Comparison between tests computed in the clear and the corresponding statistics computed using HE

Library Questions
1

Hi,
In the general case of building an application using openFHE library, how is it possible to deactivate the encryption/decryption steps in order to compare results from tests computed in the clear with the corresponding results computed using HE?
For example, using openfhe-genomic-examples, is it possible to compare statistics from Logistic Regression Approximation or Chi-Square GWAS protocols using the same code with HE activated/deactivated.
Thank you.

2

We typically write a reference implementation (e.g., in Python) using the same vector representation and floating-point arithmetic, which is always the first step in developing any FHE application.

Hypothetically it is possible to set all random masks, secret keys, and error distributions to zero in OpenFHE and observe the outputs (it is a quite tricky, though). We had a prototype of a null CKKS scheme in the past but it turned out to be challenging to maintain it and expect reasonable results. So we did not publicly release it.

3

Alright, thanks for your reply Yuriy!