Useful concentration-factor choices for general shadow ensembles

Determine whether an appropriate concentration threshold η yields a sufficiently large concentration factor for general classical shadow tomography ensembles to make the variance lower bound in Corollary~\ref{cor:shadow-variance-lower-bound} useful.

Background

The paper introduces a concentration factor measuring how much probability mass a shadow estimator assigns to outcomes whose magnitude is close to its maximum. This factor, together with the quantum Bernstein norm, gives a lower bound on the estimator’s second moment and hence on its variance.

For certain uniform classical-shadow strategies the bound is tight, but its usefulness for general measurement ensembles depends on whether one can choose a threshold η that produces a sufficiently large concentration factor. The authors explicitly identify this as unresolved and ensemble-dependent.