Consult pt2 here.
In this post I will provide a journal about the savings due to the presence of zeros in the submatrix needed to compute the sum rules.
Day 0
Problems found regarding the savings. The \({\rm det}\) and a transformation are not always valid. Sometimes, the submatrix is not Hessenberg.
Day 1
\[
T_{U, V}^{\lambda}=
\frac{1}{\sqrt{\Theta_{i}^{f}}} \frac{1}{\sqrt{\Theta_{j}^{g}}} \sum_{\cup_\gamma S_{\alpha}
\gamma S_{\beta}=S_{n}}\left(\sum_{\sigma \in S_{\alpha} \gamma S_{\beta}} \omega_{i, j}^{\lambda}(\sigma)\right) X_{f\circ\gamma, g}.
\]
According to the previous equation, if the submatrix present some zeros then at least one \(X_{f\circ\gamma, g}\) is zero. This in turn leads to avoiding compute some orthogonal irrep (OI) entries.
This is the unique saving I see at the present.
Thanks to Antti Koskinen Post