Notes. Intrinsic galaxy alignment contributions are given by the one-parameter (A₁) non-linear alignment model, non-linear structure growth is described by the one-parameter (A_bary) halo model HMCode with baryonic contributions (Mead et al. 2015, 2021), galaxy biases b_g are linear and deterministic, and scalar shifts in the mean redshift δz_i are employed for recalibration of photometric redshift distributions. Priors are flat except for δz_i, where Gaussian priors are estimated as described in Sect. 3.1.1. Central values (for non-δz_i parameters) are taken as the best-fit values from Asgari et al. (2021) (their Table A.2, Col. 3) except in the case of A_bary, which we reduce slightly in order to introduce some baryonic contributions into the fiducial data-vectors. Whilst Asgari et al. (2021) sampled directly over the S₈ parameter, it is more convenient for our implementation of CosmoPower (Spurio Mancini et al. 2022) to sample over A_s. We convert the (derived) best-fit A_s from Asgari et al. (2021) into a fiducial centre for ln(10¹⁰A_s), and assume a flat prior on ln(10¹⁰A_s) corresponding to 0.5 < 10⁹A_s < 5.0 (Secco et al. 2022). Other flat priors are taken from Asgari et al. (2021), or from Mandelbaum et al. (2018) in the case of the galaxy bias, where the latter also provides the functional form for the galaxy bias (Eq. (2.18)). The derived, fiducial S₈ value is given in the last row, and is slightly larger than that found by Asgari et al. (2021) due to our modification of the fiducial A_bary. Notice that we also adopt the best-fit for h as in Asgari et al. (2021), and while this is at the boundary of our prior, it does not affect our conclusions given that h is unconstrained.