Lion-ZS
commented
1 year ago Thank you for your attention. The main reason lies in that what we want here is to transform the entities in non-Eculidean space. In this way, we adopt multiplying the rotation to the head entity following ROTH. Moreover, you can also utilize the Mobius multiplication, practicaly, they are similar in performance for this case.
At 2023-03-01 16:05:49, "ukjung21" @.***> wrote:
In the paper, when transforming the head entity in non-Euclidean space, you used mobius multiplication.
However in the code, you just multiply the rotation to the head entity elementwise, which is used in both Euclidean and non-Euclidean space. res_c1 =logmap0(givens_rotations(rel2, lhs),c1) rot_q = givens_rotations(rel2, head)
Would you please tell me why you did not used mobius multiplication in the code?
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>
In the paper, when transforming the head entity in non-Euclidean space, you used mobius multiplication.
However in the code, you just multiply the rotation to the head entity elementwise, which is used in both Euclidean and non-Euclidean space.
res_c1 =logmap0(givens_rotations(rel2, lhs),c1)rot_q = givens_rotations(rel2, head)Would you please tell me why you did not used mobius multiplication in the code?