Skip to main content

← HomOps Reference

HomSqueeze

Remove a size-1 axis from an encrypted tensor's shape.

SzShape
RunsServer
Changes shapeYes - removes a size-1 axis
Changes scaleNo - unchanged
Levels spentNone
RotatesNo
KeysNone

What it does​

HomSqueeze removes a size-1 dimension at a chosen axis. It is layout-only: no arithmetic on the encrypted values, matching PyTorch's squeeze(dim) on encrypted tensors. Compilation fails if the target axis is not size 1.
Use it to undo an HomUnsqueeze, or to drop a kept dim after a reduction that left a trailing size-1 axis.

Signature​

HomSqueeze(dim)
No set_data.

Parameters​

ParameterTypeDefaultDescription
dimintrequiredAxis to remove; it must currently have size 1.
Rules that depend on the value
SettingRuleIf you break it
dimshape[dim] must be 1.Raises ValueError.
n axisIf the removed axis sits before the n axis, that axis index may decrement in the new shape.Not an error - but see Concepts.

Requirements​

  • Size-1 axis. The named axis must have length 1.
  • No keys, no levels. Layout only.

Shape effect - yes​

Rule: remove axis dim after confirming it has size 1.
InputdimOutput
(1, 128, 512)0(128, 512)
(16, 1, 128)1(16, 128)

Scale effect - no​

Scale is unchanged.

Level budget​

None. Squeeze does not drop a prime from the modulus chain.

Keys​

None.

Example​

Drop a leading size-1 axis after an unsqueeze or a keep-dim reduction:
pipeline.pyPYTHON
from lattica_build.operators import HomSqueeze
from lattica_build.base_classes.hom_pipeline import HomomorphicPipeline
from lattica_build.operators.composite.sequential import SequentialHomOp

pipeline = HomomorphicPipeline(
    hom=SequentialHomOp(
        HomSqueeze(0),   # (1, 128, 512) → (128, 512)
    ),
    input_shape=(1, 128, 512),
)

See also​