This happens in Kovacic' algorithm to calculate the solution of ode's. When
a,b are constants, they can be absorbed in the constants of integration:
For example,
sol: %k1*a*sin(x) +%k2*b*cos(x)
will be written as:
sol: %k1*sin(x) +%k2*cos(x)
Not strictly necessary, but for some expressions complicated constant factors
can be introduced involving sqrt, sin, etc..
Sometimes, Kovacic' algorithm needs restrictions on a,b as well, like that a,b
are positive integer constants, or that (4-a)/4 is a positive integer
constant. Else it will not be able to tell if a rational solution to the
Ricatti equation exists.
On Saturday, February 09, 2013 12:34:53 AM Richard Fateman wrote:
> On 2/8/2013 2:23 PM, Nijso Beishuizen wrote:
>
> Why are you declaring a and b to be constants?
>
> It seems to provoke a bug. Is (a^2-b>0) now returns unknown.
> Interestingly, declaring only a to be a constant does not break it.
>
> The bug, it seems, is not in atan2, but elsewhere.
> /
> /