help-octave
[Top][All Lists]

## Re: using cellfunction

 From: BOKU Subject: Re: using cellfunction Date: Sat, 13 May 2017 22:02:56 +0200 User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Icedove/45.6.0

```Hey sorry for the late answer,

```
okay now i got it, yes it works like you mentioned! Thank you very much. You really helped me out!
```
for me the following worked.

```
do a lot of repmat and cell2mat and mat2cell so that you get all the variables to the same size eg 1X371 of cell arrays and then apply the function as cellfunction. all the ingoing variables to cellfunction needs to have the same size.
```
size(var1) =  1X371X3

size(var2) =  1X371

size(var3) =  1X371

function[outp]=myfun(var1,var2,var3);

l=dosometing(var1);

domore=l*.var2;

doevenmore=domore-var3;

outp=[doevenmore./domore];

endfunction;

outcellfun=cellfun(@myfunc, var1,var2,var3,'unif',0);

cheers chris

On 2017-05-11 23:49, Juan Pablo Carbajal wrote:
```
```On Thu, May 11, 2017 at 11:47 PM, Juan Pablo Carbajal
```
```On Thu, May 11, 2017 at 10:42 PM, BOKU
```
```Thank you for your response ,

yes I do understand,

but I did already "UniformOutput", false

The problem is the function, as I want to apply a function which needs more
than just the matrix from the cell x{1,1} (for the first cell), no I need
furthermore the data which are stored in   x{3,1}  (for the first cell). And
It seems that I cannot index the matrix. I just dont know what cellfunction
does with the cell.. I mean if I do @mean or @(x) x-1 i could basicly do
mean(cell2mat(x)) and it's done! But if i have a {1,3} cell array in which
{1,1} is a matrix of let's say 300X300X300 and {1,2} is a 1X12 array, and
{1,3} is a an 200,120 array what happens then. but that is exactly what i
want to do I want to applay a function to  {1:end,1} with individual datas
stored in {1:end,3}. And idon't want to loop because it's incredible slow.
It takes about 20 min looping over this cell and I have several hundreds of
them. So I want to find another solution.

x{1,i}=foundimfilt{i};

x{2,i}=ref{i};

x{3,i}=meanval{i};

function [y]= doit(x)

idxl=x(1,1)(:,:,1)>x(3,1)(7)+0.3 | x(1,1)(:,:,1)<x(3,7)-0.3

x(1,1)(:,:,2)>x(1,3)(6)+0.3 | x(1,1)(:,:,2)<x(1,3)(6)-0.3 ...

&x(1,1)(:,:,3)>x(1,3)(8)+0.3 | x(1,1)(:,:,3)<x(1,3)(8)-0.3;

idxl=imcomplement(idxl);

pointsel=medfilt2(idxl,[10,10]);

endfunction

pointselection = cellfun(@doit, x ,"UniformOutput" ,false );

On 2017-05-11 22:11, Juan Pablo Carbajal wrote:

On Thu, May 11, 2017 at 9:41 PM, BOKU

Hello,

I would to apply a function to a 3x371 cell array

so my question would if there is a solution to the following

for i=1:numel(values)                              ####### wrapping elements
to get access in cellfunktion
####### because cellfunction does just deliver the cellaray
wrapobj{1,i}=foundimfilt{i};               #######  I hoped that it will
deliver the 3x1 cell
wrapobj{2,i}=ref{i};
wrapobj{3,i}=meanval{i};

endfor

function [pointsel]= similarpoints(f)

idxl=f(1,1)(:,:,1)>f(3,1)(7)+0.3 | f(1,1)(:,:,1)<f(3,7)-0.3&
f(1,1)(:,:,2)>f(1,3)(6)+0.3 | f(1,1)(:,:,2)<f(1,3)(6)-0.3 ...    ####### so
that i can access the data in the &f(1,1)(:,:,3)>f(1,3)(8)+0.3 |
f(1,1)(:,:,3)<f(1,3)(8)-0.3; ####### function

idxl=imcomplement(idxl);
pointsel=medfilt2(idxl,[10,10]);
endfunction

pointselection = cellfun(@similarpoints, wrapobj,"UniformOutput" ,false );
############ but unfortunately it seems that i don't
get the full  cell

error: f(7): out of bound 1
error: called from
similarpoints at line 3 column 10

is there a solution for the indexing or the delivering the needed
informations for the function. Basicly i want to apply an individual filter
to every cell and I need the information stored in wrapobj{3,i} to apply it
on the matrix stored in wrapobj{1,i}!

wrapobj is a 3x371 cell array.

i would appreciate any help on this

cheers chris

_______________________________________________
Help-octave mailing list
https://lists.gnu.org/mailman/listinfo/help-octave

Hi,
cellfun respects the size of the input, that is

c = {1,4,7,10;2,5,8,11;3,6,9,12};
f=@(x) x-1;
cp = cellfun (f,c)

cp =

0    3    6    9
1    4    7   10
2    5    8   11

If you input is a 3x371 cell, then the output of cellfun will be a
cell of the same size. If the output of the function you evaluate on
each cell element is an array or cell, then you should tell cellfun
that 'UniformOutput' is false, e.g.

f=@(x) linspace(1,x,3);
cp = cellfun (f,c,'UniformOutput', false);
whos cp

Variables in the current scope:

Attr Name        Size                     Bytes  Class
==== ====        ====                     =====  =====
cp          3x4                        288  cell

Total is 12 elements using 288 bytes

```
```cellfun, arrayfun and similar functions apply the function to each
element of the cell. so if the function can process all the elements
in the cell, e.g. different sized matrices, scalars, strings, etc..
then cellfun will just apply it to each element. It is the
generalization of the "element-wise" behavior of functions applied to
arrays, i.e. sin, tanh, exp, abs, etc...

But cellfun can also work in two cells, advancing on each other
simultaneously and implement element-wise functions of many inputs
(n-arity), for you case you could do

cellfun (your_func, your_cell(1,:), your_cell(3,:), 'unif', 0)

Does it work?
```
```you can also do arrayfun on the index...

index = 1:size(your_cell,2);
arrayfun (@(i)your_func(your_cell{1,i},your_cell{3,i}), index, 'unif', 0)
```
```

```