Interacting dark energy constraints from the full-shape analyses of BOSS DR12 and DES Year 3 measurements 2026 Interacting dark energy constraints from the full-shape analyses of BOSS DR12 and DES Year 3 measurements – References M. Tsedrik, S. Lee, K. Markovic, P. Carrilho, A. Pourtsidou, C. Moretti, B. Bose, E. Huff, A. Robertson, P. L. Taylor, J. Zuntz Abstract Dark Scattering (DS) is an interacting dark energy model characterised by pure momentum exchange between dark energy and dark matter. It is phenomenologically interesting because it is unconstrained by CMB data and can alleviate the S 8 S_{8} tension. We derive constraints on cosmological and DS parameters using three two-point correlation functions (3 × \times 2pt) from the Dark Energy Survey third year data release (DES Y3). We then add information from the multipoles of the galaxy power spectrum combined with Baryonic Acoustic Oscillation (BAO) measurements using the twelfth data release of the Baryon Oscillation Spectroscopic Survey (BOSS DR12) and external BAO measurements. We compare results from the direct combination of the probes with the joint posterior distribution calculated with a normalising flow approach. Additionally, we run a CMB analysis with the Planck Public Release 4 (PR4) for comparison of the cosmological constraints. Overall, we find that the combination of probes allows minimising the projection effects and improves constraints without the need to include CMB information. It brings the marginalised posterior maxima closer to the corresponding best-fit values and weakens the sensitivity to the priors of the spectroscopic modelling nuisance parameters. These findings are highly relevant in light of forthcoming data of surveys like DESI, Euclid, and Rubin. 1 Introduction Data from the latest generation of cosmological surveys, such as DESI , Euclid and Rubin will shed light on the nature of the dark sector (dark energy and dark matter) by providing Large Scale Structure (LSS) information with unprecedented precision. These surveys will allow us to draw unambiguous conclusions about the current cosmological tensions, for example the so-called S_{8} tension: a \sim 2\sigma disagreement in the amplitude of matter fluctuations between the one measured in LSS experiments ( Heymans et al. 2021 ; Abbott et al. 2022a ) and the one measured with the CMB and propagated to later times using the standard cosmological model ( Planck Collaboration et al. 2020 ) . Dark Scattering ( Simpson 2010 ; Pourtsidou et al. 2013 , DS, ) is one of the potential ‘new physics’ solutions for the S_{8} tension. It is a subclass of models, in which dark energy is coupled to cold dark matter via a momentum-exchange interaction. Theoretical modelling for observables in the DS model has been extensively tested on simulations and applied to data ( Pourtsidou & Tram 2016 ; Bose et al. 2018 ; Bose & Taruya 2018 ; Carrilho et al. 2021 ; Carrilho et al. 2022 ; Spurio Mancini & Pourtsidou 2022 ; Linton et al. 2022 ; Tsedrik et al. 2023 ; Carrilho et al. 2023 ; Carrion et al. 2024 ) . In this work, we provide constraints on DS cosmology with DES Y3, a photometric dataset consisting of three two-point correlation functions: cosmic shear, galaxy clustering, and the cross-correlation of source galaxy shear with lens galaxy positions. We then showcase the power of combining it with a full-shape analysis using data from BOSS DR12, a spectroscopic dataset consisting of three power spectrum multipoles and BAOs. For reference, we also provide constraints on the standard cosmological model, \Lambda CDM. 2 Model Dark Scattering is an interacting dark energy model, in which dark energy and cold dark matter exchange momentum via elastic scattering ( Simpson 2010 ; Pourtsidou et al. 2013 ; Skordis et al. 2015 ) . No energy transfer between those two species is involved in the scattering process, hence leaving the expansion history unaffected by the interaction. Additionally, this form of coupling only weakly affects the CMB and is in remarkable agreement with Planck measurements ( Pourtsidou & Tram 2016 ) . The DS model is characterised by two extra parameters with respect to \Lambda CDM: the equation of state parameter, w , and the interaction parameter, \xi . The interaction parameter is defined as the ratio between the scattering cross section and the dark matter mass. For simplicity we take the dark energy equation of state parameter to be constant. For the case of constant w , the Friedmann equation in a flat universe is given by H^{2}=H_{0}^{2}\left(\Omega_{\rm m}~a^{-3}+\Omega_{\rm DE}~a^{-3(1+w)}\right)\,, (1) where H=\dot{a}/a is the Hubble rate with the scale factor a=1/(1+z) , the dot denotes a derivative with respect to cosmic time, and H_{0} is the value of the expansion rate today; \Omega_{\rm m} is the matter density parameter today and \Omega_{\rm DE}=1-\Omega_{\rm m} is the dark energy density parameter. The total matter fraction includes contributions from cold dark matter ( \Omega_{\rm c} ), baryons ( \Omega_{\rm b} ) and massive neutrinos ( \Omega_{\nu} ). The interaction between dark matter particles and the dark energy fluid manifests itself through an additional friction or drag in the movement of dark matter particles. Under the assumption of the dark energy speed of sound being equal to the speed of light, we can write the linearised Euler equation as a~\partial_{a}\Theta+\left(2+(1+w)~\xi~\frac{\rho_{\mathrm{DE}}}{H}+\frac{a~\partial_{a}H}{H}\right)\Theta+\frac{\nabla^{2}\Phi}{a^{2}H^{2}}=0\,. (2) Here \partial_{a} is the derivative with respect to the scale factor, \Theta=\theta_{\rm c}/aH with \theta_{\mathrm{c}} the dark matter velocity divergence, \Phi is the corresponding gravitational potential, and \rho_{\rm DE} is the dark energy density. We evolve baryons and dark matter as a single species. From Equation 1 and Equation 2 we find the linear growth factor, D , and the growth rate f={\rm d}\ln D/{\rm d}\ln a by solving the linearised growth equation. In Equation 2 , we see that if w\approx-1 the drag/friction-term becomes negligible and the scattering parameter \xi becomes unconstrained. For this reason, in our analysis we vary the combined parameter A_{\rm ds}=(1+w)\xi , with a well-defined \Lambda CDM-limit of A_{\rm ds}=0 b GeV -1 and w=-1 . Since \xi is always positive (or zero), the sign of A_{\rm ds} and (1+w) has to be the same. The parameter space region that can resolve the S_{8} tension corresponds to A_{\rm ds}>0 and w>-1 . In this case, the linear growth factor decreases with respect to the \Lambda CDM scenario at low redshifts. This leads to a scale-independent suppression of the power spectrum at large scales in DS. However, on smaller scales, in collapsed objects, dark matter particles experience additional friction, which leads to a scale-dependent enhancement of structure growth. Such nonlinear behaviour is modelled within the halo model reaction framework ( Cataneo et al. 2019 ; Bose et al. 2020 ) , which was tested against the DS N-body simulations presented in Baldi & Simpson 2015 ; Baldi & Simpson 2017 ( Palma & Candlish 2023 , see also) . 