________________________________

]]>It is known that the computational complexity of forward propagation and backward propagation are about the same. So just as simulating (forward propagation) of a quantum circuit blows up exponentially with the number of qubits, so does back-prop. So, why do it? For the same reasons we do simulations, to learn new ideas and test things. For me, a bnet devotee, the fact that a bnet is built behind the scenes during back-prop, is a siren call. Imagine all the things I can learn about bnets by studying the ones generated by back-prop! Lessons that might find applicability in the quantum bayesian nets, quantum fog project.

]]>https://sociable.co/technology/darpa-exploit-quantum-computing-without-quantum-computer/ ]]>

https://www.technologyreview.com/the-download/613069/ibm-has-come-up-with-a-new-way-of-measuring-the-progress-of-quantum-computers/

https://arxiv.org/pdf/1811.12926.pdf ]]>

https://spectrum.ieee.org/computing/hardware/when-will-quantum-computing-have-real-commercial-value ]]>