280 andrey depth limitations hps

[kimandrik]

Changed floating point arithmetics to fixed point arithmetics in both: ScaleAndRound function used in multiplication and ScaleAndRound function used in decryption.

Questions:

1. Do we need to add tests for HPS and other variants? If so the test could be slow, as they require lot's of levels for BFV.
2. I keep temporary bfv-mult-bug.cpp example from https://github.com/openfheorg/openfhe-development/tree/Erabelli_280-bfv-mult-bug. Do we need some variant of it?
3. In case of decryption should we use previous method with floating point arithmetics, or move to new with fixed point arithmetics?
4. Should we rename functions in dcrtPoly, as they could be confusing? Make renamings as a part of another issue?

[yspolyakov]

1. Do we need to add tests for HPS and other variants? If so the test could be slow, as they require lot's of levels for BFV.

I would add some tests using a small ring dimension, e.g., 16.

I keep temporary bfv-mult-bug.cpp example from https://github.com/openfheorg/openfhe-development/tree/Erabelli_280-bfv-mult-bug. Do we need some variant of it?

I think we want to create a unit test based on it

In case of decryption should we use previous method with floating point arithmetics, or move to new with fixed point arithmetics?

Let us set up a conference call and discuss this in more detail.

Should we rename functions in dcrtPoly, as they could be confusing? Make renamings as a part of another issue?

Better do it as part of another issue

[yspolyakov]

We had a meeting on this. Some code/approaches from this PR will be used to address #280 and #421