1: \begin{abstract}    % Abstract of not more than 250 words.

2: % \textcolor{red}{}

3: This paper addresses a distributed convex optimization problem with a class of coupled constraints, which arise in a multi-agent system composed of multiple communities modeled by \textit{cliques}.

4: % having constraints.

5: First, we propose a fully distributed gradient-based algorithm with a novel operator inspired by the convex projection, called the clique-based projection.

6: % This operator is easily computable for typical constraints, e.g., norm and linear constraints. 

7: Next, we scrutinize the convergence properties for both diminishing and fixed step sizes.

8: For diminishing ones, we show the convergence to an optimal solution under the assumptions of the smoothness of an objective function and the compactness of the constraint set.

9: Additionally, when the objective function is strongly monotone, the strict convergence to the unique solution is proved without the assumption of compactness.

10: For fixed step sizes, we prove the non-ergodic convergence rate of $O(1/k)$ concerning the objective residual under the assumption of the smoothness of the objective function.

11: % for smooth $f$ .

12: Furthermore, we apply Nesterov's acceleration method to the proposed algorithm and establish the convergence rate of $O(1/k^2)$.

13: % under the smoothness of $f$.

14: Numerical experiments illustrate the effectiveness of the proposed method.

15: \end{abstract}

16: