First argument must be a vector.; Function call failed.

I am trying to conver a function to C using coder. I am using inbuilt function 'filtfilt'. Function executes properly but when i try to convert it to C i get following errors:

1.Function call failed.

2.First argument must be a vector.

3. Undefined function or variable 'W'. The first assignment to a local variable determines its class.

for this line in function:

W= filtfilt(digfilt1,1,y1);

where digfilt1 and y1 are both of type double.

Can anyone help me

[Darshan Ramakant Bhat]

Can you please attache a script to reproduce the issue ?

• Provide the MATLAB function which is giving this error.
• Steps / commands that you are using to generate code for the above function.

This will help us to reproduce the issue.

[Darshan Ramakant Bhat]

What is the value of input y1 in the above function ?

Answer 1

[Darshan Ramakant Bhat]

I can see that you are using MATLAB R2018a. The codegen support for filtfilt is improved significanlty in the later releases.

I tried the below command in R202a and it is working fine for me :

>> codegen -config:lib filtfilt -args {0,1,1:2} -report

Hope this is helpful for you.

[Sohan poul]

oh ok thank you...

Answer 2

[Wilson A N]

Hi Sohan,

In R2018a, the filtfilt function needs its input arguments to be constant or not changing. You can see the 'Extended Capabilities' section which says the same for R2018a.

https://www.mathworks.com/help/releases/R2018a/signal/ref/filtfilt.html?s_tid=doc_ta#d119e60665

Here the variable 'y1' you have used is not constant and its dimensions are also not determined at compile time. Hence you get this error.

Starting from R2019b, filtfilt is enhanced for code generation which will help in these inputs.

Regards,

Wilson

[Sohan poul]

y1 is a set of constant double values imported through .csv file

[Wilson A N]

I see. But since 'y1' is being passed into the function, coder might not have inferred 'y1' as a constant.

If 'y1' is indeed a constant, try putting the values (u can start with a small subset) inside the function and then call codegen on the function containing filtfilt.

Answer 3

[Sohan poul]

I tried it still I get same errors.

[Darshan Ramakant Bhat]

Can you please try in the newer release R2020a or R2019b ? As Wilson mentioned, there were many limitations for this function codegen support earlier.