BAQARO: Tracing Stochastic Black Hole Growth Histories and Quasar Lightcurves in a Cosmological Context
How supermassive black holes (BHs) assemble their mass and power the luminous quasars we observe across cosmic time remain central open questions in galaxy evolution. We present BAQARO, a semi-empirical framework for BH growth. Built on subhalo merger trees extracted from the FLAMINGO-10k simulation, the model links BH growth histories to those of their host subhalos through prescriptions that capture both average trends and stochastic variability. BAQARO is constrained by the bolometric quasar luminosity function, the clustering of UV-luminous quasars, and the conditional Eddington ratio distribution function. With six free parameters controlling BH seeding, the coupling between gas accretion and halo growth, and the stochasticity and temporal coherence of accretion, the model reproduces the available observational constraints over $0 \lesssim z \lesssim 7$. Using emulators, we perform Bayesian inference and quantify the constraining power of each observable. Our main findings are: (i) BH accretion is well described by the assembly of cold gas reservoirs in halos, without requiring an explicit dependence on cosmic time. (ii) BHs rapidly assemble their mass at high redshift through stochastic episodes of super-Eddington accretion. (iii) These episodes are radiatively inefficient and persist for timescales of $\sim1\,\mathrm{Myr}$, imprinting observable signatures on quasar lightcurves, proximity zones, and clustering measurements. (iv) The merger growth channel is always subdominant, but becomes increasingly important at $z \lesssim 1$, particularly for massive BHs. BAQARO growth histories and quasar lightcurves provide a flexible framework for interpreting the rapidly expanding landscape of quasar observations, from high-$z$ accretion probed by JWST to low-$z$ mergers constrained by pulsar timing arrays.