We commit to the hashed set in 0.5g/m, and then do set membership in (n-1)g. We could merge these to just 1 if we didn't do set membership of the hashed values, yet the hashed values - the claimed member. The issue is we'd then hash those values, which wouldn't line up with the tree in the slightest.
Taking the logarithmic derivative may be possible. Instead of 1.5n - 1, it may be 1n + 1 (0.5 for commit, 0.5 for member - claimed_member commit, 1 to eval). I have yet to review this in the slightest though.
We commit to the hashed set in 0.5g/m, and then do set membership in (n-1)g. We could merge these to just 1 if we didn't do set membership of the hashed values, yet the hashed values - the claimed member. The issue is we'd then hash those values, which wouldn't line up with the tree in the slightest.
Ideally, we do get this down to just 1g/m.