I just thought it was cool math. But Thomas suggested to me and Clément that it could be useful for discrete Gaussian sampling in the context of differential privacy. And the rest is history.

→ View original post on X — @thegautamkamath,