The bug is between Max(M) same/NOT same as max(M)
.................. Min(M) same/NOT same as min(M)

at different high/low values in "M".
"ad absurdum proof" in the work sheet.

Jean

Bug Matrix.sm (27 KiB) downloaded 50 time(s).

Confirmed even in later versions, first function with built-in max() and min() applied to the Cscale is really slow.

Max() & Min() (CustomFunctions) reworked to use the numerical engine -> if used in the first function, it works fast as the robust scale.

New Max/Min solved also this issue; not sure if my previous code buggy or still matrix manipulation issues (output values of Max and Min on the canvas were ok, apparently not in the functions).

Thanks Davide for your attention.

I think the fish is in the mud, as long it does not breath: fine.
Smath 5346 Matrix is definitively and definitely "Buggy".
"Input string !WAS! not in correct format".
The attached proof is evident: the "Zoom replicate" works in
Mathcad 11 but not in Smath. It works fine for the chess board
but not for any other data matrix. It reminds me the TI-58 NoP
[No Operation]. The bug looks at the input string in the
"zoom replicate". It takes integer values, NOT decimal values.

Jean

Bug Matrix More.sm (61 KiB) downloaded 44 time(s).

Hello Jean,

The error message of SS 5346 is strange, however in 0.98 is "Requested matrix element doesn't exist"; following the latest message the error is that in the function Zoom(M, h, v) you are using local unknowns as target of the assignments inside the loops; since you are in a programming-function, and p,N,Z are not defined before the loops, if in the canvas exists a variable with the same name, this variable is used in the function and the values locally replaced (or added if the target element is outside the original sizes). In your case the issue is triggered by Z (defined in the canvas in "Zoom integer").
You can avoid the issue placing Z:0 above the loop (or Z:matrix(rows,cols) with rows and cols as non-negative integers)

Thanks Davide, most appreciated.

I will implement as per your advice. In the mean while
finishing diner, the brilliant idea: decapsulate all
an everything, works fine and zoomed the "Persian Carpet".

There are fractional zoom up [1.25, ...] that would work
in Smath. That will be a waste of technique vs more advanced
zoom smooth: Fourier Convolution, Fourier Padding.

Cheers, Jean

Image Zoom Replicate.sm (61 KiB) downloaded 39 time(s).