Everyone is welcome to attend the online QuSoft seminar with Joseph Renes. Title: Belief propagation decoding by passing quantum messages .
Abstract:
Graphical models are a powerful tool for representing and manipulating classical random variables with structured dependency relations. A very widely-used algorithm in this setting is belief propagation, which efficiently calculates marginal distributions, and which has found applications to error-correction, free energy approximation, and satisfiability. An immediate question for enthusiasts of quantum information is: Can we construct truly quantum analogs of graphical models and belief propagation? In this talk I describe one instance of such a quantum graphical model and belief propagation algorithm for the task of decoding certain error-correcting codes for certain quantum channels. Specifically, the decoding algorithm can be given an explicit description as a message-passing algorithm which is efficient in the blocklength of the code, and which optimally decodes individual codeword bits. Moreover, it is capable of efficiently decoding capacity-achieving polar codes for the pure loss channel and the amplitude damping channel.
Joint work with Christophe Piveteau.
Please contact Subhasree Patro or Jop Briet if you like to sign up.
