Skip to content

Mathematical knowledge

This folder collects the mathematical knowledge needed to understand NCET's bound propagation and exact optimization encoding. The notes describe the formulations and their connection to the current implementation; broader frontend and GraphIR design details are documented elsewhere under docs/.

  1. Interval Bound Propagation explains how NCET obtains finite tensor bounds throughout a graph.
  2. Exact Encoding explains how a bounded GraphIR becomes an LP- or MILP-compatible CVXPY constraint system.
  3. The operator notes derive the sparse matrices or exact formulations used by the backend:
  4. Conv2d
  5. BatchNorm
  6. AdaptiveAvgPool2d
  7. ReduceMean
  8. AvgPool2d
  9. MaxPool2d

Scope of each note

Note Main topic
Interval Bound Propagation Bound rules for affine, monotone, arithmetic, and structural operators
Exact Encoding Graph-wide LP/MILP variables, constraints, and exactness
Conv2d Sparse affine matrix for convolution
BatchNorm Inference-mode per-channel affine map
AdaptiveAvgPool2d Adaptive windows and sparse averaging matrix
ReduceMean Mean reduction, dimension handling, and linear equality
AvgPool2d Sparse averaging matrix and pooling divisor
MaxPool2d Full one-hot exact maximum formulation