An FPCA-enhanced Ensemble Learning Framework for Photometric Identification of Type Ia Supernovae

Precision cosmology requires robust, data-driven methods that can handle complex survey systematics without closed-form likelihoods. Simulation-Based Inference (SBI) meets this need through forward simulations that encode complex survey characteristics. Previous SBI analyses in supernova cosmology used SALT2 light-curve parameters as summary statistics; however, SALT2's two-basis spectral template imposes rigid modeling assumptions, motivating a more flexible representation. We present the first application of Functional Principal Component Analysis (FPCA) light-curve parameters as summary statistics for SBI-based cosmological inference under a non-flat Λ CDM model. On identically generated simulations, FPCA+SBI yields constraints comparable to both SALT2-based SBI and explicit-likelihood analyses, while providing more robust constraints on out-of-domain simulations than SALT2. Applying a model trained on LSST-like light curves to a spectroscopically confirmed DES Year 5 supernova sample, we recover constraints consistent with those of the DES collaboration to within 0.12 σ and 0.23 σ in Ω m and Ω Λ, respectively, establishing generalizability to real survey data. Introducing host-dependent systematics through a mass-dependent dust extinction law, we find that FPCA-based summary statistics implicitly encode them, suggesting dedicated host modeling may be unnecessary. Combined with its demonstrated effectiveness for photometric classification, FPCA offers a foundation for unified, data-driven pipelines that jointly perform supernova classification and cosmological inference for upcoming wide-field photometric surveys.