Hello,
I am wondering if anyone in the community here is aware of any research in homomorphically generating pseudorandom numbers given a provided random seed. I haven’t found anything but my searches are confounded by the common use of PRNGs in the clear associated with generating keys and noise, etc. Ideally, I would like to generate from a gaussian distribution but I’m interested in uniform as well.
I appreciate any recommendations or optimistic/pessimistic reactions!
I suggest looking at Homomorphic Evaluation of LWR-based PRFs and Application to Transciphering It is not exactly what you are asking about, but the problem is related.
Thanks this is an interesting paper and I will have to read it more thoroughly. I don’t think this will work in my use case as I will need to apply random noise to a CKKS vector of data. Although the throughput of this method looks sufficient, I think scheme switching the output TFHE → CKKS would become prohibitively slow.