator on the non-linear quasi-static kernels. Unlike the Einstein–de Sitter kernels, the quasi-static kernels will generally depend on due to . At linear order, a mode at evolves independently
Thus, controls how the density perturbation evolves while determines how the velocity divergence evolves. In fact, these are correlated
Furthermore, we can specify the initial condition on these kernels as
At higher orders, this same linear operator acts on each kernel, while lower-order products become inhomogeneous sources. Defining , we obtain the time-dependent recursion relation for the fading dark sector model
where the source terms for are collected as
with denoting the complement of in the set and such that the kernels are symmetrized.
To compute their contributions to the matter power spectrum, let us also introduce the total and relative contributions as
where , , and consequently . With these, the one-loop matter power spectrum can then be computed as
where is the power spectrum of the initial field we used to seed our perturbation kernels. At the observed time , let and define the linear total matter perturbation as such that the associated linear power spectrum satisfies
Furthermore, we introduce the reduced kernels as
where we will use them for deducing the relevant kernels appearing in the galaxy power spectrum.
The one-loop redshift-space galaxy power spectrum
The galaxy overdensity is not determined solely by the linear matter density perturbation. In the effective field theory approach, the tracer overdensity is expanded in scalar operators constructed from the long-wavelength matter and velocity field, with coefficients absorbing unresolved astrophysical formation processes. We adopt the two-fluid Eulerian basis prescription carried out in [17, 50, 51]
where , denote the total and relative matter perturbations, respectively, ’s denote bias parameters in the theory [52, 53], indicate renormalized bias operators, indicates the stochastic field, is associated with the difference between density and velocity tidal evolutions, and denotes the trace-free tidal tensor.
Similar to the density fields, we can define the real-space tracer density kernels and velocity divergence kernels of galaxies by
where . We assume that galaxies co-move with the total matter fluid, and neglect independent relative velocity bias contributions. This assumption is further motivated by finding among the sampled points in our MCMC simulations, suggesting the correction to the redshift-space power spectrum is at the sub-percent level. With this assumption, we have
where no independent relative velocity galaxy bias parameters is sampled. Inserting into (A35), we obtain
where , , , and
Galaxies observed by DESI are measured via their angular positions and spectroscopic redshifts. The galaxy positions are hence shifted along the line-of-sight by their peculiar velocities. To compare with the full-shape data from DESI, we need to map the previous real-space kernels to the redshift-space. In the plane-parallel limit, the redshift-space position is [54, 55]
where is the comoving real-space position, , and is the conformal time. Conservation of galaxy number under this change of coordinates yields
Upon a Fourier transform and removing the homogeneous contribution yields
with and . Assuming an irrotational large-scale velocity , we have . The redshift-space kernels can then be defined by
Then, plugging (A46), (A36) and chugging (A37) into (A45) while expanding the exponential, we obtain
where we have used
The above one-loop convention removes the independent local quadratic-bias contribution from . We can see this simplification by introducing the standard local quadratic coordinate . The projection acts on the third-order terms
where enters via . Therefore, the third-order redshift-space kernel contributing to is
where we have suppressed the time dependence.
From here, we can construct the perturbative contributions to the galaxy power spectrum
where .
Counterterms and stochastic contributions
In addition to the one-loop contributions, the power spectrum also include analytic counterterms for unresolved short-distance dynamics and a stochastic contribution associated with discrete tracers
where are the redshift-space EFT coefficients while are the native stochastic amplitudes. Given a galaxy sample, we analytically marginalize four Gaussian-prior variables , as was done in [13], with their mapping to the nuisance parameters as
where is the effective linear growth rate, and are tracer-specific shot-noise normalizations, satellite fraction, and stochastic velocity scale used in the likelihood.
Anisotropic IR resummation
Long-wavelength displacements can smear the BAO features of the galaxy power spectrum. To resum these motions, we can split the linear matter power spectrum into a smooth component and an oscillatory component ,
| Type | Parameter | Prior | Units |
| nuisance (sampled) | — | ||
| — | |||
| — | |||
| nuisance (analytically marginalized) | |||
| — | |||
| — |
Table IV. The nuisance parameters and priors used for each DESI full-shape tracer. Here, denotes a uniform prior between and , while denotes a Gaussian prior with mean and variance . The bias combinations , , and are sampled explicitly in the MCMC chains. The counterterm parameters and and stochastic amplitudes and enter linearly and are analytically marginalized in the likelihood. The same nuisance prescription is applied independently to each full-shape redshift/tracer bin.
