Hi. I don't know if this is a know one. I guess they are related to the fact that functions with if can neither be derived nor integrated. Notice that seems that there are a bug in cases() too.

FunctionBug.sm (13 КиБ) скачан 29 раз(а).
FunctionBug.pdf (48 КиБ) скачан 27 раз(а).

Best regards.
Alvaro.

... and issues in composite functions with loops.

FunctionBug2.sm (17 КиБ) скачан 29 раз(а).
FunctionBug2.pdf (76 КиБ) скачан 30 раз(а).

Best regards.
Alvaro.

Wrote

... and issues in composite functions with loops.

FunctionBug2.sm (17 КиБ) скачан 29 раз(а).
FunctionBug2.pdf (76 КиБ) скачан 30 раз(а).

Best regards.
Alvaro.

Debugging with trace(), what I see is that when calling g, f takes phi=1, (this is the first phi=1 in the output window) but don't do the loop. Calling f isolated shows another phi=1 and subsequent values for phi.



With for(3) gives an error, so I can't debug that. Additionally, calling with for(4) seems to have the same behavior than while.

Wrote

I don't know if this is a know one. I guess they are related to the fact that functions with if can neither be derived nor integrated. Notice that seems that there are a bug in cases() too.

1. You must assign "cases" like in your last example.
2. Before Davide introduced "cases" it was a damned job integrate
unless from discrete finite differences like in the old days.
The min(x) style [undocumented] is quite smart, mostly OK Simpson.

Cheers ... Jean

Integrate Discontinuous [min(x)].sm (16 КиБ) скачан 27 раз(а).

As I've already mentioned here and there, assignments shouldn't be made inside cases() arguments, because:

Wrote

Please be careful however, since a symbolic evaluation to show the function content or a failed evaluation with an error triggered on the conditions leads to the evalution of all the options.
In general currently is better to use cases to host just dead content (numbers, strings, variables).

This is a downside that currently can't be prevented. As you can see making the assignment outside cases() works as expected.