Information Loss Is All You Need: TC^0 Pseudorandomness from Worst-Case LatticesResearch Paper
Learning with errors (LWE) is the standard source of post-quantum hardness: given many noisy inner products of a secret , distinguish them from uniform. Regev showed in 2005 that this is as hard, for quantum algorithms, as worst-case approximate lattice problems (Reg05). Removing the random error makes the samples deterministic: learning…