Function output produces double array while only a single value is expected

Hello,

I created a fairly straightforward function that accepts lat and lon values in a particular format and converts them into decimal format. Specifically, the input format is a character/string of the style "DDmm.mm ..." where D is degrees and m is minutes. It simply takes the sum of the first two digits (DD) and the minutes (mm.mm ....) divided by 60 to get the lat/lon values in decimals:

function [lat, lon] = latlon_decimal(lat, lon, varargin)

    lat = char(string(lat)); 
    lon = char(string(lon)); 

    lat_deg = str2double(lat(:,1:2)); 
    lon_deg = str2double(lon(:,1:2)); 
    
    lat_min = str2double(lat(:,3:end)); 
    lon_min = str2double(lon(:,3:end)); 

    lat_nans = lat_deg + lat_min/60;
    lon_nans = lon_deg + lon_min/60;
    
    if varargin == "omitnan"
        
        lat = lat_nans(~isnan(lat)); 
        lon = lon_nans(~isnan(lon)); 
        
    else
        
        lat = lat_nans; 
        lon = lon_nans; 
        
    end
   
end

When I use this function with array inputs, it works perfectly. However when I use it with a single-value input, it outputs an array of values I don't understand, but I'm expecting just a single-value output. To give an example:

[lat, lon] = latlon_decimal('1234.5678', '8765.4321')

lat =

  Columns 1 through 7

    0.8167    0.8333    0.8500    0.8667    0.7667    0.8833    0.9000

  Columns 8 through 9

    0.9167    0.9333


lon =

  Columns 1 through 7

    0.9333    0.9167    0.9000    0.8833    0.7667    0.8667    0.8500

  Columns 8 through 9

    0.8333    0.8167

What's really strange about this is that if I just copy/paste all the lines within the function into the command window and run them with the same values, it works perfectly fine:

lat = '1234.5678'; lon = '8765.4321'; 

lat = char(string(lat)); 
lon = char(string(lon)); 

lat_deg = str2double(lat(:,1:2)); 
lon_deg = str2double(lon(:,1:2)); 

lat_min = str2double(lat(:,3:end)); 
lon_min = str2double(lon(:,3:end)); 

lat_nans = lat_deg + lat_min/60;
lon_nans = lon_deg + lon_min/60;

lat = lat_nans; 
lon = lon_nans; 

disp([lat lon])
   12.5761   88.0905
   

I've also run the function with the 'omitnan' condition, and that didn't change the weird result.

So for that reason I expect that there is something wrong with defining the function, but everything in it seems to be okay. Is it possibly something weird about how str2double behaves within a function? I've had issues before where typing >> double('insert number here') will produce an unexpected array of two-digit numbers, and I suspect that isn't an error but more so something I don't understand. All the same, str2double doesn't cause any issues when I use it in the command window.

I assume this is probably something simple that I'm overlooking. Thanks in advance!

[Stephen23]

The reason for those strange values is that your output is simply calculated from the character codes:

V = [0.8167,0.8333,0.8500,0.8667,0.7667,0.8833,0.9000,0.9167,0.9333]

V = 1×9

0.8167 0.8333 0.8500 0.8667 0.7667 0.8833 0.9000 0.9167 0.9333

char(round(60*V))

ans = '1234.5678'

Possibly this is a flaw in your algorithm, or it may be a flaw in how you call it: e.g. it may be a bug that does not correctly handle both strings and (cell arrays of) char vectors (if you attempted to make you code work with both). You did not give enough context to debug this.

[Matt Sprague]

I think I should take a look at character codes more generally. I assumed that using double() or char(), etc. would simply convert from one type to another - i.e. you could write

double('123456')

and simply get an output of 123456. (And that this would work as long as your argument had no letters, symbols, etc.)

Either way the method you provided works well, much appreciated.

Answer 1

[Stephen23]

I would avoid all of those superfluous and inefficient type conversions and messing around with indexing into character arrays. Just convert all of the values to numeric at once and then efficiently perform the required adjuments (as numeric operations). It is quite simple and much more efficient, e.g.:

[lat, lon] = latlon_decimal('1234.5678', '8765.4321')

lat = 12.5761

lon = 88.0905

[lat, lon] = latlon_decimal('0123.4567 1234.5678 2345.6789', '9876.5432 8765.4321 7654.3210')

lat = 1×3

1.3909 12.5761 23.7613

lon = 1×3

99.2757 88.0905 76.9053

function [lat, lon] = latlon_decimal(lat, lon, varargin)
lat = [1,1/60]*sscanf(lat,'%2d%f',[2,Inf]);
lon = [1,1/60]*sscanf(lon,'%2d%f',[2,Inf]);
if any(strcmpi(varargin,'omitnan'))
    idx = isnan(lat)|isnan(lon);
    lat = lat(~idx);
    lon = lon(~idx);
end
end

[Matt Sprague]

Thank you, this seems to work. I had never seen sscanf before and probably would have used it if I had known.