FHElium

Appearance

fhelium.backend.ckks.codec._periodic ​

Prepare periodic CKKS messages using their sparse polynomial embedding.

periodic_extent ​

function View source

python
def periodic_extent(period: int, ring_dimension: int, generator: int) -> int: ...

Return the coefficient count in the embedded negacyclic subring.

periodic_materials ​

function View source

python
def periodic_materials(config, context, period: int, depth: int, basis: str, state): ...

Build compact embedding and indexed NTT tables with the full-ring roots.

prepare_periodic_tensor ​

function View source

python
def prepare_periodic_tensor(message, pre, twister, state, parameters, twice_moduli, even, odd, twiddles, *, ring_dimension, scale, min_modulus, polynomial_domain='ntt'): ...

Compute compact NTT rows for p(X)=a(X**(N/U)), with a of degree below U.

The U-point inverse embedding uses the original generator's slot order. Rounding selects word j*N/U of the full coefficient stream and advances by N words per batch item. NTT roots are psi_N**(N/U), so each returned bit-reversed NTT entry represents N/U contiguous entries of the full NTT. FFT roundoff can differ from a full-size FFT near rounding thresholds.

NativePrepareCompressedPlaintextImplementation ​

class View source

python
NativePrepareCompressedPlaintextImplementation()

Prepare compact NTT plaintext rows from one period and supplied tables.

Attributes ​

NameTypeDefault/value
name'native-prepare-compressed-plaintext'
supports_in_placeFalse

resource_requirements ​

method

python
def resource_requirements(invocation): ...

execute ​

method

python
def execute(invocation, inputs, resources, *, in_place): ...