Then, the displacement variances associated to the density and line-of-sight displacement responses are
where is the characteristic BAO separation, is the IR/UV separation wavenumber, the numerical lower limit of the displacement integration, the spherical Bessel function of order , , and . The anisotropic damping exponent is
with . In the scale-independent case, we can introduce a constant such that our variances can be identified with introduced in [54] where , , , , reproducing its known behavior in this limit.
At one loop, the IR prescription associated to one wiggle in the power spectrum is [56]
where . Thus, includes two permutations of one smooth and one wiggly power spectrum while includes wiggle insertions in its external or internal linear power factor. The EFT counterterm uses the damped combination while the stochastic term is unaffected. Hence, the final galaxy power spectrum is
Alcock–Paczynski mapping
To compare a theory with different expansion history to measurements in the coordinates adapted by observations, we apply the Alcock–Paczynski (AP) transformation [57]. At redshift , we adapt the notation used in [58]
where is the transverse comoving distance and the subscript refers to the fiducial cosmology used in constructing the survey coordinates. Let be the wavenumber and line-of-sight angle in the fiducial cosmology, then the theory coordinates and volume Jacobian are [18]
Using this last fiducial power spectrum, the observed-coordinate multipoles are
where and is the Legendre polynomial. The behavior of the matter power spectrum along with the multipole moments of the redshift-space galaxy power spectrum for the FDS model is shown in Figure 4.
Figure 4. The best-fit power spectrum from CMB + DESI1.5 as an illustrative example of corrections to the matter power spectrum in the quasi-linear regime .
Mapping to DESI multipoles
The galaxy-bias expansion (A36) was carried out using the Eulerian coefficients ’s. However, the bias parameters relevant to the DESI full-shape analysis is done with Lagrangian bias coefficients. In Lagrangian perturbation theory, galaxies associated with an initial position has the following expansion
where this is related to the Eulerian positions as . For a conserved, co-moving tracer population, galaxy number conservation implies
Then, the mapping between the Eulerian and Lagrangian bases at one-loop order becomes [52, 59]
In this basis, the DESI nuisance parameters then appear explicitly. The simulation results for the sampled quadratic bias parameters are shown in Figure 5.

Figure 5. Comparison of the quadratic galaxy bias parameters among CDM, the CPL parameterization, and the FDS model where the cosmology is fitted to CMB+DESI1.5. Here, and correspond to the quadratic local and tidal Lagrangian bias parameters. The superscript runs from 1 to 6 indicating the galaxy bias parameter for BGS, LRG1, LRG2, LRG3, ELG2, and QSO, respectively, in the DESI DR1 database.
Additionally, to make contact with observed galaxy surveys, we need to select an angular mask and redshift that is measured. Let index an observed power-spectrum multipole and wavenumber bin, and index the ordered pre-window theory entries. Let denote the multipole moments and the wavenumber bins , then the window matrix maps the fiducial coordinate theory into the observed data space via
where denotes cosmological parameters and denotes the galaxy-bias, counterterm, and stochastic nuisance parameters. The actual window differs between observations. The relevant DESI window matrix can be found in [60]. Hence, combining eqs. (78), (A71) and (B6), we can specify the observable one-loop redshift-space multipoles adequate for fitting to the DESI data appropriate for the fading dark sector.
References
- [1]A. G. Adame et al. (DESI), DESI 2024 VI: cosmological constraints from the measurements of baryon acoustic oscillations, JCAP 02, 021, arXiv:2404.03002 [astro-ph.CO].DOI
- [2]M. Abdul Karim et al. (DESI), DESI DR2 results. II. Measurements of baryon acoustic oscillations and cosmological constraints, Phys. Rev. D 112, 083515 (2025), arXiv:2503.14738 [astro-ph.CO].
