There is no error, as well SS 6179.
Smath native vectorize is superior to MCD, then more instructions.
Cheers ... Jean

Vectorize Errors in Nested Arrays Doctored.sm (10kb) downloaded 16 time(s).

I think the observed behaviour matches the following hypothesis on how vectorize() operates:
- search for array-valued operators in the expression (before executing any part of the vectorized expression). in your case it is Z.
- Determine length of these operators. In your case length(Z)=4.
- If all non-scalar operators share the same length, then generate a loop for index = 1..length. Otherwise complain about length mismatch. In your case this iterates over the four sub-matrices.
- within the loop, apply the original expression, with the non-scalar expression replaced by the indexed one. In your case extraction of first element and subsequent rounding is applied to each of the four subvectors.
- return a vector with the stacked results of each iteration.

In the second case, the error results from failing to extract element number four from the last two vectors (having less than 4 elements).

The hypothesis doesn't, however, explain the last error "Result is above...".

Radovan, unfortunately your example isn't consistent with my hypothesis
Single and double index el(Z,1) and el(Z,1,1) both return exactly the same result (first submatrix) if applied on the canvas.
So there is no hint why to expect different results when applying vectorize().

Can the developers please shed some light on this?

If you have red or not as demanded, IMHO: the bug is SS 7251.
Revisit the last attached Vectorize Nest_Sub.sm

Thank you Jean. No errors in 7251.


Also many thanks to Martin and Omorr for your comments.

Yes. I also did it using a traditional loop method.
But yours is more elegant.
Thanks.