Dirk Laurie on Tue, 12 May 2015 14:32:12 +0200

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: incorrect type in gpow(0,n) (t_VEC)

2015-02-12 7:33 GMT+02:00 Kevin Acres <research@research-systems.com>:

> I've noticed recently that something like [1,2,3]^[2] works fine,

Then you have probably also noticed that the result is a vector of
one-element floating-point vectors. I'm curious about why anyone
would want precisely that.

> whereas if any element is zero I get an error:
> ? [0,2,3]^[2]
>   ***   at top-level: [0,2,3]^[2]
>   ***                        ^----
>   *** _^_: incorrect type in gpow(0,n) (t_VEC).
>   ***   Break loop: type 'break' to go back to GP prompt
> Is there a work around for this?

The reason is that floating-point a^b is in effect computed
as exp(log(a)*b). log(0) => bang!

> As background I need to perform sqrt(vecsum(x^[2])) so maybe there is a
> better way of obtaining the same result.

I would have expected `norm(x)`, whose help message says
"norm(x): norm of x.",  to do that, but gp 2.5.5 (which comes with
Ubuntu 14.04 LTS) returns [0,4,9] when x=[0,2,3]. Doubtless all
will become clear if one peruses the user's manual.

sqrt(x*x~) works.