41CL Quiz: Determinant of 30x30 anti-Identity matrix

[Valentin Albillo]

.
Hi, Werner:

(05-20-2018 09:21 AM)Werner Wrote:  The venerable 42S solves 2x2 cases separately, and so does not fall into this trap. (but Free42 does ;-)) That, however, does not solve the problem as the equivalent

3 1 0
3 1 0
0 0 1

yields the determinant 3.e-12 as well. Not much to be done about that, save for the 48G's (and successor's) approach to return an integer determinant for a matrix with integer coefficients.

I think that something very simple could and should have been done, namely to use the expanded formula for 2x2 and 3x3 cases, which entails very, very few multiplications and additions and no divisions whatsoever. This would be both faster and exact as compared to using the general reduction algorithm on such simple cases.

HP did whatever needed to be done to ensure that 2^3 gives exactly 8 but didn't care that det(3,1,3,1) isn't 0 as it should.

Quote:(it's actually even better than that but that would require a longer explanation ;-)

As we say here, "prisa no tenemos y tarde no es", i.e. "It's not too late and we're not in a hurry" so go ahead with that longer explanation, please, I'm all ears ... :-D

Regards.
V.
.