Bioequivalence and Bioavailability Forum

Dear Supreme Grand Maestro,

❝ PS: Any wise guy wanting to tell me I shouldn't find b through inversion but use a QR factorisation in stead?

I must confess that I'm surely not an expert in numerical Al-Gore-Rhythms optimized for speed for the solution of a general linear model fit. And my last encounter with matrices calculus is also some long years ago.
Nowadays I had only coincidences with Matratzen every night .
And I'm surely not wise even so old.

But just some small observations from what R itself is doing:
Type in your R console lm without brackets (!) to see in the code that the working horse of this function is the function lm.fit().

?lm.fit will give you the information that it uses a QR factorisation as the solely method for solving.

Type lm.fit (again without brackets) and notice that here the calculation burden is assigned to a call of compiled FORTRAN code via function .Fortran(...), namely to the LINPACK routine DQRDC as the help tells us. The rest are some native R code lines surely not important for speed.

Thus I highly suspect that you don't get great speed gains if you implement the whole process in FORTRAN or C by yourself.