3 Data and methods We consider the publicly released measurements of the three two-point correlation functions (3 \times 2pt) from DES Y3 ( Abbott et al. 2022b ; Abbott et al. 2023b ) . They contain cosmic shear, galaxy clustering and galaxy-galaxy lensing information from sources in 4 redshift-bins and lenses from the first 4 redshift-bins of the MagLim sample ( Porredon et al. 2021 ) . The corresponding covariance matrix is obtained analytically as described and validated in Friedrich et al. 2021 . We use the same scale-cuts as in the DES Y3 \Lambda CDM baseline analysis, since we apply the reaction approach, a robust prescription of nonlinearities for DS ( Carrilho et al. 2022 ) . The study of these scale-cuts is presented in Krause et al. 2021 and justifies the fact that we do not model effects of baryonic feedback on small scales. We use the official DES-pipeline, CosmoSIS , in which we substitute the computation of linear and nonlinear matter power spectra by CAMB ( Lewis & Challinor 2011 ) with an emulator, DS-emulator . This emulator is discussed in detail in Carrion et al. 2024 , where it is also applied to the shear data of the Kilo-Degree Survey ( Joachimi et al. 2021 ; Asgari et al. 2021 , KiDS-1000, ) . For DS (and \Lambda CDM being a DS limit with w=-1 and A_{\rm ds}=0 ), the DS-emulator provides linear power spectra trained on CLASS ( Lesgourgues 2011 ; Blas et al. 2011 ) outputs, and nonlinear power spectra trained on ReACT outputs. Both models are trained with CosmoPower ( Spurio Mancini et al. 2022 ) . The emulation error is under 1\% at all scales of interest. In the following bullet-points we summarise the difference with respect to the analysis set-up from Abbott et al. 2022b . • Modelling of the nonlinear power spectrum is done with HMCode2020 ( Mead et al. 2021 ) for \Lambda CDM motivated by a better agreement with numerical simulations ( Abbott et al. 2023a ) . • The priors are identical to the ones summarised in Table 1 of Abbott et al. 2022b , except of \{\Omega_{\rm b},h,w\} (for instance, w\in[-1.3,-0.7] ). For these parameters we have narrower flat priors, explained by the parameter-range in the training set of the DS-emulator . Additionally, the prior on the interaction parameter is A_{\rm ds}\in[-30,30] b GeV -1 , motivated by the constraints found in Carrilho et al. 2023 ; Carrion et al. 2024 . • We fix the total mass of one massive and two massless neutrinos to M_{\nu}=0.06 eV. This analysis choice is motivated by the weak constraints from Stage III surveys ( Abbott et al. 2022b ; Moretti et al. 2023 ) . Moreover, varying M_{\nu} introduces additional prior volume effects ( Porredon et al. 2022 ) . • We do not include information from the small-scale shear ratio ( Sánchez et al. 2022 ) motivated by Aricò et al. 2023 . Table 1: Mean values and 68% c.l. values for DS and \Lambda CDM with fixed one massive neutrino with M_{\nu}=0.06~{\rm eV} for the different probes and their combination considered in this work. We show the MAP values in parentheses, and include derived constraints on \sigma_{8} . We report the number of data-points N_{\rm data} , the number of parameters N_{\rm par} varied in the MCMC and number of analytically marginalised parameters in parentheses, \chi^{2} and \chi^{2}_{\rm prior} computed with MAP-values and log-evidence \log{\mathcal{Z}} from the sampler. Unconstrained parameters are denoted by a dash-line. The BBN prior is applied in the FS+BAO and joint analyses. 3\times 2 pt from DES Y3 FS + BAO from BOSS DR12 and external BAOs joint N_{\rm data} 462 456+(8+4) 930 \Lambda CDM DS \Lambda CDM DS \Lambda CDM DS \Omega_{\rm m} 0.305\pm 0.024 (0.317) 0.303\pm 0.029 (0.342) 0.311^{+0.012}_{-0.014} (0.306) 0.297\pm 0.016 (0.309 | 0.311) 0.316\pm 0.011 (0.311) 0.311\pm 0.014 (0.317) S_{8} 0.803\pm 0.018 (0.799) 0.811^{+0.031}_{-0.039} (0.799) 0.746\pm 0.048 (0.842) 0.639^{+0.043}_{-0.063} (0.711 | 0.864) 0.792\pm 0.015 (0.808) 0.790\pm 0.018 (0.814) \sigma_{8} 0.799^{+0.039}_{-0.047} (0.777) 0.811^{+0.057}_{-0.076} (0.748) 0.733\pm 0.050 (0.833) 0.643^{+0.043}_{-0.063} (0.701 | 0.849) 0.773\pm 0.024 (0.794) 0.775^{+0.024}_{-0.028} (0.792) w -1 -1.02^{+0.14}_{-0.19} (-0.81) -1 -1.17^{+0.13}_{-0.08} ( \approx -1 | -0.94) -1 -1.04^{+0.10}_{-0.08} (-0.93) A_{\rm ds} 0 -5^{+13}_{-20} (6) 0 -21^{+16}_{-10} (-22 | 0) 0 -0.2^{+4.6}_{-5.9} (0.0) h – (0.677) – (0.640) 0.681\pm 0.010 (0.676) 0.713^{+0.018}_{-0.024} (0.678 | 0.664) 0.682\pm 0.009 (0.680) 0.689^{+0.016}_{-0.020} (0.668) n_{\rm s} – (0.904) – (0.870) 0.974^{+0.063}_{-0.055} (1.036) 0.945\pm 0.059 (1.022 | 1.049) 0.957^{+0.041}_{-0.047} (0.977) 0.955^{+0.041}_{-0.048} (0.985) N_{\rm par} 27 29 17(+32) 19(+32) 39(+32) 41(+32) \chi^{2} 511.9 509.8 470.03 469.7 | 468.9 513.4+469.4 513.2+468.8 \chi^{2}_{\rm prior} 9.5 9.3 7.99 7.5 | 8.3 19.6 18.9 \log{\mathcal{Z}} 5749.2 \pm 0.2 5748.8 \pm 0.2 -123.3 \pm 0.2 -126.9 \pm 0.2 5630.5 \pm 0.3 5627.8 \pm 0.2 Additionally, we consider the galaxy power spectrum multipoles from BOSS DR12 ( Alam et al. 2015 ; Gil-Marín et al. 2016 ; Beutler et al. 2017 ) . The data consists of two galaxy samples: CMASS and LOWZ with effective redshifts z_{1}=0.38 and z_{3}=0.61 , respectively. Each sample covers two different sky cuts, NGC and SGC. Hence in total, we fit 4 independent sets of multipoles. The multipole measurements and covariance are provided in Philcox & Ivanov 2022 and are obtained with the windowless estimator ( Philcox 2021a ; Philcox 2021b ) . The covariance is based on 2048 ‘MultiDark-Patchy’ mock catalogues ( Kitaura et al. 2016 ; Rodríguez-Torres et al. 2016 ) . Complementarily, we use BAO data from BOSS DR12 ( Philcox & Ivanov 2022 , with a cross-covariance, both provided by ) , as well as pre-reconstruction BAO measurements at low redshift from the 6DF survey ( Beutler et al. 2011 ) at z\approx 0.11 and SDSS DR7 MGS ( Ross et al. 2015 ) at z=0.15 . We also add information from high redshift measurements of the Hubble factor and angular diameter distance from the Ly- \alpha forest auto and cross-correlation with quasars from eBOSS DR16 ( du Mas des Bourboux et al. 2020 ) at z\approx 2.33 . Further in text ‘FS+BAO’ means the full shape (FS) analysis with the BOSS DR12 data with the inclusion of all BAOs mentioned above. For the BOSS DR12 FS analysis we add an EFTofLSS ( Baumann et al. 2012 ; Carrasco et al. 2012 ; Perko et al. 2016 ; Fonseca de la Bella et al. 2017 ; D’Amico et al. 2020 ) modelling module to CosmoSIS . The modelling code is discussed in detail in Moretti et al. 2023 . It was also extensively validated on N-body simulations in \Lambda CDM ( Oddo et al. 2020 ; Oddo et al. 2021 ; Rizzo et al. 2023 ; Tsedrik et al. 2023 ) . This approach allows us to model the power spectrum multipoles in redshift space including mildly nonlinear scales. We use the same set-up of parameters and priors as in Carrilho et al. 2023 . In the FS+BAO analysis the chosen priors on cosmological parameters are very broad and confined only by the \Lambda CDM cosmological parameter ranges of the linear bacco emulator ( Aricò et al. 2021 , trained on CLASS outputs, ) . The only informative prior we impose is a Gaussian baryon (BBN) prior: \omega_{b}\in\mathcal{N}(0.02268,0.00038) ( Aver et al. 2015 ; Cooke et al. 2018 ; Schöneberg et al. 2019 ) . The priors on the DS parameters are specified by w\in[-3,-0.5] and A_{\rm ds}\in[-500,500] b GeV -1 . We compute the growth parameters in \Lambda CDM and DS scenarios using the evogrowthpy code. We sample posterior distributions using Polychord ( Handley et al. 2015 ) with the ‘publication quality’ parameters from Lemos et al. 2023 . For the DS model we increase live_points and num_repeats to 1000 and 100 to guarantee sufficient sampling in the highly non-Gaussian posterior, with the ‘butterfly’-shaped prior in