Scalable Density-Based Bootstrap Embedding with Perturbative Triples for Large Molecular Systems

Extending density-based bootstrap embedding (BE) with perturbative triples to large molecules requires control of both fragment solver cost and tensor storage. We present a restricted and unrestricted BE-CCSD(T) implementation combining density-fitted embedding transformations, fragment parallelism, and streamed contractions of triples density contributions. Explicit one-and two-particle terms and the correction to the coupled-cluster multipliers are retained without assembling a molecular two-particle density matrix or storing a complete triples tensor. Fixed-order fragmentation bounds the local correlated problem under stated locality assumptions, while blockwise evaluation controls temporary storage. Timings for