FTransform inverse/compose strange results

[anonymous_user_7ebabb4e]

I have a simple example:

and it will log:

Then, if you change scale to repeated same values, for example [1, 1, 1]:

it will log:

If I’m not mistaken, then

For every invertable matrice A.

But here is happening something strange.
If you swap pins on compose, than it will always log correct result.

I don’t believe, that Epic could make a mistake in such core feature.

Then, most likely,
the problem is in my knowledge of linear algebra.

Pls, help.

[redbox]

I would make initial transform editable, add this to construction script, and test the results when changing initial transform to different values. Also it would be useful to check c++ version of invert transform function, maybe it have some restrictions for non-uniform scales.