On 10/4/10 7:52 AM, Raymond Toy wrote:
>>
>> What am I doing wrong?
> Nothing, except that your initial starting point causes the lbfgs
> algorithm to try to compute the derivative at r = 0, which causes a
> division by zero. If you choose a different starting point, say, 0.8,
> lbfgs converges to 0.436.
>
> Maxima should check for that possible error condition and return a noun
> form.
I think that's ok for a symbolic routine, but I think numeric routines
should do something more. I propose that we catch the error and signal
an error.
Any objections?
Ray