- [3]D. Brout et al., The Pantheon+ Analysis: Cosmological Constraints, Astrophys. J. 938, 110 (2022), arXiv:2202.04077 [astro-ph.CO].DOI
- [4]D. Rubin et al., Union Through UNITY: Cosmology with 2,000 SNe Using a Unified Bayesian Framework, Astrophys. J. 986, 231 (2025), arXiv:2311.12098 [astro-ph-CO].arxiv.org/abs/2311.12098
- [5]T. M. C. Abbott et al. (DES), The Dark Energy Survey: Cosmology Results with ∼1500 New High-redshift Type Ia Supernovae Using the Full 5 yr Data Set, Astrophys. J. Lett. 973, L14 (2024), arXiv:2401.02929 [astro-ph.CO].arxiv.org/abs/2401.02929
- [6]B. Popovic et al. (DES), The Dark Energy Survey supernova program: a reanalysis of cosmology results and evidence for evolving dark energy with an updated Type Ia supernova calibration, Mon. Not. Roy. Astron. Soc. 548, stag632 (2026), arXiv:2511.07517 [astro-ph.CO].DOI
- [7]K. Lodha et al. (DESI), Extended dark energy analysis using DESI DR2 BAO measurements, Phys. Rev. D 112, 083511 (2025), arXiv:2503.14743 [astro-ph-CO].arxiv.org/abs/2503.14743
- [8]D. Shlivko, P. J. Steinhardt, and C. L. Steinhardt, Optimal parameterizations for observational constraints on thawing dark energy 10.1088/1475-7516/2025/06/054 (2025), arXiv:2504.02028 [astro-ph.CO].DOI
- [9]P. Agrawal, G. Obied, and C. Vafa, H₀ tension, swampland conjectures, and the epoch of fading dark matter, Phys. Rev. D 103, 043523 (2021), arXiv:1906.08261 [astro-ph.CO].DOI
- [10]J. Khoury, M.-X. Lin, and M. Trodden, Apparent w<-1 and a Lower S8 from Dark Axion and Dark Baryons Interactions, Phys. Rev. Lett. 135, 181001 (2025), arXiv:2503.16415 [astro-ph.CO].DOI
- [11]H. Bagherian, M. Reece, L.-T. Wang, and H. Xiao, A Natural Dark Matter-Dark Energy Interaction from Axion-Monopole Dynamics, (2026), arXiv:2610.27059 [hep-ph].arxiv.org/abs/2610.27059
- [12]M. Ishak et al., Modified gravity constraints from the full shape modeling of clustering measurements from DESI 2024, JCAP 09, 053, arXiv:2411.12026 [astro-ph.CO].DOI
- [13]A. G. Adame et al. (DESI), DESI 2024 VII: cosmological constraints from the full-shape modeling of clustering measurements, JCAP 07, 028, arXiv:2411.12022 [astro-ph.CO].DOI
- [14]M. Montero, C. Vafa, and I. Valenzuela, The dark dimension and the Swampland, JHEP 02, 022, arXiv:2205.12293 [hep-th].DOI
- [15]C. Vafa, The Phenomenology of the Dark Dimension and Non-Supersymmetric Strings, (2026), arXiv:2610.04909 [hep-th].arxiv.org/abs/2610.04909
- [16]A. Bedroya, G. Obied, C. Vafa, and D. H. Wu, Evolving Dark Sector and the Dark Dimension Scenario, (2025), arXiv:2507.03090 [astro-ph.CO].arxiv.org/abs/2507.03090
- [17]S. Bottaro, E. Castorina, M. Costa, D. Redigolo, and E. Salvioni, Unveiling Dark Forces with Measurements of the Large Scale Structure of the Universe, Phys. Rev. Lett. 132, 201002 (2024), arXiv:2309.11496 [astro-ph.CO].DOI
- [18]A. Chudaykin, M. M. Ivanov, O. H. E. Philcox, and M. Simonović, Nonlinear perturbation theory extension of the Boltzmann code CLASS, Phys. Rev. D 102, 063533 (2020), arXiv:2004.10607 [astro-ph.CO].
- [19]J. Torrado and A. Lewis, Cobaya: Code for Bayesian Analysis of hierarchical physical models, JCAP 05, 057, arXiv:2005.05290 [astro-ph.IM].
- [20]N. Aghanim et al. (Planck), Planck 2018 results. V. CMB power spectra and likelihoods, Astron. Astrophys. 641, A5 (2020), arXiv:1907.12875 [astro-ph.CO].DOI
- [21]Y. Akrami et al. (Planck), Planck intermediate results. LVII. Joint Planck LFI and HFI data processing, Astron. Astrophys. 643, A42 (2020), arXiv:2007.04997 [astro-ph.CO].DOI
- [22]E. Rosenberg, S. Gratton, and G. Efstathiou, CMB power spectra and cosmological parameters from Planck PR4 with CamSpec, Mon. Not. Roy. Astron. Soc. 517, 4620 (2022), arXiv:2205.10869 [astro-ph.CO].DOI
- [23]M. S. Madhavacheril et al. (ACT), The Atacama Cosmology Telescope: DR6 Gravitational Lensing Map and Cosmological Parameters, Astrophys. J. 962, 113 (2024), arXiv:2304.05203 [astro-ph.CO].DOI
- [24]F. J. Q. Qu et al. (ACT), The Atacama Cosmology Telescope: A Measurement of the DR6 CMB Lensing Power Spectrum and Its Implications for Structure Growth, Astrophys. J. 962, 112 (2024), arXiv:2304.05202 [astro-ph.CO].
