[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: LevenbergMarquardt/summary
It might be better to set this up as a -ml- problem.
The text of Sophia Rabe-Hesketh and Brian Everitt
discusses the two Gaussians case in detail and poses
your case as an exercise, with hints for solution.
> In austrian centre of physics we use Stata to do
> please look at my posting from january 20th with
> short summary :the number of persons who is exposed to a little
> dosis-interval /about 0.01 mSv/ yields in sum a curve which is mostly
> composed of
> two sometimes three humps.above each dosis-interval /about
> 0.01 mSv/there is
> the number
> of persons as function. with Stata it was possible to show that in 397
> cases we have the sum of Gauß-distributions.the results are
> very useful.
> because of lv-command we can talk here about skewness of our curves.
> median-test is also used very often.we decided to take Stata
> to use for our
> work becaue it is so easy to split in subgroups and make tests after.
> My question: if I have two Gauß-curves to be fitted,it is no
> problem to find
> in Stata.but if I have to find three Gauß-curves,I mean to
> come from 1) to
> 2) for
> Ai for i = 1,2.... means amplitude
> A1 12209.2
> µ1 4.382678
> delta1 0.3097696
> A2 21683.69
> µ2 4.200002
> delta2 0.550282
> A1 11509.05
> µ1 4.403567
> delta1 0.3050716
> A2 1305.195
> µ2 4.18422
> delta2 1.100233
> A3 21540.19
> µ3 4.198081
> delta3 0.5104234
> I need long time to try.with another famous statistical
> software I have
> possibility to make
> constraints and it takes no long time.
> I "fear the worst" that someone(my boss?) will ask me during my
> presentation,if there are add-inns in Stata to use or
> user-written-programs that I have
> this feature.
> it is a pity that it is not possible for me to come to 2nd
> German Stata
> User's meeting in April.
* For searches and help try: