Index in position 2 must be positive or logical value error?

I am about to pull my hair out. Why is this giving me an error. I checked my math about nine times and its fine. Either I am insane or MATLAB is giving me the wrong error (I attached the "picture" variable and the "longitude" and "latitude" as .mat files in the google drive link cause they are more than 5 MB). I appreciate any help. https://drive.google.com/drive/folders/1Nw50uOfz_fHVQtmk0awbdKeGn4okiM-X?usp=sharing

FULL_PICTURE = zeros(36001, 18001, 3);
for i = 1:length(latitude(:,1))
    for j = 1:length(latitude(1,:))
        I = 100*(round(longitude(i,j),2)+180) +1;
        F = 100*(round(latitude(i,j),2)+90) +1;
        FULL_PICTURE(I, F, :) = picture(i,j,:);
    end
end

[dpb]

If I had to guess, I'd say you've probably got negative latitude/longitude values and it just happens to be the latitude one that popus up first.

Try

any(latitude<0,'all')

[Ali Almakhmari]

I do have negative latitudes but the equation that calculates the index always ends up positive because my minmum value of latitude is -58 ish.

Answer 1

[Voss]

You'd think that 100*round(x,2) would be an exact integer value and in fact would be the same as round(100*x), but that's not necessarily true:

format long
x = 77.2071507771809848463817615993320941925; % latitude(1,15), more-or-less

x1 = 100*round(x,2)

x1 = 
     7.720999999999999e+03

x2 = round(100*x)

x2 = 
        7721

isequal(x1,x2) % not the same

ans = logical
   0

In particular, rounding to 2 decimal places and then multiplying by 100 is not guarateed to give you an integer:

x1 == floor(x1) % not an integer

ans = logical
   0

But multiplying by 100 first and then rounding to the nearest integer is (?):

x2 == floor(x2) % integer

ans = logical
   1

So when you get a not-exactly-an-integer number and try to use in as an index, you get that error.

Instead try your code like this, where round is done after the multiplication:

FULL_PICTURE = zeros(36001, 18001, 3);
for i = 1:length(latitude(:,1))
    for j = 1:length(latitude(1,:))
        I = round(100*longitude(i,j)) + 18001;
        F = round(100*latitude(i,j)) + 9001;
        FULL_PICTURE(I, F, :) = picture(i,j,:);
    end
end

https://en.wikipedia.org/wiki/Floating-point_arithmetic

[Ali Almakhmari]

https://youtube.com/clip/UgkxQFQaVrY3y5RYafMdq5gg_9pzcyJYJk5b

[Voss]

Hahaha! You're welcome!

[Voss]

@Ali Almakhmari: By the way, a useful approach to solving this type of problem - where the code fails sometimes and you don't see any reason why - is to issue the command

dbstop if error

and then run the code. The code will stop executing when an error happens and give you a chance to inspect the values of the variables, so you can see what causes the error.

An alternative that I tend to use (for no reason other than that's what I'm used to doing) is to wrap the erroneous line in a try/catch with a keyboard in the catch part. That is, something like this:

FULL_PICTURE = zeros(36001, 18001, 3);
for i = 1:length(latitude(:,1))
    for j = 1:length(latitude(1,:))
        I = round(100*longitude(i,j)) + 18001;
        F = round(100*latitude(i,j)) + 9001;
        try
            FULL_PICTURE(I, F, :) = picture(i,j,:);
        catch
            % when the error happens, you'll get control on the
            % command line, so you can see what's going on
            keyboard
        end
    end
end

Answer 2

[Jon]

I think the problem is that you are rounding to 2 decimal place rather than just rounding to an integer. This makes indices for I and F that are not integers

So replace

        I = 100*(round(longitude(i,j),2)+180) +1;
        F = 100*(round(latitude(i,j),2)+90) +1;

with

        I = 100*(round(longitude(i,j))+180) +1;
        F = 100*(round(latitude(i,j))+90) +1;

[Jon]

@Voss approach (multiply by 100 first then round) is better if you need to maintain the extra two digits of precision in your integer value