Sharpness of the exponent in the radial eigenvalue sum bound; counterexample for β < 1
Construct a counterexample showing that, for Schrödinger operators H = −Δ + V on L^2(R^d) with complex radial potentials V ∈ L^q(R^d) (q ∈ ((d+1)/2, d)) and any p > ((d−1)q)/(d−q), the strengthened inequality (∑_j δ(z_j)^β |z_j|^{β p (1 − d/(2q)) − 1})^{q/(β p)} ≤ C_{p,q,β} ∫_{R^d} |V|^q dx fails for some β < 1, thereby establishing that the exponent q/p in Theorem 1 cannot be increased.
References
We conjecture that the exponent $q/p$ in eq. sums of eigenvalues main theorem cannot be increased, i.e.\ that the inequality
\begin{align}
\left(\sum_{j}\delta(z_j){\beta}|z_{j}|{\beta p(1-\frac{d}{2q})-1}\right){q/(\beta p)}\leq C_{p,q,\beta}\int_{Rd}|V|qd x
\end{align}
fails for $\beta<1$. We leave it as an open problem to find a counterexample.