about conv and pooling in paper

[nigelzzzzz]

The paper is very interesting, and i read two paper about difflogic.
https://arxiv.org/abs/2210.08277
https://arxiv.org/html/2411.04732v1

I still have some questions.

• why need to do flatten in begin.
• I checked issue in github. (about "convolutional" differentiable logic gate neworks paper #28), it says the source code can be released lately. it's about conv and max pooling, right? because i can't see the logic in repo.
• The logic about conv and pooling, can i know why the purposal can transfer conv and pooing to below formula.

---

---