[SciPy-dev] Trouble with optimize.fmin_ncg
Tue Jul 24 15:23:26 CDT 2007
I still have no idea...
I checked small example (below), it works ok
I don't know, maybe I did something wrong, but I removed
/site-packages/scipy, re-installed the package from
http://packages.ubuntu.com/feisty/python/python-scipy (ver 0.5.2), and
your problem still makes my CPU hanging on.
can anyone who didn't update scipy from svn during last 10-11 days run
>>> f = lambda x: ((x-arange(len(x)))**2).sum()
>>> fprime = lambda x: 2*(x-arange(len(x)))
>>> r = optimize.fmin_ncg(f, *50, fprime)
Optimization terminated successfully.
Current function value: 0.000000
Function evaluations: 3
Gradient evaluations: 4
Hessian evaluations: 0
array([ 0., 1., 2., 3., 4., 5., 6., 7., 8., 9., 10.,
11., 12., 13., 14., 15., 16., 17., 18., 19., 20., 21.,
22., 23., 24., 25., 26., 27., 28., 29., 30., 31., 32.,
33., 34., 35., 36., 37., 38., 39., 40., 41., 42., 43.,
44., 45., 46., 47., 48., 49.])
Nils Wagner wrote:
> On Tue, 24 Jul 2007 18:33:51 +0300
> dmitrey <firstname.lastname@example.org> wrote:
>> Nils, are you sure that troubles raised after last svn
> Well, I believe it is less than 3 month ago that
> ncg worked as expected.
>> All my changes are in func _cubicmin from optimize.py
>> but when I placed a breakpoint there, the hanging cycle
>> didn't reached
>> the one.
>> Can't you do the same trick?
>> line 309,
>> d1 = empty((2,2))
>> I have found the hanging cycle (optimize.py, line 1030,
>> while numpy.add.reduce(abs(ri)) > termcond: )
>> but numpy.add.reduce(abs(ri)) is constantly growing
>> maybe you had changed x0 and now it's too far from
> Even if I start with a vector near x_opt ncg hangs.
>> btw if 2nd derivatives are not supplied, then other
>> cycle is hanging:
>> line 1013:
>> while (numpy.add.reduce(abs(update)) > xtol) and (k <
>> I don't know howto fix the problem.
>> Please inform me about the breakpoint.
>> BTW your func seems to be very suspicious to me
>> def R(v):
>> rq = dot(v.T,A*v)/dot(v.T,B*v)
>> res = (A*v-rq*B*v)/linalg.norm(B*v)
>> return rq
>> are you sure that the func(v)=dot(v.T,A*v)/dot(v.T,B*v)
>> is convex?
>> I'm not.
>> So using 2nd derivatives (or their approximating by
>> fmin_ncg (if user
>> didn't provide that ones) , in line 1033:
>> Ap =
>> will handle non-convex funcs much more bad than 1-st
>> order do.
>> HTH, D.
> Scipy-dev mailing list
More information about the Scipy-dev