Big O notation for complexity of a program

Hello,

I am writing a paper and have written a program where I have to calculate the time complexity for it. the program has to find the summation of a vector of elements

v=sum(x) where x=[x1^2,x2^,…….,x(n^2)]; where n could be 1,10,100,1000,10000. each element has random value. I have used (tic toc) to find the calculation time but I got strange results; I expect that the time will be increase linearly while changing n from 1 to 10000 but I got graphs as shown in the attachment where results differ for each run. I would like to ask: is the time of summation process varies with respect to the values; for example sum[1.2^+2.3^] is differ from sum[11546.233665^2+9866.89^2] ?

Also, I have did the following test:

d=1000;
VarMin = -10;        
VarMax = 10; 
VarSize=[1 d]; 
x=unifrnd(VarMin,VarMax,VarSize);
tic
sum(x);
toc
disp('%%%%%%%%%%%%%%%%%%%5')
tic
sum(x);
toc

the results is:

Elapsed time is 0.000100 seconds.

%%%%%%%%%%%%%%%%%%%5

Elapsed time is 0.000056 seconds.

So what is the best way is this case to find the compelxity. Thank you in advance

Hazim

[Tommy]

Your results are possibly related to https://www.mathworks.com/matlabcentral/answers/374578-why-does-runtime-decrease-the-more-times-i-execute-a-command and https://www.mathworks.com/matlabcentral/answers/137363-significant-differences-in-execution-time

Consider running the function multiple times for each n rather than just once. i.e. timeit.

[Hazim Nasir]

Thank you very much