On Jun 17, 8:33 am, smichr <[email protected]> wrote:
> I've been working on extending the capabilities of the solver and have
> something that is passing the tests as well as being able to make
> general u-substitutions to turn expressions into polynomials
> (including trig expressions) and expressions with radicals. I am stuck
> right now trying to think of how to do a symmetric test for something
> like 2**x-5**x which has a trivial-like solution of x=0 which sets
> both terms to 1 (not 0) and they cancel. Does anyone have any ideas of
> how to recognize when such an attempt might succeed...and when if
> might fail?

I'm not sure what exactly you mean. a**x - b**x => x = 0 or something
more general?

>
> e.g. Given x + (1-x**2)**(1/2) if we try to see if there are any
> values of x that set both terms to a cancelling constant, we proceed
> as follows:
>
>   we need the value of x that sets x=c and (1-x**2)**(1/2)=-c but
> solving the second term for x gives x=(1-c**2)**(1/2) and if we equate
> that with the negative of the x obtained from x=c we are led back to
> where we started:
>
>    c+(1-c**2)**(1/2) = 0
>
> ==> infinite solving loop.

Could you please post an example where this method actually works? For
radicals I'd choose another strategy (isolating the radical on one
side of the equation, squaring, solving, checking afterwards for false
solutions).

>
> Does anyone understand what I am trying to do and have any suggestions
> on how to accompllish it? Is there a way that this could be extended
> to more than 1 term, perhaps through pattern matching?

I'd try some routine that intelligently isolates terms (if it makes
sense).

Vinzent
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"sympy" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [email protected]
For more options, visit this group at http://groups.google.com/group/sympy?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to