- [25]J. Carron, M. Mirmelstein, and A. Lewis, CMB lensing from Planck PR4 maps, JCAP 09, 039, arXiv:2206.07773 [astro-ph.CO].DOI
- [26]A. G. Adame et al. (DESI), DESI 2024 V: Full-Shape galaxy clustering from galaxies and quasars, JCAP 09, 008, [Erratum: JCAP 02, E02 (2026)], arXiv:2411.12021 [astro-ph.CO].DOI
- [27]A. Chudaykin, M. M. Ivanov, and O. H. E. Philcox, Reanalyzing DESI DR1. I. ΛCDM constraints from the power spectrum and bispectrum, Phys. Rev. D 113, 063502 (2026), arXiv:2507.13433 [astro-ph-CO].arxiv.org/abs/2507.13433
- [28]D. Forero-Sánchez et al., Cosmological constraints from a joint DESI DR1 Full-Shape and DR2 BAO, JCAP 06, 043, arXiv:2602.18761 [astro-ph-CO].
- [29]T. J. Hoyt, D. Rubin, G. Aldering, S. Perlmutter, A. Cuceu, and R. Gupta, Union3.1: Reducing Systematics in Supernova Cosmology with Self-consistent Measurements of Host Galaxy Properties for 2000 Type Ia Supernovae, (2026), arXiv:2601.19424 [astro-ph.CO].arxiv.org/abs/2601.19424
- [30]D. Rubin, T. Hoyt, G. Aldering, and S. Perlmutter, Banana Split: Improved Cosmological Constraints with Two Light-curve-shape and Color Populations Using Union3.1+UNITY1.8, Astrophys. J. 1012, 6 (2026), arXiv:2601.19854 [astro-ph.CO].arxiv.org/abs/2601.19854
- [31]F. Bernardeau, S. Colombi, E. Gaztanaga, and R. Scoccimarro, Large scale structure of the universe and cosmological perturbation theory, Phys. Rept. 367, 1 (2002), arXiv:astro-ph/0112551.DOI
- [32]Y. Donath and L. Senatore, Biased Tracers in Redshift Space in the EFTofLSS with exact time dependence, JCAP 10, 039, arXiv:2005.04805 [astro-ph.CO].
- [33]G. D’Amico, Y. Donath, L. Senatore, and P. Zhang, Limits on clustering and smooth quintessence from the EFTofLSS, JCAP 03, 032, arXiv:2012.07554 [astro-ph.CO].DOI
- [34]A. Clini, P. Zhang, E. Castorina, and M. Archidiacono, The Equivalence Principle in the Dark Sector in light of DESI, (2026), arXiv:2609.17423 [astro-ph.CO].arxiv.org/abs/2609.17423
- [35]Y. Mellier et al. (Euclid), Euclid - I. Overview of the Euclid mission, Astron. Astrophys. 697, A1 (2025), arXiv:2405.13491 [astro-ph.CO].DOI
- [36]M. Rigault et al., ZTF SN Ia DR2: Overview, Astron. Astrophys. 694, A1 (2025), arXiv:2409.04346 [astro-ph-CO].arxiv.org/abs/2409.04346
- [37]R. Mandelbaum et al. (LSST Dark Energy Science Collaboration), The LSST Dark Energy Science Collaboration (DESC) Science Requirements Document 10.2172/1471560 (2018), arXiv:1809.01669 [astro-ph.CO].arxiv.org/abs/1809.01669
- [38]R. Kessler, R. Hounsell, B. Joshi, D. Rubin, M. Sako, R. Chen, V. Miranda, and B. M. Rose, Cosmology Constraints from Type Ia Supernova Simulations of the Nancy Grace Roman Space Telescope Strategy Recommended by the High-Latitude Time Domain Survey Definition Committee, Astrophys. J. 993, 116 (2025), arXiv:2506.04402 [astro-ph.CO].arxiv.org/abs/2506.04402
- [39]E. V. Linder, Cosmic growth history and expansion history, Phys. Rev. D 72, 043529 (2005), arXiv:astro-ph/0507263.DOI
- [40]R. J. Turner et al., The DESI DR1 Peculiar Velocity Survey: growth rate measurements from galaxy and momentum correlation functions, Open J. Astrophys. 9, 164736 (2026), arXiv:2512.03230 [astro-ph.CO].DOI
- [41]H. Desmond and P. G. Ferreira, Galactic warps constrain strain equivalence-principle violation in the dark sector, (2026), arXiv:2610.24618 [astro-ph.CO].arxiv.org/abs/2610.24618
- [42]D. Fulcher, G. Obied, C. Vafa, D. H. Wu, and O. Zier, Galaxies and Decaying Tower of Dark Matter, to Appear.
