Proper improvements of well-known perturbation estimations for the parallel sum (Q6083987)

The parallel sum in electrical circuits is one of the most important tools to study how parallel resistances add together. \textit{W. N. Anderson jun.} and \textit{R. J. Duffin} [J. Math. Anal. Appl. 26, 576--594 (1969; Zbl 0177.04904)] proposed the definition for the parallel sum of two positive semidefinite matrices, which is analogous to the form of parallel sum with scalars studied by \textit{K. E. Erickson} [IEEE Trans. Circuit Theory 6, No. 1, 124--126 (1959; \url{doi:10.1109/TCT.1959.1086519})]. As a generalization, the parallel sum and the shorted operators have been extended to several different contexts in the literature. This paper discusses the properties and applications on perturbations of parallel sum under the condition of positive semidefinite matrices. In particular, the authors establish an operator inequality with parameter \(v\in [0,1]\) involving the parallel sum of positive semidefinite matrices. Some new norm upper bounds related to the perturbations for the parallel sum of positive semidefinite matrices are presented. More specifically, the authors present a nontrivial improvement for the well-known perturbation estimations of parallel sum when the positive semidefinite matrices satisfy the condition of \(A+B\neq 0\), \(X\neq 0\) or \(A+B\neq 0\), \(X+Y\neq 0\). In addition, some intuitive examples are provided to show further that these norm upper bounds are much better than the results obtained by W. N. Anderson jun. and R. J. Duffin [loc. cit.].