w-A_{\rm ds} . To plot and derive constraints from the posterior distributions we use getdist ( Lewis 2019 ) . While for the best-fit, i.e. the Maximum A Posteriori (MAP) values, we use iminuit ( James & Roos 1975 ) . When the MAP value disagrees with the peak of the marginalised posterior obtained from the chain, it signifies the presence of projection effects ( Gómez-Valent 2022 ; Hadzhiyska et al. 2023 ) . Projection effects arise when a high-dimensional non-Gaussian posterior distribution is compressed in a lower-dimensional parameter space. Projection effects are especially prominent for extended cosmologies, in which beyond- \Lambda CDM parameters are strongly degenerate with cosmological parameters ( Moretti et al. 2023 ; Maus et al. 2024 ) . For a combination of probes with CombineHarvesterFlow we train and combine individual marginalised posterior-distributions in cosmological parameters [\Omega_{\rm m},h,\omega_{\rm b},n_{\rm s},S_{8},(A_{\rm ds},w)] with 14 flows ( Taylor et al. 2024 , for details see) . Additionally, we run a CMB analysis for DS using a modified version of CLASS and Cobaya with the Planck low- \ell TT, the lollipop TE, EE and the hillipop TTTEEE likelihoods for the PR4 data release ( Tristram et al. 2024 ) . We do not include CMB lensing information to show constraints that are as independent as possible from the late-time data from LSS. 4 Results The constraints on cosmological parameters are given in Table 1 with the means, the 68 per cent credibility intervals, and the MAP values. We do not quote the constraints on \omega_{\rm b} as they are BBN-prior dominated for the FS and joint (3 \times 2pt+FS+BAO) analyses. In Figure 1 we show the posterior distributions of cosmological parameters in \Lambda CDM (left panel) and DS (right panel). For DS we omit n_{\rm s} and h because it shows a trend similar to the \Lambda CDM case. Based on \chi^{2} statistics and comparison of the Bayes factors, the performance of both models is very similar, with no clear preference for \Lambda CDM or DS. Figure 1: Marginalised posterior distribution for cosmological parameters in \Lambda CDM (left panel) and DS (right panel). Contours for DES Y3, BOSS DR12 and their combination are shown in blue, orange and green, respectively. Grey lines correspond to CMB constraints from Tristram et al. 2024 without lensing information. Figure 2: Projection effects in S_{8} and the DS parameters. The error bars show the 68% marginalised posteriors, the stars denote the MAP values, the triangles denote the second-best MAP values in FS+BAO with DS. The grey shaded line corresponds to constraints on S_{8} from CMB data ( Tristram et al. 2024 ) . We first note that the DES-only re-analysis in \Lambda CDM yields a higher value of S_{8} than the official results in Abbott et al. 2022b . This is due to our analysis choices listed in section 3 and in agreement with the DES shear re-analysis in Abbott et al. 2023a . DS expands the uncertainty on S_{8} and drives the mean value higher, while the extended parameters are prior dominated and consistent with the \Lambda CDM-limit. However, the best-fit values in S_{8} in both models are very close. The primordial amplitude computed at the MAP values equals to \log{10^{10}A_{s}}=2.865 in \Lambda CDM and \log{10^{10}A_{s}}=3.042 in DS. The latest CMB constraint without lensing information with Planck in PR4 for \Lambda CDM from Tristram et al. 2024 is \log{10^{10}A_{s}}=3.040\pm 0.014 . Our PR4 analysis with DS constraints the primordial amplitude to \log{10^{10}}A_{\rm s}=3.035\pm 0.014 . From this we see that DS offers a solution that can consistently connect early-time measurements of the matter density fluctuations in the CMB with late-time LSS measurements. Additionally, our PR4 analysis with DS constraints the following parameters: \omega_{\rm b}=0.0223\pm 0.0001 , \omega_{\rm c}=0.118\pm 0.001 , and n_{\rm s}=0.968\pm 0.004 . However, it leaves the highly degenerated parameters w,A_{\rm ds},h unconstrained with CMB data alone. The FS+BAO analysis reproduces the results of Carrilho et al. 2023 as expected with a strong shift in S_{8} towards lower values in the marginalised posterior. Note that the authors quote best-fit values of the analytically marginalised posterior. Based on the MAP values (see Figure 2 ) we see that the shift is caused by the projection effects in both cosmologies; the data prefer higher values of S_{8} than the posterior averages. Moreover, from minimising the full posterior in DS we find very close \chi^{2} values that correspond to two very different values of the extended parameters. This implies that the data are not constraining enough to draw any conclusion about the DS constraints in this analysis setup. In the joint (3 \times 2pt+FS+BAO) analysis for \Lambda CDM the S_{8} -value remains slightly lower than the value extrapolated from the CMB measurements without lensing information. For instance, the tension is quantified by 1.9\sigma and 1.3\sigma with Planck Collaboration et al. 2020 and PR4, respectively. The corresponding primordial amplitude computed at the MAP values equals to \log{10^{10}A_{s}}=2.975 . Meanwhile, DS constraints in the joint analysis are significantly improved, especially on A_{\rm ds} . The corresponding primordial amplitude computed at the MAP values equals to \log{10^{10}A_{s}}=3.029 . Similar to the DES-only analysis, in the joint analysis the MAP values are more in agreement with the CMB in DS than in \Lambda CDM. In Carrilho et al. 2023 the authors demonstrated that CMB-informed priors on the primordial parameters, n_{\rm s} and A_{\rm s} , improve constraints on the extended parameters, decrease the shift in S_{8} and the projection effects. As we show above, constraints on these parameters from the CMB in \Lambda CDM and DS are similar, hence the choice of the priors is justified. We find similar trends in the combination of the spectroscopic and photometric probes without CMB-informed priors on the primordial parameters. In the joint analysis our constraints on A_{\rm ds} have similar uncertainty, constraints on w are \sim 3-times weaker, while S_{8} constraints are \sim 2-times stronger than the ones from the FS+BAO analysis with the informative priors. Additionally, in the joint analysis, the difference between the MAP and marginalised mean values is reduced (see Figure 2 ), signifying a decrease in the projection effects with respect to the individual LSS probes. This is also demonstrated in Figure 3 , where we vary the standard deviation of the Gaussian prior of the EFTofLSS nuisance parameters. The effects of enlarging the prior are small in \Lambda CDM: only lowering n_{\rm s} and slightly decreasing S_{8} . In DS, the shift towards lowest prior-bound in n_{\rm s} is more prominent and resembles DES-only constraints for the largest EFTofLSS priors. We also observe a slight decrease in S_{8} and increase in h . However, the ‘butterfly’-posterior in w-A_{\rm ds} around its \Lambda CDM-limit remains for all prior-choices. This implies that the photometric probes effectively constrain the amplitude of the power spectrum and help to break the degeneracy between S_{8} and the extended parameters. Finally, an application of the normalising flows approach in \Lambda CDM is straightforward and in agreement with the brute-force joint analysis (see Figure 3 ). For DS we have two complications: (a) the overlapping region in the marginalised S_{8} -contours between