Dot indexing is not supported for variables of this type.

Error in ferroskyrPos (line 36)

M=A.data;

[Sudhakar Shinde]

can you add 'CoPd-JT8b.out' file also so that we can check A=importdata(infile,' ',28) contains a proper structure.

Error may say that A is not a structure and hence we could not use dot indexing.

[Sateesh Kandukuri]

Ok sir, I attached a new matlab file along with the out put file to process.

[Sateesh Kandukuri]

Dear Sudhakar Shinde,

I am a novice to Matlab, I am still in the learning stage. Could you please resolve the error?

Answer 1

[Bjorn Gustavsson]

When you get a problem like this it is because the variable is not of a struct as you expect. In general you can add a line with something like this:

whos A

just after your importdata-call. That will tell you what the variable is.

Another thing to do is to set a debug-stop like this:

dbstop if error

and rerun your script/function, then matlab will stop at the point of error, giving you access to the workspace where the error occured, so that you can have a look at the variables with interactive keyboard commands, you can even step up and down in the call-stack to

investigate how you ended up where you ended up. (This is a necessary skill to learn, not a funny one to do, but as someone said "to err is human"...)

HTH

[Sateesh Kandukuri]

As a novice to Matlab, I am still in the learning stage. Could you please resolve the error?

[Walter Roberson]

Your code is expecting importdata() to return a structure, with the numeric data in a field named data. However, importdata() is instead returning a cell array of character vectors.

importdata() is convenient when it works, but it can return any of three different output formats, and that makes it undesirable for writing robust programs.

The files are a mix of a text header and binary data. The text part of it has field names then a colon and then data, but the data is not in a consistent format (some of it is text, some of it has one numeric value, some of it more than one numeric value.) I therefore recommend that you use low-level file I/O operations such as fopen() and fgetl() and sscanf() or regexp() in order to analyze the input lines.

[Sateesh Kandukuri]

Dear Walter Roberson,

Thank you so much for the reply. But, How it is working for the person who shared this programe to me?

[Bjorn Gustavsson]

Unfortunately not, since when I tried to load one of your data-files matlab crashed. But I can help you learn how to solve these types of problems - that would be better for you in even the medium short-term.

Start with running the lines in your script line-by-line (copy and paste works fine), then instead of running the whole loop just set jj to 1, run the next couple of lines, then set t to one and run the 3 lines down to:

    A=importdata(infile,' ',28);

After that at the command-line type:

    whos A

and check what data-type that variable has. If it for example is a cell-array one of its components might

be a struct that you think you should have. In that case you have to extract that cell-element into another variable. Or something else depending on what A is.

HTH

[Walter Roberson]

How it is working for the person who shared this programe to me

Over the years, importdata() has changed how it detects what is in a file.

Also, you are perhaps not using the same files they are.

[Sateesh Kandukuri]

Dear Walter Roberson,

The similar .ovf files, I am using for the post processing of the output data.

[Walter Roberson]

https://pypi.org/project/OOMMFTools/ is python tools that can convert ovf files into .mat files.

[Walter Roberson]

https://math.nist.gov/oommf/doc/userguide11b2/userguide/vectorfieldformat.html defines the structure of ovf files, by the way.