Re: [Boost-users] [MultiArray] How to create / handle a collection of multi-array with varying dimensions?

Subject: Re: [Boost-users] [MultiArray] How to create / handle a collection of multi-array with varying dimensions?
From: Oliver Fochler (janitor048_at_[hidden])
Date: 2009-09-28 08:47:34

2009/9/28 Joel Falcou <joel.falcou_at_[hidden]>

> Oliver Fochler wrote:
>
>> Hi,
>>
>> I'm trying to write some dimension independent routine that makes use of
>> Boost.MultiArray containers. The thing is that I would need several such
>> MultiArrays within this routine with the number of MultiArrays depending on
>> the overall number of dimensions.
>> And here is where it becomes complicated (at least for me). Let N be the
>> overall number of dimensions the routine works with. Then I would a
>> collection of N MultiArrays - but not with fixed dimensions but rather with
>> varying (internal) dimension. That is, a collection like this:
>>
>> boost::multi_array<int, N> A
>> boost::multi_array<int, N-1> B
>> boost::multi_array<int, N-2> C
>> ...
>>
>> However, since N may vary when the routine is called I cannot hardcode the
>> array as above. I was thinking of a vector of multi_array or something like
>> that. But then I don't know how to treat the second template parameter in
>> defining the type of the vector (array, ...).
>>
>> Is there a simple and efficient way to do this?
>>
>
> If you know N at compile why not using a fusion::vector of multi_array ?
> _______________________________________________
> Boost-users mailing list
> Boost-users_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/boost-users
>

Well, I don't really know N at compile time. Eventually I'd like to have a
routine that can be called with varying N.
BTW: The algorithm is meant for interpolating values from higher dimensional
grids (N=4 or N=8 for example). Is anyone aware of easy-to-integrate, solid,
fast library solutions to this problem? Linear interpolation would be fine
for starters.

If I was to write the code with fixed N known at compile time (which would
be sort of ok as a first step), how would I instantiate the fusion::vector?

vector< boost::multi_array<int, 3>, boost::multi_array<int,2>,
boost::multi_array<int,1> > test

wouldn't make that much sense, since it would not be flexible at all.

Oliver