According to J. Gough, one of the bottlenecks in the current development of large-scale quantum computing may be the lack of our ability to simulate large scale quantum system, which is a NP-hard problem and also requires exponential memory with classical hardware if I understand correctly.
I'm currently developing a certain powerful discrete approximate algorithm. In principle, it can, for example, find an approximate solution for a given quantum Ising model efficiently or optimize the configuration of quantum gates to minimize the error a la U. Las Heras et. al. by replacing its GA with mine.
1) Can we reduce digital/analog quantum simulation to a NP-hard discrete optimization problem (like solving Ising model), so that it can be approximately solved by a classical algorithm using classical device?
2) What are some other crucial bottlenecks? Do they need discrete optimization?
Edit: I guess one of other bottlenecks is quantum error correction, which was already mentioned above.