CODEWITHSUNDEEP
matlabexponential
chilliefiber
chilliefiber

Reputation: 651

Matlab power function randomly malfunctions

tk=linspace(0,1,101);
R = 35.9; L = 3.98; C = 0.000360; P = ceil(R/(2*L));
M = ceil(sqrt( (1/(L*C)) - ( (R^2)/ (4*(L^2)) )));
syms q(t);
syms i(t);
q(t) = 40*exp((-R/(2*L))*t).*cos(t*sqrt((1/(L*C))-((R^2)/(4*L^2))));
i(t) = diff(q);
exact_intensity_values = 1:101;
aux = 1:101;
for ix = 1:101
    exact_intensity_values(ix) = i(tk(ix));
    disp('exact intensity value');
    disp(exact_intensity_values(ix));
    aux(ix) = (exact_intensity_values(ix))^(-1/(P * tk(ix)));
    disp('aux value exact');
    disp((aux(ix))^(-P * tk(ix)));
end

So, sometimes the two values displayed are the same, as it is supposed to be, but sometimes they are very different. What am I doing wrong?

Here are some examples, I won't post all of them. Usually when the exact intensity value is negative the aux value is its symmetric, as in the same absolute value but positive. Other times the numbers have nothing to do with each other. Other times they are the same. I'm completely lost, I've been debugging this for a long time now.

This one is for tk(ix) = 0. It makes no sense whatsoever.

exact intensity value -180.4020

aux value exact 1

Edit - Now that I think about it a bit more it probably makes sense, since tk(1) = 0 we have exact_intensity_values(ix)^(-1/0), and that probably causes the problems here. This can probably be safely ignored.

These ones have the correct absolute value, but incorrect sign.

exact intensity value -422.8061

aux value exact 422.8061

exact intensity value -616.2485

aux value exact 616.2485

These ones are correct: exact intensity value 464.3460

aux value exact 464.3460

exact intensity value 412.2708

aux value exact 412.2708

exact intensity value 337.3326

aux value exact 337.3326

exact intensity value 246.4757

aux value exact 246.4757

These ones have a negative exact value but are correct:

exact intensity value -48.4391

aux value exact -48.4391 + 0.0000i

exact intensity value -132.0104

aux value exact -1.3201e+02 + 2.2673e-14i

exact intensity value -199.6144

aux value exact -1.9961e+02 + 6.9122e-14i

I can't see any pattern in what causes some values to be correct and others incorrect, it seems absolutely random. I've tried both ^ and the power function but the results are exactly the same...

MATLAB Version: 9.5.0.944444 (R2018b)

Edit #2: So the values are only incorrect for the first values of the curve. Here are the graphs, the first one has all the correct values and the second one has the incorrect values as well as some correct. enter image description here

enter image description here

Upvotes: 3

Views: 102

Answers (1)

Dev-iL
Dev-iL

Reputation: 24169

I know you said that you solved this problem, but I'm providing this answer so that the question doesn't remain unanswered.

It appears that the "very different" values simply have a small imaginary component. Otherwise, the real part appears to be the same (within numerical error), and only the notation is different (which is likely the default display format for complex doubles):

  • -1.3201e+02 likely means the same as -132.0104.
  • -1.9961e+02 likely means the same as -199.6144.
  • 0.0000i, 2.2673e-14i and 6.9122e-14i mean 0 practically, but not numerically.

I would advise to surround values that should be real with real().

See also: Why is 24.0000 not equal to 24.0000 in MATLAB?

Upvotes: 1

Related Questions