Thread Subject: Reducing matrix size by finding similar elements

I have read CFD contaminant data into Matlab. I would like
to reduce the size of that data by finding groups of points
with similar concentrations. Is there a command for finding
like elements in a matrix? Please help, thanks!

"Lisa Chen" <yhc22@drexel.edu> wrote in message
<fm64na$53s$1@fred.mathworks.com>...
> I have read CFD contaminant data into Matlab. I would
like
> to reduce the size of that data by finding groups of
points
> with similar concentrations. Is there a command for
finding
> like elements in a matrix? Please help, thanks!

Could you provide a bit more detail? I don't anything about
CFD -- could you just give, say, an example of a 5x5 matrix
with a before and after shot? Say you want to turn:

[1 2 2.0001 3] -> [1 2 3]

Is that what you mean by finding like elements?

If that's the case, maybe something like:

>>data=[1.83 2.845 2.001 3.2343 3.24 3.25 3.26];
>>roundData=round(data/0.1)*0.1; % rounds to nearest tenth
>>unique(roundData)
ans =
    1.8000 2.0000 2.8000 3.2000 3.3000

"Ian Clarkson" <ian.clarkson@gesturetek.com> wrote in message
<fm65km$o1h$1@fred.mathworks.com>...
> "Lisa Chen" <yhc22@drexel.edu> wrote in message
> <fm64na$53s$1@fred.mathworks.com>...
> > I have read CFD contaminant data into Matlab. I would
> like
> > to reduce the size of that data by finding groups of
> points
> > with similar concentrations. Is there a command for
> finding
> > like elements in a matrix? Please help, thanks!
>
> Could you provide a bit more detail? I don't anything about
> CFD -- could you just give, say, an example of a 5x5 matrix
> with a before and after shot? Say you want to turn:
>
> [1 2 2.0001 3] -> [1 2 3]
>
> Is that what you mean by finding like elements?
>
> If that's the case, maybe something like:
>
> >>data=[1.83 2.845 2.001 3.2343 3.24 3.25 3.26];
> >>roundData=round(data/0.1)*0.1; % rounds to nearest tenth
> >>unique(roundData)
> ans =
> 1.8000 2.0000 2.8000 3.2000 3.3000


Suppose D is the number of accurate digits, you can also do
this by using vpa function

>> data=[1.83 2.845 2 2.001 3.2343 3.24 3.25 3.26]
>> d = 2;
>> double(unique(vpa(data,d)))
            
ans =

   1.8000 2.0000 2.8000 3.2000 3.3000


>> d = 3;
>> double(unique(vpa(data,d)))

ans =

    1.8300 2.0000 2.8400 3.2300 3.2400
3.2500 3.2600


Anh Huy Phan
RIKEN - BSI

This helps a lot, thanks Anh!

Lisa


"Anh Huy Phan" <phananhhuy@mathworks.com> wrote in message
<fm7gdg$km0$1@fred.mathworks.com>...
> "Ian Clarkson" <ian.clarkson@gesturetek.com> wrote in
message
> <fm65km$o1h$1@fred.mathworks.com>...
> > "Lisa Chen" <yhc22@drexel.edu> wrote in message
> > <fm64na$53s$1@fred.mathworks.com>...
> > > I have read CFD contaminant data into Matlab. I would
> > like
> > > to reduce the size of that data by finding groups of
> > points
> > > with similar concentrations. Is there a command for
> > finding
> > > like elements in a matrix? Please help, thanks!
> >
> > Could you provide a bit more detail? I don't anything
about
> > CFD -- could you just give, say, an example of a 5x5
matrix
> > with a before and after shot? Say you want to turn:
> >
> > [1 2 2.0001 3] -> [1 2 3]
> >
> > Is that what you mean by finding like elements?
> >
> > If that's the case, maybe something like:
> >
> > >>data=[1.83 2.845 2.001 3.2343 3.24 3.25 3.26];
> > >>roundData=round(data/0.1)*0.1; % rounds to nearest
tenth
> > >>unique(roundData)
> > ans =
> > 1.8000 2.0000 2.8000 3.2000 3.3000
>
>
> Suppose D is the number of accurate digits, you can also
do
> this by using vpa function
>
> >> data=[1.83 2.845 2 2.001 3.2343 3.24 3.25 3.26]
> >> d = 2;
> >> double(unique(vpa(data,d)))
>
> ans =
>
> 1.8000 2.0000 2.8000 3.2000 3.3000
>
>
> >> d = 3;
> >> double(unique(vpa(data,d)))
>
> ans =
>
> 1.8300 2.0000 2.8400 3.2300 3.2400
> 3.2500 3.2600
>
>
> Anh Huy Phan
> RIKEN - BSI
>
>

Thanks, Ian!!!

"Ian Clarkson" <ian.clarkson@gesturetek.com> wrote in
message <fm65km$o1h$1@fred.mathworks.com>...
> "Lisa Chen" <yhc22@drexel.edu> wrote in message
> <fm64na$53s$1@fred.mathworks.com>...
> > I have read CFD contaminant data into Matlab. I would
> like
> > to reduce the size of that data by finding groups of
> points
> > with similar concentrations. Is there a command for
> finding
> > like elements in a matrix? Please help, thanks!
>
> Could you provide a bit more detail? I don't anything
about
> CFD -- could you just give, say, an example of a 5x5
matrix
> with a before and after shot? Say you want to turn:
>
> [1 2 2.0001 3] -> [1 2 3]
>
> Is that what you mean by finding like elements?
>
> If that's the case, maybe something like:
>
> >>data=[1.83 2.845 2.001 3.2343 3.24 3.25 3.26];
> >>roundData=round(data/0.1)*0.1; % rounds to nearest tenth
> >>unique(roundData)
> ans =
> 1.8000 2.0000 2.8000 3.2000 3.3000