How to pass large Arrays into a dotnet library via Matlab?

I am calling a dot net function which works in C# directly but just doesnt work via matlab. I am thinking matlab internally converts something when I pass large data. Note that in the dotnet code the Array that is passed is internally converted inplace. i.e Arr when passed into the function and when returned are different. Arr will be converted by the function and I think this is causing the issue. It would be helpful if you can guide me to resolve this issue.

In the below code if nsamples is a large number like 1000. Then when I create an array and pass the result is

% Result of the above code when nsamples is large number like 1000

No method 'FrequencyCalibration' with matching signature found for class
'xxx.CalculationContext'.


% result when I use nsamples a very small number
methods(ctx, '-full')
[uint32 scalar RetVal, System.Double[] p] FrequencyCalibration(xxx.CalculationContext this, int32 scalar numt, System.Double[] p)


p =

    10    20    30    40    50
    
ctx.FrequencyCalibration(2,2*p)
ans =

  uint32

   524288

The above code directly works in C# just fails to work in Matlab. It would be helpful if someone can guide me in resolving this issue.

thank you for your time.

with regards,

Kartik

Answer 1

[Karan Singh]

Hi Kartik,

From what I understand based on the code and error message you provided, it seems that you are encountering an issue while calling a specific “.NET” function (“FrequencyCalibration”) in MATLAB. The code you shared works correctly in C# but fails to work when called from MATLAB.

Here are a few suggestions to resolve the issue:

1. Array Initialization: Instead of using “NET.createArray” to initialize the “genArr” array. You can try using MATLAB's array creation syntax directly.

For example, you can replace the line genArr = NET.createArray('System.Double', nsamples*probetype); with genArr = zeros(1, nsamples*probetype);.

1. Check the Method Signature: Ensure that the method signature of the “FrequencyCalibration” function in the “.NET” code matches the way you are calling it from MATLAB. Check the number and types of arguments being passed, as well as the return type. Make sure that they match the expected signature in the “.NET” code.
2. Data Type Conversion: Verify that you are correctly converting the data types when passing data between MATLAB and “.NET”. Ensure that the data types of the arguments being passed to the “FrequencyCalibration” function in MATLAB match the expected data types in the “.NET” code.

Attached below are some documentation links that you may find helpful:

• Create array of all zeros - MATLAB zeros (mathworks.com)

Hope this helps!

[Stefan Bilz]

Hello Karan,

Thanks for the help. But the issue got resolved. The issue was the usage of unsigned int. Hence the signature mismatch. I didnot know matlab had specific type casting. After type casting everything worked just fine. By default all variables in matlab are int. we need to cast it to unsigned int. otherwise the dotnet functions donot work if they are using unsigned variables

Thank you for your time. have a good day!!

with regards,

Kartik