1: \begin{abstract}\normalsize

2: Distributed learning offers a practical solution for the integrative analysis of multi-source datasets, especially under privacy or communication constraints.

3: However, addressing prospective distributional heterogeneity and ensuring communication efficiency pose significant challenges on distributed statistical analysis.

4: In this article, we focus on integrative estimation of distributed heterogeneous precision matrices, a crucial task related to joint precision matrix estimation where computation-efficient algorithms and statistical optimality theories are still underdeveloped.

5: To tackle these challenges, we introduce a novel HEterogeneity-adjusted Aggregating and Thresholding (HEAT) approach for distributed integrative estimation.

6: HEAT is designed to be both communication- and computation-efficient, and we demonstrate its statistical optimality by establishing the convergence rates and the corresponding minimax lower bounds under various integrative losses.

7: To enhance the optimality of HEAT, we further propose an iterative HEAT (IteHEAT) approach. 

8: By iteratively refining the higher-order errors of HEAT estimators through multi-round communications, IteHEAT achieves geometric contraction rates of convergence.

9: Extensive simulations and real data applications validate the numerical performance of HEAT and IteHEAT methods.

10: \end{abstract}

11: