Reading a file, storing data as variables, but unable to access them?

Original:

Hello! I'm trying to access the variable 't' from my function, which is designed to read the information from a data file. The file has three columns and two headers, so I skip over the two headers and store the columns in the variables t, y, and s. I need to then compute the Jacobian of the 't' value with respect to the initial equation I was given. My issue is, however, when I run my program, I get 'ans' instead of any variable 't', even though I seem to be able to plot the variables read from the file just fine.

function [t, y, s] = read_datafile(filename) 
  fid = fopen(filename, 'r');
  header = fgetl(fid); % Omit first header. 
  header2 = fgetl(fid); % Omit second header. 
  blank_line = fgetl(fid);
  t = []; y = []; s = [];
  counter = 0;
  while (feof(fid) == 0)
    counter = counter + 1;
    line = fgetl(fid);
    data = str2num(line);
    t(counter, 1) = data(:, 1);
    y(counter, 1) = data(:, 2);
    s(counter, 1) = data(:, 3);
  end
  fclose(fid);
  errorbar(t, y, s, 'o'); % Plot the data from the file in an xy-graph with error bars. 
  xlabel('t(seconds)');
  ylabel('Amplitude(V)');

This becomes a problem when I have to move on to the second function, which calculates the Jacobian. It's a user-defined function as opposed to the built-in Jacobian (requirements of the assignment).

function J = jacob(a, t)
  A = [1; 1; 1]
  dyda1 = A(1)*t; dyda2 = A(2)*-10.*t.*sin(t); dyda3 = A(3)*t.^2;
  J = [dyda1, dyda2, dyda3];

These functions are being run on separate m-files. I could compute the Jacobian after running read_datafile('data02.dat') and then entering jacob(A, ans) in the command prompt, but I want to be able to store the value of 'ans' from the previous function as a variable, 't'. How can I do this, so that I can instead write jacob(A, t)?

Thanks in advance for any assistance and insight!

Later edit:

Hey guys, I took a long hard look at my code and figured out how to fix it, thanks to some insightful comments. Thanks for your assistance!

EDIT: I wanted to be really sure that my program worked before posting it, I'll show the corrections I made below.

% Main script:
t = []; y = []; s = [];

% Call the read_datafile function:
[t, y, s] = read_datafile('data02.dat'); 

% Define the model parameter vector alpha (a):
a = [1; 1; 1];

% Compute the Jacobian by calling the jacob function:
J = jacob(a, t);

%%%%%%%%Read Datafile Function - separate m-file %%%%%%%%
function [t, y, s] = read_datafile(filename) 
  fid = fopen(filename, 'r');
  header = fgetl(fid);
  header2 = fgetl(fid);
  blank_line = fgetl(fid);
  t = []; y = []; s = []; 
  counter = 0;
  while (feof(fid) == 0)
    counter = counter + 1;
    line = fgetl(fid);
    data = str2num(line);
    t(counter, 1) = data(:, 1);
    y(counter, 1) = data(:, 2);
    s(counter, 1) = data(:, 3);
  end
  fclose(fid);
  errorbar(t, y, s, 'o');
  xlabel('t(seconds)');
  ylabel('Amplitude(V)');

%%%%%%%%Jacobian Function (user-defined) - separate m-file %%%%%%%%
function J = jacob(a, t)

% Manually compute the partial derivatives for the Jacobian:
dyda1 = a(1)*t; dyda2 = a(2)*-10.*t.*sin(t); dyda3 = a(3)*t.^2;
J = [dyda1, dyda2, dyda3];

[The Matlabinator]

How exactly are you calling read_datafile? If these are functions all in one m-file, I would try using different variable names within my function, i.e,

[tmp1,tmp2,tmp3] = read_datafile(filename);
Jtmp = jacob(A,tmp1);

[per isakson]

Erasing the question makes the thread useless and discourage me (and others) from answering your questions in the future.

[Sarah Malik]

Hi! The problem ended up being really simple and easy to solve--a result of my misunderstanding of how to call functions. I didn't think it would be useful to most people, so I removed it and rewrote the program before posting it again.

[Cedric]

Thanks Sarah, your mistakes will be useful to other people, look at the forum, it is full of ours ;)

[dpb]

In the function

function [t, y, s] = read_datafile(filename) 
  ...
  t = []; y = []; s = []; 
  counter = 0;
  while (feof(fid) == 0)
    counter = counter + 1;
    line = fgetl(fid);
    data = str2num(line);
    t(counter, 1) = data(:, 1);
    y(counter, 1) = data(:, 2);
    s(counter, 1) = data(:, 3);
  end

you read every record one at a time while the data (apparently) are just a delimited numeric text file with a couple of header lines. You build the array by dynamic allocation/reallocation of arrays in the loop. This is a very inefficient programming technique and while it has its place in certain instances that are difficult to solve in more elegant manners, reading a regular text file isn't one of those cases. That's why I recast the subroutine in my earlier posting to use one of the builtin TMW-supplied import functions.

While the above works and won't be too bad performance-wise as long as the files remain small, as dataset sizes grow, the overhead of allocating a variable, then reallocating space for a new variable of larger size and copying that new array back to the original will really show up and if you persist in using such techniques you'll be back wondering why your code takes minutes to read a file and how to fix that.

If you'll look, there's an orange line on the RH margin of the editor screen on those lines -- click on that and you'll get a more complete discussion of the issues and possible ways to solve the problem...for this kind of problem best is to use one of the file import functions that best fits the data structure; they deal with the memory allocation internally for you.

[dpb]

"being really simple and easy to solve--a result of my misunderstanding of how to call functions. I didn't think it would be useful to most people, ..."

But it wasn't so simple before the answer was pointed out, was it? (Else't you wouldn't have needed to post in the first place). Just because end solution is simple doesn't necessarily imply the question isn't worthy of note and removing the question leaves all the responses hanging out of context.

Going forward, make updates/modifications as comments; reserve the Question for the original question with, perhaps Addenda to amplify or clarify questions responders have or to add sample data or the like.

[Rena Berman]

(Answers Dev) Restored edit

Answer 1

[dpb]

"...when I run my program, I get 'ans' instead of any variable 't'"

Because you're calling the function without a LHS into which to store the returned variables from the function; ans is the default Matlab variable for a return value when no variable is given. It also is normal Matlab syntax that all except the first variable of multiple return variables from a function are discarded unless there are explicit variables provided on LHS of the calling expression for them to be assigned into.

I'd rewrite the function more like

function data = read_datafile(filename) 
  data = importdata(filename,);
  data=data.data;
  errorbar(data(:,1), data(:,2), data(:,3), 'o'); % Plot the data from the file in an xy-graph with error bars. 
  xlabel('t(seconds)');
  ylabel('Amplitude(V)');

and return the data as an array instead of three variables. Alternatively, consider using a table, kinda' the best of both; a single structure but with names for variables within.

[Sarah Malik]

Thanks so much for your comment, I'll give it a try!

[Cedric]

Don't forget to accept the answer if it helped.

[Sarah Malik]

I accepted it this morning after thoroughly testing my corrections, thanks for reminding me.