A Tolerant Independent Set Tester

We give nearly optimal bounds on the sample complexity of ( Ω(ǫ), ǫ)-tolerant testing the ρ-independent set property in the dense graph setting. In particular, we give an algorithm that inspects a random subgraph on O(ρ 3 /ǫ 2 ) vertices and, for some constant c, distinguishes between graphs that have an induced subgraph of size ρn with fewer than ǫ c log 4 (1/ǫ) n 2 edges from graphs for which every induced subgraph of size ρn has at least ǫn 2 edges. Our sample complexity bound matches, up to logarithmic factors, the recent upper bound by Blais and Seth (2023) for the non-tolerant testing problem, which is known to be optimal for the non-tolerant testing problem based on a lower bound by Feige, Langberg and Schechtman (2004) . Our main technique is a new graph container lemma for sparse subgraphs instead of independent sets. We also show that our new lemma can be used to generalize one of the classic applications of the container method, that of counting independent sets in regular graphs, to counting sparse subgraphs in regular graphs.