Closed BigUglySpider closed 3 years ago
It should be noted that this has been classed as lower priority due to Inverse
(the main operation one would likely want in scenarios where MatrixOfMinorsLaplace
may be called) already being implemented via gaussian elimination.
As of commit dd9b0ea60f0eb8efa70f5ee4efc023839cf61f37, this issue has been address. Taken verbatim from the commit:
No longer brute forcing minors function, and allowed optional possibility to transpose the output automatically (this uses an altered shuffle for the output, and thus has the same overhead as constructing the non-transposed version).
This addresses the recently published issue regarding the performance of MatrixOfMinors
.
As of commit 4b8b1a10a3ed2d88283c2cd04369d68550fb3eeb,
FastMatrix4x4f_CM
has access to the laplace expansion methods that the scalar matrix has access to.However, this is implemented highly inefficiently due to the root function for each step in the laplace expansions methods:
MatrixOfMinors
.As the matrix of minors is the first step for any
...Laplace
function, their performance is strongly tied to that ofMatrixOfMinors
. An improvement of efficiency to this would, as a result, be an improvement to the efficiency of all...Laplace
functions inFastMatrix4x4f_CM
.