Re: [Gretl-devel] bug?
On Mon, 28 Jan 2019, Sven Schreiber wrote:
Am 28.01.2019 um 00:20 schrieb Allin Cottrell:
> On Sun, 27 Jan 2019, Sven Schreiber wrote:
>> open denmark
>> eval (LRY .< 10) * IBO
>
> Your first formulation is apperently confusing gretl by employing a
> dot-operator (".<"). Such operators are explicitly intended for matrices
> only, though a degree of slop is tolerated in some contexts. If you do
>
> open denmark
> eval (LRY < 10) * IBO
>
> you get the expected answer without complaint.
Ah OK, thanks. Personally I would perhaps favor less tolerance because
that would render things clearer and more understandable, but maybe
that's just me.
(Meaning that series s = (LRY .< 10) would also give an error.)
I take your point, but here's my understanding of what's going on
here: In the expression "(LRY .< 10)" LRY is being "promoted" to
matrix status in view of the matrix operator ".<". So the result would
naturally be a matrix, but it can be converted to a series s, per
specification, if it has the right number of elements.
The reason the initial "eval" version above didn't work is that
"(LRY .< 10) * IBO" has the form matrix * series which is not well
defined. But
eval (LRY .< 10) .* IBO
would work OK (giving a column vector) since both series get promoted
to vector status.
So I guess my use of the word "slop" was not really appropriate.
Allin