Add determinant, laplace and cross products to math_vector.c3

[jopadan]

• [Feature Request] [2,3,4 and n]-dimensional vector product using cross[2,3,4]/hodge function #2033 except for n-dim
• add test_vec_cross to test/unit/stdlib/math/vector_math.c3

[lerno]

Can you please add some actual tests, look into test/unit how to construct them. You can then use c3c compile-test <path to folder> to test.

[lerno]

Sorry, that part wasn't supposed to be there. Not that it harms anything but..

[lerno]

One question I have is whether the vector operations are the most efficient this way (I'm thinking of the 4 cross product) or not, have you measured the resulting output compared to writing it out on -O1? Also, possibly the two different windings should be two different functions instead.

[lerno]

I am confused. There is this: macro float[<*>].determinant2(self, float[<*>] b) => determinant2(self,b);

But this looks like it's assuming

| self.x self.y | 
| b.x       b.y  |

Which means it's not applicable to any float vectors that aren't len = 2. Also, it is only defined for floats, as opposed to both doubles and floats.

I am also a little concerned that only some of them have tests.

[jopadan]

I don't see why the implementation should not work for vectors that aren't len = 2.
It supports len = n vectors with only taking m elements specified by determinant[m].

[jopadan]

See updated implementation for double

[jopadan]

Is your 2x2 post concern about column/row major format?

[lerno]

I think it's fairly confusing to have this. If you really want the example taking a subset of the vector then that isn't hard. Like for the determinant case for a 4 vector and you want to do the 2x2 determinant, why not simply do it explicitly? x.xy.det(y.xy)