XRD-VEC_DB: A simulated crystal structure and X-ray diffraction dataset for materials research

XRD-VEC DB is a simulated paired crystal structure and powder X-ray diffraction dataset for materials research. This release contains 6,568,959 crystal structure records paired with normalized powder diffraction representations. The dataset spans all 230 space groups and 89 elements, and was constructed through stochastic crystal-structure exploration, physics-informed structure filtering, machine-learning-potential relaxation, powder XRD simulation and spectral-vector indexing. The release provides structure records, XRD vector representations, metadata links between structures and spectra, and files required for retrieval-oriented reuse. The released XRD vectors are 180-point normalized powder diffraction representations over 10°–90° 2θ. The associated project repository provides scripts for data querying, XRD perturbation, molecular-dynamics-derived perturbation workflows, validation and candidate structure export. This dataset is intended for reuse in powder XRD structure retrieval, symmetry recognition, crystallographic representation learning, spectral vector search and inverse structure modelling.