Self-calibrated cluster counts as a probe of primordial non-Gaussianity

We show that the ability to probe primordial non-Gaussianity with cluster counts is drastically improved by adding the excess variance of counts which contains information on the clustering. The conflicting dependences of changing the mass threshold and including primordial non-Gaussianity on the mass function and biasing indicate that the self-calibrated cluster counts break the degeneracy between primordial non-Gaussianity and the observable-mass relation. Based on the Fisher matrix analysis, we show that the count variance improves constraints on fNL by more than an order of magnitude. It exhibits little degeneracy with dark energy equation of state. We forecast that upcoming Hyper Suprime-Cam survey and Dark Energy Survey will constrain primordial non-Gaussianity at the level sigma(fNL) approximately 8, which is competitive with constraints from next-generation cosmic microwave background experiments.