In Matlab, the following line of code generates 100000 random numbers from an Alpha-stable pdf (here Alpha=0.5):
Rand = random('Stable',0.5,0,1,0,[1,100000]);
The distribution of Rand exactly matches the the 'theoretical' curve generated by
PDF = makedist('Stable','alpha',0.5,'beta',0,'gam',1,'delta',0);
x = -5:.1:5;
PDF = pdf(PDF,x);
figure
plot(x,PDF,'r-.');
(to check, use, e.g.:)
Data=Rand;
Middle=0.01;
PosBinsUP=10.^(log10(Middle):0.05:log10(max(abs(Data))));
PosBinsDown=10.^(log10(Middle):0.05:log10(abs(min(Data))));
xbins=[-flip(PosBinsDown) -Middle:0.2:Middle PosBinsUP];
[xpdf ypdf]= plotpdfc(Data, xbins);
plot(xpdf(1:end-1),ypdf(1:end-1),'Or');
xlim([-5,5]);
My question is:
How do I generate these Rands in a loop over 't', such that their distribution will expand in time, namely P(Rand) = t^(-1/Alpha) W(Rand/ t^(1/Alpha))?
Aucun commentaire:
Enregistrer un commentaire