R2025b Update 4 does not support element-wise division (./) for tf objects?

The following should divide 1/(s+1) by 2, and 1/(s+2) by 3:

s = tf('s');
sys1 = [1/(s+1), 1/(s+2)];
sys2 = [2, 3];
result = sys1 ./ sys2;

Error using ./
Invalid data type. Argument must be numeric, char, or logical.

However, I get the following error message:

Error using ./

Invalid data type. Argument must be numeric, char, or logical.

Error in

test (line 4)

result = sys1 ./ sys2;

Answer 1

[Walter Roberson]

The same is true at least as far back as R2018a. I don't think the ./ operator has ever been defined for tf

[Ross]

Wow. This is the first time that I am trying it. I will make a note. Thanks!

[Steven Lord]

The ./ operator is not defined for tf objects, but you can reorder the operations a bit to take advantage of the fact that the .* operator is defined for tf objects.

s = tf('s');
sys1 = [1/(s+1), 1/(s+2)];
sys2 = [2, 3];
A = sys1.*(1./sys2)

A =
 
  From input 1 to output:
   0.5
  -----
  s + 1
 
  From input 2 to output:
  0.3333
  ------
  s + 2
 
Continuous-time transfer function.

[Paul]

I'm quite surprised that operator .* works here. I don't believe that's documented. If .* is really intended for the user, then it's a bit surprising that ./ isn't supported. I'll take a wild guess that .* is intended to be used under the hood by the toolbox functions.

Also, .* doesn't have the same functionality as might be expected insofar as singleton expansion is not supported

M = [tf(1,[1,1]),tf(1,[1,2])]

M =
 
  From input 1 to output:
    1
  -----
  s + 1
 
  From input 2 to output:
    1
  -----
  s + 2
 
Continuous-time transfer function.

try
    M.*[2,3;2,3];
catch ME
    ME.message 
end

ans = 'Model I/O dimensions must agree.'

Note that M is a transfer function matrix, which isn't really an array of models (Model Arrays).

@Ross,

I suspect that a model array is what you really want because .* isn't a defined operation from a control theoretic perspective.