Why do I get: "Error using sum Invalid option. Option must be 'double' " even though my variable is a double?

12 views (last 30 days)
lil brain on 25 Mar 2022
Commented: Walter Roberson on 1 Apr 2022
Hi,
I have this for loop:
n = numel(first_three);
test_first_three = {};
for dd = 1:n
test_first_three{dd} = sqrt(sum(first_three{dd}(:,1:3),1,1).^2);
end
Whenever I try to run it I get the error,
Error using sum
Invalid option. Option must be 'double', 'native', 'default', 'omitnan', or 'includenan'.
It isn't clear to me why that is happening since my cell of doubles array is a double.
What I am trying to achieve with this loop is to apply this function: dist = sqrt( dx.^2 + dy.^2 + dz.^2 ); to every row and its adjacent row.
Here dx, dy and dz are the differences (subtracted from one another) between one row (ex: row 2) and its adjacent row (ex: row 1). That way row 2 - row 1, row 3 - row 2, row 4 - row 3 ... and so on.
Does anyone know how I can solve this?
Walter Roberson on 1 Apr 2022

Walter Roberson on 25 Mar 2022
At the moment you have code which is effectively
sum(x,1,1)
where x happens to have 3 columns.
The first 1 there would give the dimension to sum over (so, sum each column), but the second 1 there is in the slot where MATLAB expects only output class 'double', 'native', 'default', or else the flag about how to treat nans. No numeric value is expected in that third parameter.
Walter Roberson on 25 Mar 2022
temp = first_three{dd}(:,1:3);
sqrt(sum( (temp(2:end,:) - temp(1:end-1,:)).^2, 2))
But you should expect exactly the same problem with secod order derivative later, as subtracting adjacent rows like this is exactly what diff() is going to do (but with a more efficient internal implementation.)