“…III, that for n A = n B we have max I n,n = max{n 2 − 2(k − l) 2 }, by adding the marginals max I ′ n,n = max{n 2 − 2(k − l) 2 + (2k − n) − (2l − n)} = max{n 2 − 2(k − l)(k − l + 1)} = n 2 , thus the local bound does not change. Let us notice, that I ′ 2,2 specified by n = 2 in (14) is just the I 3322 Bell inequality [40,41], which is known to be tight [42]. In both cases, I n,n−1 and I ′ n,n , we suspect that these Bell inequalities are tight for any higher values of n > 4, as well.…”