Assumptions and limitations =========================== PackLab provides complementary models, not interchangeable ways to obtain the same result. Record the physical inputs and choose the workflow whose assumptions match the question. Common inputs ------------- Radii, number fractions, volume fraction, domain dimensions, and boundary conditions define the physical system. Use explicit units for every dimensional input. A radius distribution is not merely a numerical setting: it changes the mixture composition and its structure. RSA --- Random sequential adsorption produces non-overlapping configurations by irreversible deposition. Accepted spheres never move, so early random choices can block later placements. RSA is appropriate for deposition-like packings and for workflows that require explicit sphere centres. It is not an equilibrium hard-sphere sampler. Check the stopping criterion, random seed, realised packing fraction, and box size. A configuration can be valid yet still be too small or too sparse for a stable bulk correlation estimate. Metropolis Monte Carlo ---------------------- Metropolis moves target an equilibrium hard-sphere system at fixed particle count and volume. A valid RSA packing is a convenient initial state, but its deposition history must be forgotten through burn-in before collecting data. Assess stationarity, autocorrelation, acceptance rate, and sensitivity to box size; a fixed sweep count alone is not evidence of equilibration. Percus--Yevick -------------- The analytical workflow is a Percus--Yevick approximation for an equilibrium hard-sphere mixture. It supplies pair correlations and structure factors quickly, but does not create centres, represent deposition history, or remove finite-size effects from a simulation. Verify numerical resolution by refining the wavenumber grid when real-space correlations are important. Scattering ---------- Scattering is optional and uses PyMieSim for the single-particle optical amplitudes. A mixture phase function additionally depends on the supplied structure tensor. When that tensor comes from Percus--Yevick, the scattering result is structure-aware but remains an analytical equilibrium prediction, not a scattering calculation from an RSA configuration. Comparisons ----------- When comparing workflows, match radii, composition, total volume fraction, boundary conditions, and the definition and resolution of the observable. Agreement between equilibrated Metropolis samples and Percus--Yevick tests an analytical approximation. Differences between RSA and Percus--Yevick are often expected because they describe different physical processes.