I keep getting NaN

For some reason my T values keeps giving me NaN when putting greater values. I dont understand why the error occurs.

clc, clear all


T=0

summa=0;

x=60;%distance x=60 whole distance
s=0;
n=1%amount of steps (change this depending on the x)
route=input('1 or 2:')
%check approx anwser
fun= @(x) 1./(velocity(x,route));

check=60.*integral(fun,0,x) %in min
T=60.*time_to_destination(x,route,n)


while abs(check-T)>=1 %+-0.5 will give a span of 1
T=60.*time_to_destination(x,route,n); %in min
n=n+1;
end
svar=n %antal steg
t=T %integral approx värde (in min)


function T= time_to_destination(x,route,n)
s=0; %start vilkkor
summa=0;
h=x./n; 

for i=0:h:x;
v=1./(velocity(s,route));
v2=1./(velocity((s+h),route));
summa=(h./2).*(v+v2)+summa; %trapets def
s=s+h;

end
T=summa;
end

[John D'Errico]

How can someone help you, if you do not provide sufficient information to do so? For example, what is velocity? (Don't tell me rate of change of distance traveled.) At least, in terms of matlab, what is velocity? It is not defined, so it must be a function that you have kept secret.

A NaN results when one of several indeterminate things happens. For example 0/0, inf/inf, inf-inf, etc. But we have no way to track that down, since we are not you. However, you can learn to use the debugger, which is what someone else would probably do anyway.

[Adam Danz]

NaNs spread like wildfire. Somewhere in your script a NaN is produced and it's likely affecting other variables, too. As John mentioned, we have no change of troubleshooting without being able to run your script.

[John D'Errico]

I'd suggest this command:

dbstop if infnan

Then look at what you have. Where did the inf or NaN arise? What caused it? Look carefully at your code. See that infs can often then turn into NaNs. For example

>> inf/inf
ans =
   NaN
>> inf - inf
ans =
   NaN

So infs can be precursors to NaNs. And once NaNs happen, they just breed like wire coat hangars in back of your closet.

[Daniel Hodgson]

Im sorry that i did not provide sufficient infomation the velocity function is (i thought it might have been a easy fix):

function v=velocity(x, route);

if route==1 
load speed_anna 

elseif route==2
load speed_elsa

end

v=interp1(distance_km,speed_kmph,x);

end

where speed_anna and speed_elsa gives two vectors being distance_km and speed_kmph both provide the vectors though they have different values depending on which one that loads (both values are vector of roughly 1000 elements so i have not included them for now, if needed i can provide them). Both x and route are number that get injected via inputs, route being 1or2 and x can be any positive number.

(PS: In the future i will leave my code less messy and with better descpritions, ignorance and fustration occured which prevented me to such things.)

[Adam Danz]

That's helpful, Daniel Hodgson, but we need to be able to reproduce the NaN values on our end before we can trace back to their origins.

If there are any NaN values in the "v" output, please attach the two files.

If there aren't any NaN values in "v" you could produce v, save it to a mat file, and attach that mat file (or share the two files).

To determine if an array has any NaNs,

any(isnan(v),'all')

[Daniel Hodgson]

i really appreaciate the help. I beleive i have fixed the issue for the for loop i needed to change it to i=0:h:x-h as i was trying to calculate an integral using the trapezoidal method and i beleive i had exceeded the boundries of my integral by including x at to my understanding the x also breaks the methods rules.