Near-Optimal Convex Optimization with Lazy Second-Order Oracles

This paper studies the complexity of convex optimization using lazy second-order oracles (Doikov, Chayti, and Jaggi, ICML 2023), where an algorithm queries gradients every iteration and Hessians once per $m$ iterations. Under this setting, we show a lower bound of $\mathrm{\Omega }(m+m^{1\mathrm{/}7}{\epsilon }^{-2\mathrm{/}7})$ on the number of total iterations to find an $\epsilon$-solution using a novel block zero-chain construction. Then we propose a novel method that achieves a new upper bound of $\tilde{\mathcal{O}}(m+m^{1\mathrm{/}7}{\epsilon }^{-2\mathrm{/}7})$, which significantly improves the prior one (Chen, Liu, Luo, and Zhang, COLT 2026) of $\tilde{\mathcal{O}}(m+m^{13\mathrm{/}21}{\epsilon }^{-2\mathrm{/}7})$ and is tight up to logarithmic factors.