[racket] matrix-solve and approximation errors

Neil, a two-layer design, using option contracts, could be a solution here. 


------------------------------------------------------------------------


On Apr 16, 2014, at 3:18 PM, Konrad Hinsen <konrad.hinsen at fastmail.net> wrote:

> Any linear-algebra library meant for "serious" work does not do any
> non-essential work "just" for error checking. Performance is king.
> Users are supposed to know the pitfalls and take care to avoid them.
> Unfortunately, many don't. 

It is incredible how often I find this scenario. Just yesterday I had 
a colleague tell me (on a related subject) that "this is how you test 
people who want to work in this field. If they can't figure out these
assumptions, they are not well suited. Flush them." 

My proposal had been to create a type-like system to check some invariant
about geometric relations. His response in the end was "yeah, it may help
more novices to overcome this obstacle, but the field wont' care in the end. 
Experts don't make these mistakes." 

At this age, I am 1000% convinced that experts make these mistakes too, but
what can we say. 

-- Matthias