.mat files not loading correctly

Hi all,

I am very new to MATLAB, and the first program i wrote which loaded a file loaded a file named data_vector. Now, whenever I load any file it gets automatically stored in a variable of name data_vector. This means that the following code will not compile on my copy of MATLAB R2016b.

clear
load test_data;

plot(test_data(1,:)) %test_data is a 1000x500 2D array

Please help!

Edit: the specific error message is

Reference to a cleared variable test_data.

Error in phase_detector (line 4)
plot(test_data(1,:))

[Matt J]

It's a little too hard to diagnose from the code shown. Modify it as follows and report all output to us.

clear
disp 'Contents of file:',
whos -file test_data
load test_data;
disp 'Contents of workspace:',
whos
plot(test_data(1,:)) %test_data is a 1000x500 2D array

[dpb]

load will retrieve the variable(s) stored in the .mat file; they may or may not be named the same as the filename used to store them.

Show us the result of

whos
whos -file test_data.mat

in the workspace...the first will show the variables in the workspace and their size and class; the second the content of the file on disk.

It looks like you may have inadvertently re-SAVEd the file after having cleared the variable so that it's empty; the second above will show you what is actually in the file.

[Daniel Guilbert]

after replacing the entire code with what @Matt J provided, the output is:

>> phase_detector
Contents of file:
  Name                Size               Bytes  Class     Attributes

    data_vector      1000x500            4000000  double              

Contents of workspace:
  Name                Size               Bytes  Class     Attributes

    data_vector      1000x500            4000000  double              

Reference to a cleared variable test_data.

Error in phase_detector (line 7)
plot(test_data(1,:)) %test_data is a 1000x500 2D array

[Daniel Guilbert]

I just realised my OP is incorrect, data_vector isn't the name of the first file i loaded - sorry! that was data_matrix.

I generated the test_data file using another script which stores the data in a file called data_vector, but at the end of the script I do:

save test_data data_vector

Sorry for the mistake, hopefully this makes the problem a lot less odd

[dpb]

The only problem appears to be your expectation...the variable is named data_vector, the file you wrote it to is test_data.mat.

Naming a file doesn't change the variable name inside it; the

whos -file test_data

line clearly reveals the content of the file is, as you asked it to be, the array data_vector. When you load the file, the variable(s) in the file are restored to the workplace with their original name.

This is all in

doc save
doc load

There simply is no variable test_data in sight here...if you want that as a variable name you'll have to assign something to it to make it such.

[Daniel Guilbert]

So what's the point of being able to save them under different names if you can only ever reference them using the initial name? Is there anyway to change the name of a variable besides cloning? Seems a bit weak.

[dpb]

This doesn't save them under different variable names, only as a different file; that's the point. SAVE/LOAD is primarily a shorthand for precisely that purpose; save/retrieve a given variable or set thereof for quickly being able to reproduce a snapshot of a previous session or the like without having to worry about extraneous code.

If you really want to save a set of data but read it back in a more general sense, use something other than SAVE (like fwrite/fread for unformatted) that doesn't include such additional information with it.

Or, as Steven notes, if you want to force rename a variable from a .mat file, there's always the structure route albeit it is a fair amount of overhead often compared to using simple arrays.

Answer 1

[Steven Lord]

1. Calling clear to clear all variables is rarely necessary in a function.
2. If you call load with an output argument, the output argument will be a struct array and the variables that were in the MAT-file will be read into fields in that output argument. This allows you to avoid conflicts with variables that already existed in the workspace. I recommend you use this approach; that way the name of the variable you have to access to use your data is predictable. You can use dynamic field names in conjunction with fieldnames if you don't know the names of the variables in your MAT-file but need to iterate over them.

[Daniel Guilbert]

Thankyou.