Same data - different answers

In the first scenario i type the values in by hand and get this outcome:

In the second scenario i type the values in like this:

And my output is then:

I change nothing more, but these two parameters, but the outputs are completely different. Why is that? I'd rather type the values in by hand, but the second output is the one i expect to have

Answer 1

[Walter Roberson]

In the first case, you are typing in 14 significant digits; MATLAB stores at most 15 1/2 significant digits.

In the second case, the gain is displayed rounded to 2 decimal places. We do not know immediately whether it is only stored to two decimal places, or if the display is rounding. The actual numbers might be up to 15 1/2 significant digits.

In the second case, the coefficients are displayed with 5 significant places. We do not know immediately whether it is only stored to 5 decimal places, or if the display is rounding. The gain was displayed to 5 signicant places, so possibly it is a display artificat. The actual numbers might be up to 15 1/2 significant digits.

At the very least, we run into the possibility that the actual coefficients are stored to 15 1/2 significant digits, but that you only typed in 14 significant digits -- which would be consistent with the possibility that you displayed them with "format long g" turned on and typed what you saw there, instead of using fprintf() to get all 15 1/2 digits.

[Arkadiusz]

Thank you, I didn't find a way to get the numbers from fprintf() function, but guided by your tip, I clicked those three dots when i put the kp(7) then kp (base workspace) -> open and it showed me the full numbers. I entered those numbers and got the desired output. Thank you very much again!

[Walter Roberson]

fprintf('%.1100g\n', kp(7))

will show kp(7) in all of its glory details .

fprintf('%.17g\n', kp(7))

will show kp(7) to enough significant digits to be unique. (There are a few numbers where %.16g would not be unique.)

[Arkadiusz]

That's cool! I will probably use that in future, thanks again!