> What do you think about the papers showing mathematical proofs that GNNs (i.e. GATs/transformers) are dynamic programmers and therefore perform algorithmic re

> What do you think about the papers showing mathematical proofs that GNNs (i.e. GATs/transformers) are dynamic programmers and therefore perform algorithmic reasoning? Do you have a reference?