1: \begin{abstract}%%%%%%%%%%%%% 概要 %%%%%%%%

2: 	% Although the recent development of related theory has revealed

3: 	% the connection between the attainable convergence rate of kernel quadrature

4: 	% and the spectrum of the corresponding integral operators,

5: 	% little has been studied about positively weighted rules.

6:   We study kernel quadrature rules with positive weights for probability measures on general domains.

7:   % The key ingredients which can exactly integrate some appropriately chosen test functions

8: 	% such as eigenfunctions of the integral operator or their Nyst{\"o}m approximation.

9: 	Our theoretical analysis combines the spectral properties of the kernel with random sampling of points.

10:   This results in effective algorithms to construct kernel quadrature rules with positive weights and small worst-case error.

11:   Besides additional robustness, our numerical experiments indicate that this can achieve fast convergence rates that compete with the optimal bounds in well-known examples.

12: \end{abstract}

13: