Thanks for all the help. This has got me going just fine (at least for now)
Yes I understand my typo: +/@*: should have been +/@:*: Will try things - have an intensive minimization problem and want to see how things compare to "amoeba" which is (for my problem) slow and gets stuck in local apparent minima. lgbfgs used to work find for me under J6. So I miss it under J7. Advantage was that it had derivatives and Hessians which are important for likelihood calculations in statistical problems. One comment re deoptim. Why not define it as an adverb? On Fri, Sep 7, 2012 at 7:31 PM, Ric Sherlock <tikk...@gmail.com> wrote: > By the way, if you use getDEoptim instead of deoptim the output is as > a table with labels in the first column and maybe easier to interpret > i.e. > getDEoptim 'ssq_base_'; _3 3 > > On Fri, Sep 7, 2012 at 8:43 PM, Ric Sherlock <tikk...@gmail.com> wrote: >> Yes Mike has got you going. If you haven't already I'd recommend >> looking through the test script which I've fixed the link to on the >> wiki page for the addon. It has some simple examples there. >> >> http://www.jsoftware.com/jwiki/Addons/math/deoptim >> >> Glad to see that someone else has use for this. >> >> I suspect that the LBFGS routine just needs testing and updating for >> J7. I admit that I never got it going on J6 so am not sure what is >> required. >> >> On Fri, Sep 7, 2012 at 8:23 PM, Mike Day <mike_liz....@tiscali.co.uk> wrote: >>> New to me but I've just had a look, in J602. >>> >>> As you said, there seem to be a number of issues: >>> 1) your provided function needs to return a scalar result. >>> (+/@*:) 2 3 >>> 4 9 >>> (+/@:*:) 2 3 >>> 13 >>> So ssq should be +/@:*: >>> >>> 2) The locale for deoptim starts as pdeoptim >>> One way round appears to be to do >>> deoptim 'ssq_base_';_3 3 >>> ====================== >>> Generation: 1 >>> Best Value: 0.5667 >>> Best Var set: >>> _0.2958 _0.6922 >>> ====================== >>> Generation: 50 >>> Best Value: 0.0000 >>> Best Var set: >>> 0.0000 0.0000 >>> ====================== >>> Generation: 100 >>> Best Value: 0.0000 >>> Best Var set: >>> 0.0000 0.0000 >>> +------------------------+-----------+---- >>> |1.06998e_15 _3.06813e_15|1.05583e_29|2020| >>> | | | | >>> ....... >>> >>> Perhaps that's enough to get you going. >>> >>> As for lbfgs, it's in my addons\math folder >>> under J602. but not under J701. Perhaps the >>> addons gurus can tell you why. >>> >>> Mike >>> >>> >>> On 07/09/2012 5:35 AM, Piet de Jong wrote: >>>> >>>> I 'm trying to use the "deoptim" math addon package. >>>> >>>> However can't figure it out. Tried the following trite example to >>>> understand how things work. >>>> >>>> >>>> NB. Minimize the sum of squares within the range _3 to 3 >>>> >>>> require 'math/deoptim' >>>> >>>> ssq=:+/@*: >>>> deoptim 'ssq';_3 3 >>>> >>>> This fails for a variety of reasons which I cannot seem to grasp/fix up. >>>> >>>> Can anyone help? Once I can figure out a simple example like this I >>>> think I will be able to >>>> implement it on non trite examples. >>>> >>>> Also what has happened to the LBFGS minimization routine? >>>> ---------------------------------------------------------------------- >>>> For information about J forums see http://www.jsoftware.com/forums.htm >>>> >>> >>> ---------------------------------------------------------------------- >>> For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm -- Piet de Jong -------------------------------------------------- View my current research at http://ssrn.com/author=619154 -------------------------------------------------- ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
