Prev: Stand alone application (gui) crashes with unexpected java error
Next: pcmcia parallel port emulator
From: Jonas Baumann on 7 Jan 2010 08:56
When doing simulation I often end uf with a data structures along the line of:

MyData:
Value1:
Value: 1x10000
Time: 1x10000
Value2:
Value: 1x10000
Time: 1x10000
..
..
..
ValueN:
Value: 1x10000
Time: 1x10000

What I often need to do is extract the same time window out of all values, i.e. from vectors only indizes 1000:2000.
Is there a quick solution to this for large N (10-100)?
From: Andy on 7 Jan 2010 09:06
"Jonas Baumann" <baumann.jonas(a)gmail.com> wrote in message <hi4p5i$8vv$1(a)fred.mathworks.com>...
> When doing simulation I often end uf with a data structures along the line of:
>
> MyData:
> Value1:
> Value: 1x10000
> Time: 1x10000
> Value2:
> Value: 1x10000
> Time: 1x10000
> .
> .
> .
> ValueN:
> Value: 1x10000
> Time: 1x10000
>
> What I often need to do is extract the same time window out of all values, i.e. from vectors only indizes 1000:2000.
> Is there a quick solution to this for large N (10-100)?

If this is your goal, then you should form a structure array from the beginning. That is, instead of MyData with fields Value1, Value2, etc., have MyData be a structure array with MyData(1) storing the value and time for Value1, MyData(2) storing value and time for Value2, and so on. Then I think (I'm not at MATLAB at the moment) to pull out the time, you would just say [MyData(:).Time].
From: Jonas Baumann on 7 Jan 2010 09:15
"Andy " <theorigamist(a)gmail.com> wrote in message <hi4poq$ob8$1(a)fred.mathworks.com>...
> "Jonas Baumann" <baumann.jonas(a)gmail.com> wrote in message <hi4p5i$8vv$1(a)fred.mathworks.com>...
> > When doing simulation I often end uf with a data structures along the line of:
> >
> > MyData:
> > Value1:
> > Value: 1x10000
> > Time: 1x10000
> > Value2:
> > Value: 1x10000
> > Time: 1x10000
> > .
> > .
> > .
> > ValueN:
> > Value: 1x10000
> > Time: 1x10000
> >
> > What I often need to do is extract the same time window out of all values, i.e. from vectors only indizes 1000:2000.
> > Is there a quick solution to this for large N (10-100)?
>
> If this is your goal, then you should form a structure array from the beginning. That is, instead of MyData with fields Value1, Value2, etc., have MyData be a structure array with MyData(1) storing the value and time for Value1, MyData(2) storing value and time for Value2, and so on. Then I think (I'm not at MATLAB at the moment) to pull out the time, you would just say [MyData(:).Time].

Thanks for the reply.
This would work, correcto.
However, I often prefer using substructeres the way I described it because you immediatly know which subvector has which meaning. Numbering them is often easier, but for me, not always.
Any other solutions?
From: Andy on 7 Jan 2010 09:39
But ValueN is just numbering as well. In any case, your problem can be easily solved in a loop if you need to keep your structure like this:

% generating some data
for ix=1:100
s.(['v' num2str(ix)]).val=ix;
s.(['v' num2str(ix)]).time=rand;
end

% extracting the times
T=zeros(1,100);
for jx=1:100
T(jx)=s.(['v' num2str(jx)]).time;
end

You'll need to slightly modify this to extract time vectors instead of single numbers.
From: Jan Simon on 7 Jan 2010 10:24
Dear Jonas!

> Value1:
> Value: 1x10000
> Time: 1x10000
> Value2:
> Value: 1x10000
> Time: 1x10000
> ...
> ValueN:
> Value: 1x10000
> Time: 1x10000

You have to decide if you want descriptive fieldnames and substructs *or* fast access to certain subvectors of the data.
Nevertheless, you could copy(!) all subvectors together to a big matrix:
S.a.time = rand(1, 10);
S.b.time = rand(1, 10);
S.b.time = rand(1, 10);
SCell = struct2cell(S);
SArray = cat(1, S{:});
TimeMatrix = cat(1, SArray.time)

This creates several copies of your data, so it may polute your memory. Some CLEAR command may help.

Good luck, Jan
 |  Next  |  Last
Pages: 1 2
Prev: Stand alone application (gui) crashes with unexpected java error
Next: pcmcia parallel port emulator