Can variable size matrices be used as inputs with coder.ceval?

I would like to use from MATLAB a C function which takes as an argument a matrix of variable size and returns the sum of its elements. The resulting code needs to be compatible for use with Matlab coder. The function is defined in the header "my_sum.h":

#ifndef __MY_SUM_H__
#define __MY_SUM_H__

#ifdef MATLAB_MEX_FILE
#include <tmwtypes.h>
#else
#include "rtwtypes.h"
#endif

double my_sum(double * data, int rows, int cols);

#endif

The function is defined inside the file "my_sum.c":

#include "my_sum.h"
double my_sum(double * data, int rows, int cols){
    double tmp=0;
    for (int ind=0; ind<rows*cols; ++ind){
        tmp += data[ind];
    }
    return tmp;
}

I defined the following MATLAB interface (file "call_my_sum.m"):

function out = call_my_sum(mat)
%#codegen
assert(isa(mat, 'double'));

if coder.target('MATLAB')
    out = call_my_sum_mex(mat);
else
    [rows, cols] = size(mat);
    out = double(0);
    out = coder.ceval('my_sum', coder.ref(mat), rows, cols);
end

Finally I created the mex interface "call_my_sum_mex" using the command:

>> codegen -config:mex call_my_sum my_sum.h my_sum.c

The compilation completed but I can use the call_my_sum only with a scalar input. If I run the following code

>> mat = ones(3,3);
>> call_my_sum(mat)

I get the error: MATLAB expression 'mat' is not of the correct size: expected [1x1] found [3x3].

How should I change my code to make it work with an input of any size?

Thanks

Carlo

Answer 1

[Carlo Savorgnan]

I found how to solve my problem. I share the solution for other people running into the same error.

The error can be eliminated by using coder.varsize inside the file "call_my_sum.m":

function out = call_my_sum(mat)
%#codegen
coder.varsize('mat', [inf, inf], [1 1]);
assert(isa(mat, 'double'));

if coder.target('MATLAB')
    out = call_my_sum_mex(mat);
else
    [rows, cols] = size(mat);
    out = double(0);
    out = coder.ceval('my_sum', coder.ref(mat), rows, cols);
end

Answer 2

[Walter Roberson]

You initialized out to a scalar before the call. It needs to initialized to zeros(rows, cols)

[Carlo Savorgnan]

Dear Walter, I didn't get what I should do exactly. Could you be more specific? Thanks Carlo

Answer 3

[upol]

I am trying to convert this simple code into excutable using matlab coder.

function y = hello_world

%#codegen

y = 'Hello World!';

converting to source code C works but when i change the build type to Executable

It gives me this error: Build error: C compiler produced errors. See the Build Log for further details.

C:/PROGRA~3/MATLAB/SUPPOR~1/R2018b/3P778C~1.INS/MINGW_~1.INS/bin/../lib/gcc/x86_64-w64-mingw32/6.3.0/../../../../x86_64-w64-mingw32/lib/../lib\libmingw32.a(lib64_libmingw32_a-crt0_c.o):crt0_c.c:(.text.startup+0x2e): undefined reference to `WinMain'

collect2.exe: error: ld returned 1 exit status

gmake: *** [C:/dummy/hello_world.exe] Error 1

The make command returned an error of 2

'An_error_occurred_during_the_call_to_make' is not recognized as an internal or external command,

operable program or batch file.

Error(s) encountered while building "hello_world":

### Failed to generate all binary outputs.