the probes is small due to projection effects; (b) the ‘butterfly’-shaped posterior in w-A_{\rm ds} is highly non-Gaussian. To address these caveats we train the normalising flows on boosted posteriors (an approximation of true posteriors): this increases the size of the training set by approximately one order of magnitude. To better avoid projection effects in S_{8} , we repeat the FS+BAO analysis with a flat prior on the primordial amplitude, \log{10^{10}A_{\rm s}}\in[2.6,4] , motivated by the results of the DES-only analysis. We also re-run the DES analysis by imposing a BBN prior ( DESI Collaboration et al. 2024 , this is known to improve CombineHarvesterFlow performance, see ) . Overall, we reach a good agreement between the brute-force combination and the normalising flows approach with the boosted posterior distributions and adjusted A_{\rm s} -prior runs. CombineHarvesterFlow successfully passes this stress-test on highly non-Gaussian posterior distributions with strong projections. \begin{overpic}[wACDM_cosmology_without_Omb_varyEFTofLSSpriors_newlike_w_Harvester.pdf] \put(56.0,55.0){\includegraphics[scale]{wACDM_cosmology_butterfly_varyEFTofLSSpriors_newlike_w_Harvester.pdf}} \end{overpic} Figure 3: Marginalised posterior distribution for cosmological parameters in the joint \Lambda CDM (left panel) and DS (right panel) analyses for three different choices for the priors on EFTofLSS nuisance parameters: baseline case (green) and the cases with standard deviations of nuisance parameters, except for b_{1} , increased by factors of 3 (orange) and 10 (pink). Dashed black line shows baseline joint analysis with the normalising flow, CombineHarvesterFlow ( CHF ), approach. 5 Conclusions In this letter we combine photometric and spectroscopic probes from Stage III surveys: the 3x2pt correlation functions of DES Y3, and the FS+BAO measurements of BOSS DR12, including external BAOs. The methods and findings of this study are important for forthcoming beyond- \Lambda CDM analyses by Stage-IV surveys like DESI, Euclid, and Rubin, which will simultaneously provide spectroscopic and photometric datasets with unprecedented precision. Currently, the significance of the apparent S_{8} tension between the CMB and joint LSS analysis for \Lambda CDM depends on the reference CMB data ( Tristram et al. 2024 ; Roy Choudhury & Okumura 2024 , see, e.g.) . For instance, for our analysis setup the tension is reduced to \sim 1.3\sigma when compared with Planck PR4 within the standard cosmology. In case the tension would remain with Stage-IV data, Dark Scattering (DS), an interacting dark energy model, presents a potential resolution. Here we provide the first DS constraints from DES Y3 data, from the joint analysis of DES Y3 and BOSS DR12, and from Planck PR4 data. The joint analysis constraints on the DS parameters are w=-1.04^{+0.10}_{-0.08} and A_{\rm ds}=-0.2^{+4.6}_{-5.9} b GeV -1 . There is no significant detection of DS. However, in the DES-only and joint analyses, the MAP-values of the model demonstrate an exemplary solution of the tension. We also show that the joint LSS analysis without CMB information is comparable with the single-probe BOSS analysis with CMB-informed priors. The combination of probes allows minimising the projection effects without the inclusion of CMB information: it brings the marginalised posterior maxima closer to the corresponding best-fit values and weakens the sensitivity to the priors of the EFTofLSS nuisance parameters. Finally, we find good agreement between the direct joint analysis and the combination of posteriors via normalising flows. Data Availability Links and references with sources of the data and analysis pipelines are provided in the main text.