- [43]D. Forero-Sánchez et al., Cosmological constraints from the DESI DR1 Bispectrum Full-Shape and DR2 BAO, (2026), arXiv:2606.23936 [astro-ph.CO].
- [44]A. Chudaykin, M. M. Ivanov, and O. H. E. Philcox, Reanalyzing DESI DR1. II. Constraints on dark energy, spatial curvature, and neutrino masses, Phys. Rev. D 113, 123506 (2026), arXiv:2511.20757 [astro-ph.CO].arxiv.org/abs/2511.20757
- [45]S. Novell-Masot et al., Cosmological constraints from the DESI DR1 joint power spectrum and bispectrum analysis, JCAP 09, 145, arXiv:2603.19356 [astro-ph.CO].
- [46]Y. Wang et al., The High Latitude Spectroscopic Survey on the Nancy Grace Roman Space Telescope, Astrophys. J. 928, 1 (2022), arXiv:2110.01829 [astro-ph.CO].DOI
- [47]L. Amendola, Perturbations in a coupled scalar field cosmology, Mon. Not. Roy. Astron. Soc. 312, 521 (2000), arXiv:astro-ph/9906073.DOI
- [48]V. Miranda, M. Carrillo González, E. Krause, and M. Trodden, Finding structure in the dark: coupled dark energy, weak lensing, and the mildly non-linear regime, Phys. Rev. D 97, 063511 (2018), arXiv:1707.05694 [astro-ph.CO].DOI
- [49]C.-P. Ma and E. Bertschinger, Cosmological perturbation theory in the synchronous and conformal Newtonian gauges, Astrophys. J. 455, 7 (1995), arXiv:astro-ph/9506072.DOI
- [50]F. Schmidt, Effect of relative velocity and density perturbations between baryons and dark matter on the clustering of galaxies, Phys. Rev. D 94, 063508 (2016), arXiv:1602.09059 [astro-ph.CO].DOI
- [51]S.-F. Chen, E. Castorina, and M. White, Biased Tracers of Two Fluids in the Lagrangian Picture, JCAP 06, 006, arXiv:1903.00437 [astro-ph-CO].DOI
- [52]V. Desjacques, D. Jeong, and F. Schmidt, Large-Scale Galaxy Bias, Phys. Rept. 733, 1 (2018), arXiv:1611.09787 [astro-ph.CO].DOI
- [53]P. McDonald and A. Roy, Clustering of dark matter tracers: generalizing bias for the coming era of precision lss, Journal of Cosmology and Astroparticle Physics 2009 (08), 020.DOI
- [54]M. M. Ivanov and S. Sibiryakov, Infrared Resummation for Biased Tracers in Redshift Space, JCAP 07, 053, arXiv:1804.05080 [astro-ph.CO].DOI
- [55]M. Lewandowski, L. Senatore, F. Prada, C. Zhao, and C.-H. Chuang, EFT of large scale structures in redshift space, Phys. Rev. D 97, 063526 (2018), arXiv:1512.06831 [astro-ph.CO].DOI
- [56]D. Blas, M. Garny, M. M. Ivanov, and S. Sibiryakov, Time-Sliced Perturbation Theory II: Baryon Acoustic Oscillations and Infrared Resummation, JCAP 07, 028, arXiv:1605.02149 [astro-ph.CO].DOI
- [57]C. Alcock and B. Paczynski, An evolution free test for non-zero cosmological constant, Nature 281, 358 (1979).DOI
- [58]R. Neveux et al. (eBOSS), Combined full shape analysis of BOSS galaxies and eBOSS quasars using an iterative emulator, Mon. Not. Roy. Astron. Soc. 516, 1910 (2022), arXiv:2201.04679 [astro-ph.CO].DOI
- [59]S.-F. Chen, Z. Vlah, and M. White, Consistent Modeling of Velocity Statistics and Redshift-Space Distortions in One-Loop Perturbation Theory, JCAP 07, 062, arXiv:2005.00523 [astro-ph.CO].DOI
- [60]A. G. Adame et al. (DESI), DESI 2024 II: sample definitions, characteristics, and two-point clustering statistics, JCAP 07, 017, arXiv:2411.12020 [astro-ph.CO].DOI
