Exact Jensen–Shannon Divergence in Gaussian Secure Aggregation

This release provides the exact asymptotic expansion of the Jensen–Shannon divergence between neighboring datasets in the Gaussian secure aggregation model. We proveJSD = SNR/ (8n) − SNR²/ (64n²) + O (n⁻³), with a negative second-order coefficient, highlighting a qualitative contrast with discrete randomized response. The result establishes the universal leading-order relation JSD = χ²ₐgg/8 + O ( (χ²ₐgg) ²). This version updates and supersedes the previous Zenodo record, incorporating a complete mathematical audit and independent numerical verification.