problem with numbers in for loop that gives NaN

function  [s, a]= importfile(t, v)
N=length(t)
s(1)=v(1)*t(1);
for i=2:N
    s(i)= s(i-1)+ v(i)*(t(i)-t(i-1))
end

This loop works for the 16 first iterations after that I get NaN:

s =
  Columns 1 through 8
         0         0    9.7223   15.3335   21.3891   24.5002   27.6391   32.6669
  Columns 9 through 16
   56.3337   62.5004   85.0560   90.7227  107.9451  131.2785  139.1119  142.9730
  Columns 17 through 24
       NaN       NaN       NaN       NaN       NaN       NaN       NaN       NaN

I have checked the numbers in t and they are defined as the first column in the file running.txt that I uploaded and are all normal numbers and not infinity. I have to deliver an assignment so I am a bit lost. Anyone who can help out? Thanks! :)

EDIT: I added the whole script just in case but it should not matter function

 [s, a]= importfile(t, v)

N=length(t) for i=1:N-1

    a(i)=(v(i+1)-v(i))/(t(i+1)-t(i));

end

s(1)=v(1)*t(1);

for i=2:N

    s(i)= s(i-1)+ v(i)*(t(i)-t(i-1))

end

ta=t([1:1217]); sa=s([1:1217]); plot( ta, sa, 'linewidth', 4)

xlabel('tid')

ylabel('posisjon')

title('posisjon')

figure(2)

ta=t([1:1218]);

plot(ta,a, 'linewidth', 4)

title('akselerasjon')

xlabel('tid')

ylabel('akselerasjon')

[Guillaume]

Can you show the values of t and v that give you this result?

Note that NaN is nothing to do with infinity, it means not a number and is the result of an undefined operation such as dividing 0 by 0, or Infinity by Infinity, etc.

[Stephen23]

Please format your code properly by selecting it and clicking the {} Code button above the text box. And don't put an empty line as every alternate line, it just makes it harder to read.

Answer 1

[Stephen23]

Although the code is very inefficient (see below), it does not generate NaN's for me:

>> s(1:24)
ans =
 Columns 1 through 7:
     0.00000     0.00000     9.72230    15.33350    21.38910    24.50020    27.63910
 Columns 8 through 14:
    32.66690    56.33370    62.50040    85.05600    90.72270   107.94510   131.27850
 Columns 15 through 21:
   139.11190   142.97300   146.72300   163.94540   168.27870   172.61200   190.61200
 Columns 22 through 24:
   202.86190   206.72300   210.58410

Because what you originally uploaded did not import that data file, I had to import it using my own code:

% My code to import data file:
fid = fopen('running.txt','rt');
C = textscan(fid,'%f%f','Delimiter',',');
fclose(fid);
t = C{1};
v = C{2};
% Your original code:
N = length(t)
s(1) = v(1)*t(1);
for k=2:N
    s(k) = s(k-1)+ v(k)*(t(k)-t(k-1));
end

Better Code

That code is extremely inefficient, because it increments an array with every loop iteration. This means MATLAB has to expand its memory allocation on every iteration. SLOW!

The solution is to stop using low-level loops to solve every small problem, and learn how to make use of MATLAB's much more powerful code vectorization abilities. This one line is much faster and more memory-efficient that your loop, and creates exactly the same output:

>> X = cumsum([v(1)*t(1);v(2:end).*diff(t)])
X =

     0.00000
     0.00000
     9.72230
    15.33350
    21.38910
    24.50020
    27.63910
    32.66690
    56.33370
    62.50040
    85.05600
    90.72270
   107.94510
   131.27850
  ... etc

And here is a timing comparison of your loop vs. this vectorized code, for 100 repetitions:

Elapsed time is 55.9212 seconds.    % your loop
Elapsed time is 0.00800109 seconds. % vectorized code