Title: Universal construction of decoders from encoding black boxes

Abstract: This work considers transformations of encoding operations of quantum information that are given as isometry operations. We propose a universal protocol for isometry inversion that constructs a decoder from multiple calls of an encoding black box transforming a $d$-dimensional system to a $D$-dimensional system. This protocol is probabilistic but exact, and uses the black boxes in parallel. Our protocol can significantly outperform other protocols based on embedding isometry inversion in $D$-dimensional unitary inversion or quantum process tomography. This is because the success probability of our protocol is independent of $D$. The advantage of the performance of this protocol can be understood as being originated in the impossibility of exact isometry complex conjugation for $D \geq 2d$. This means that isometry inversion cannot be achieved by concatenating isometry complex conjugation and isometry transposition. We also perform semidefinite programming to obtain the optimal success probability of isometry inversion, isometry complex conjugation, and isometry transposition in parallel, sequential, and general protocols including indefinite causal order to investigate the performance improvement beyond the parallel protocols. We numerically find a sequential ``success-or-draw'' protocol for isometry inversion for $d=2$ and $D=3$. This implies that there exists a sequential protocol whose success probability is exponentially improved over parallel protocols in the number of calls of the input isometry operation.