RE : [eigen] LinSpaced() API

[ Thread Index | Date Index | More lists.tuxfamily.org/eigen Archives ]


Hi,

speaking of this method, it would be useful to have a LinSpace(low, high) version with increment of 1 between two numbers
(maybe in this case the name would be Range) or more generaly something mimicking the start:end and start:incr:end.
It would be especially useful to select rows/cols of a matrix...

Adrien


-------- Message d'origine--------
De: Listengine de la part de Hauke Heibel
Date: jeu. 22/07/2010 14:06
À: eigen@xxxxxxxxxxxxxxxxxxx
Objet : Re: [eigen] LinSpaced() API
 
Done.

On Thu, Jul 22, 2010 at 1:41 PM, Hauke Heibel
<hauke.heibel@xxxxxxxxxxxxxx> wrote:
> Hi Jitse,
>
> I wrote those functions....
>
> On Thu, Jul 22, 2010 at 1:13 PM, Jitse Niesen <jitse@xxxxxxxxxxxxxxxxx> wrote:
>> I'm adding some bits to Tutorial page 5 (advanced initialization).
>
> Thank you for doing this since I am hardly finding any time at the moment.
>
>> While
>> doing this, I noticed the order of the arguments for the static methods
>> Constant() and LinSpaced() for vectors:
>>
>>   Constant(size, value)   versus   LinSpaced(low, high, size)
>>
>> I guess that the order of arguments for LinSpaced() mirrors the order for
>> the Matlab function linspace. However, it does seem inconsistent when
>> compared to the order of arguments for Constant(). I think that consistency
>> within Eigen is more important than consistency with Matlab, so what about
>> changing the order to LinSpaced(size, low, high)?
>
> That's true and I am fine with switching. Last chance for breaking the
> API, right?
>
>> Additionally, I note that there is no fixed-size variant without the size
>> argument. Is there any reason for that?
>
> I cannot recall why I did not implement this. It should be simply to
> add an overload. I can do both changes right now.
>
> - Hauke
>



<<winmail.dat>>



Mail converted by MHonArc 2.6.19+ http://listengine.tuxfamily.org/