When does allow the Hardy inequality to calculate an exact Poincaré constant on a line? (Q959015)

Consider two symmetric measures \(\mu\), \(\nu\) on \(\mathbb{R}\), \(\mu\) being a probability. Denote by \(A(\mu,\nu)\) the best constant in the Poincaré inequality: \[ \text{Var}(f,\mu)\leq A(\mu,\nu)\int_{\mathbb{R}} |f'|^2\,d\nu, \] and denote by \(\varphi\) the density of the absolutely continuous part of \(\nu\). Set \[ B(\mu,\nu):= \max\Biggl\{\sup_{x< 0}\int^0_x {dy\over\varphi(y)}\times \mu(]-\infty,x]);\;\sup_{x> 0} \int^x_0 {dy\over \varphi(y)}\times \mu([x,\infty[)\Biggr\}. \] The purpose of the author is to establish that \[ B(\mu,\nu)\leq A(\mu,\nu)\leq 4B(\mu,\nu), \] thereby improving (by a factor 2) in the present case bounds got by Bobkov and Götze, by other methods. The author proves also that these bounds are optimal, by yielding criterions ensuring that \(JA= B\), or that \(A= 4B\).