DSWIM:Efficient and Stable Deterministic Computation of Warm Inflation Perturbations
Warm inflation perturbations are sourced by both thermal and quantum fluctuations and are commonly computed through stochastic realizations of the perturbation equations, as implemented in the publicly available code SWIM. Deterministic formulations based on correlation matrix evolution provide a computationally efficient alternative, but can become numerically ill-conditioned when the perturbation variables evolve over widely different scales. In this work, we extend SWIM by introducing a deterministic module, DSWIM, based on correlation matrix evolution. We introduce a physically motivated scaling matrix transformation derived from the effective Hubble scaling of the perturbation variables. The transformed system preserves the primordial curvature power spectrum exactly while substantially improving the numerical conditioning of the deterministic evolution equations. Using representative warm inflation models, we show that the scaled framework suppresses numerical artifacts, improves the robustness of the deterministic evolution, and yields substantial computational speedups while preserving accuracy. We further show that correlated thermal noise contributions arise naturally through the diffusion matrix structure, resolving previously observed discrepancies between stochastic and deterministic implementations. Our results establish DSWIM as a numerically robust and computationally efficient framework for computing warm inflation scalar perturbations.