* mex memory break - 1 messages. 1 compose * Finding dips in a data set - 1 messages. 1 compose * plot a line ?? - 2 messages. 2 authors * Does anybody know a non-linearly inequality constrainted non-linear least square solver called ENLSIP? - 2 messages. 1 compose * Updating interpret or plan in realtime - 4 messages. 3 authors * prediction using statistics - 6 messages. 2 authors * NI DAQ 6008 Comunication problem - 1 messages. 1 compose * Help! - 1 messages. 1 author * how to deliver mex DLLs to open files? - 1 messages. 1 author * parsing a txt register by keywords - 1 messages. 1 author * GUI Axes_desire an Oscilloscope - 1 messages. 1 author * Meaning??? - 1 messages. 1 author * Wishlist for R2007b - 1 messages. 1 compose * remove mex can't sight my_mexfile mexw32 - 1 messages. 1 compose * Linking a Slider to alter Box - 1 messages. 1 author
"Florian " <f pappenberger@NOSPAMlancater ac uk> wrote in message <fbh18d$6vc$1@fred mathworks com>...> Hello all,> I am trying my first bit of mex code but somehow create a> memory leak - could you gratify help?> > > plhs[0]=mxCreateDoubleMatrix(arr_si,1,mxREAL);> all_values = mxGetPr(plhs[0]);> > /* Read from file */> while(read_from_register_func(&val))> {> all_values[n] = val;> }> > mxSetPr(plhs[0],all_values);> > ?? Do I use mxGetPr & mxSetPr correctly? > Thansk,> flo
% the data (your example) y=[ 3.150 3.200 3.130 3.080 3.100,... 2.980 2.550 2.282 2.220 2.300,... 3.170 3.190 3.150 3.228 3.190,... 3.150 3.200 3.130 3.080 3.100,... 2.890 2.550 2.282 2.266 2.390,... 3.170 3.190 3.150 3.300,... ]; x=1:numel(y);% the engine% .. a macro fun=@(v) find(.. v(1:end-2)>=v(2:end-1) & .. v(2:end-1)<=v(3:end)... )+1;% - this requires the <spline tbx>! nf=5; sf=.75; % <- play with the smoothness xs=linspace(x(1),x(end),5*numel(x)); ys=csaps(x,y,.75,xs).'; id=fun(y); is=fun(ys); [it,it]=histc(xs(is),x); it=it+1; % the result r=[x(it);y(it)]; disp(r); % .. on show line(x,y,... 'marker','o',... 'alter',[1,0,0]); lie(xs,ys); line(xs(is),ys(is),... 'marker','+',... 'linestyle','none'); lie(x(it),y(it),... 'marker','o',... 'markerfacecolor',[0,0,0],... 'linestyle','none');
On Sep 2. 8:50 pm. "Gino Linus" <housing2...@gmail com> wrote:> "Hans Mittelmann" <mittelm...@asu edu> wrote in message>> news:1188791010.709079.252560@y42g2000hsy googlegroups com...>>>> > On Sep 2. 7:12 pm. "Gino Linus" <housing2...@gmail com> wrote:> >> I saw in some previous posts populate mentioned about it.>> >> I've looked into it from the decision channelise - it is very old and there is> >> no> >> documentation no tutorial etc. only Fortran label.>> >> I don't experience how to use it.>> >> If anybody know where to find the user command or documentation it would> >> be> >> great! If there is Matlab interface floating around that's even> >> greater...>> >> Moreover. I would acknowledge if anybody can furnish me some pointers about> >> non-linearly inequality constrainted non-linear least square solvers that> >> more modern?>> >> Thanks a lot!>> > Hi,> > I list it on> > HM>> Thanks Hans! Indeed. I got it from your website but there is no> documentation and userguide...
On Sep 2. 8:50 pm. "Gino Linus" <housing2...@gmail com> wrote:> "Hans Mittelmann" <mittelm...@asu edu> wrote in message>> news:1188791010.709079.252560@y42g2000hsy googlegroups com...>>>> > On Sep 2. 7:12 pm. "Gino Linus" <housing2...@gmail com> wrote:> >> I saw in some previous posts people mentioned about it.>> >> I've looked into it from the decision channelise - it is very old and there is> >> no> >> documentation no tutorial etc. only Fortran code.>> >> I don't experience how to use it.>> >> If anybody know where to find the user guide or documentation it would> >> be> >> great! If there is Matlab interface floating around that's change surface> >> greater...>> >> Moreover. I would appreciate if anybody can give me some pointers about> >> non-linearly inequality constrainted non-linear least square solvers that> >> more modern?>> >> Thanks a lot!>> > Hi,> > I enumerate it on> > HM>> Thanks Hans! Indeed. I got it from your website but there is no> documentation and userguide...
"Lorenzo " <lorenz4matlab@gmail com> wrote in message <fban4t$dvc$1@fred mathworks com>...> Hi all,> I am developing an algorithm to predict a fill profile. To do this. I assume > that the fill in a given time window can be expressed as an analytical > function of the measure. I know that this load is somewhat regular that is it > looks like a sequence of similar shapes. By turn fitting. I am able to > characterize a 1-minute window (300 data points) by using 9 parameters.> I undergo a lot of recorded data and by curve fitting thousands of 1-minute > cycles I can sight the statistical distributions of these 9 parameters. The > distributions are of different kind (some uniform some normal some > logarithmic). Most of the parameters are correlated to each other even if not > perfectly.> Now. I would like to guess the values of these 9 parameters for one window > in the immediate future. I tried generating a set of random values based on > the distributions using copulas to take the correlations into account but this > doesn't really help: my prediction is comfort way off.> > To improve the situation here is my idea. The 9 parameters are actually three > sets of three each set describing ~1/3 of the window: I could determine 3 of > the parameters by curve fitting the first third of the window then use these 3 > values and the distributions to cause the remaining parameters... (I > would thus guess 2/3 of the window but that's ok)> My challenge is: is there a way to do this? > (Of course any other suggestion would be appreciated!)> Thanks,> Lorenzo
"Lorenzo " <lorenz4matlab@gmail com> wrote in message<fbhc77$j19$1@fred mathworks com>...> "Lorenzo " <lorenz4matlab@gmail com> wrote in communicate > <fban4t$dvc$1@fred mathworks com>...> > Hi all,> > I am developing an algorithm to guess a fill profile. To do this. I assume > > that the load in a given measure window can be expressed asan analytical > > answer of the measure. I experience that this load is somewhatregular that is it > > looks desire a grade of similar shapes. By curvefitting. I am able to > > characterize a 1-minute window (300 data points) byusing 9 parameters.> > I undergo a lot of recorded data and by turn fittingthousands of 1-minute > > cycles I can find the statistical distributions of these9 parameters. The > > distributions are of different kind (some furnish somenormal some > > logarithmic). Most of the parameters are correlated toeach other even if > not > > perfectly.> > Now. I would like to predict.
Forex Groups - Tips on Trading
Related article:
http://matlab-blog.blogspot.com/2007/09/25-new-messages-in-15-topics-digest.html
comments | Add comment | Report as Spam
|