Hi Bill,
Thanks for your reply.
In the pde toolbox, under the 'Boundary' option, I used the export decomposed geometry matrix to get 'g' into my workspace.
You were right about the significant digits. I have fixed the problem. Thank you.
"Bill Greene" wrote in message <khv4r7$ht1$1@newscl01ah.mathworks.com>...
> "Sashankh Rao" wrote in message <kht5k2$l97$1@newscl01ah.mathworks.com>...
> > Hi,
> >
> > I create a geometry using the pde toolbox GUI. I then export the decomposed geometry matrix g into the command space. Now I try to use the
> > [p,e,t] = initmesh(g) and it works fine.
> >
> > However, if I copy the same g matrix from the command space to an .m file and try to use initmesh I receive the following error:
> > 
> > ??? Error using ==> pdevoron at 32
> > Geometry error.
> >
> > Error in ==> initmesh at 208
> > [p,t,c,h]=pdevoron(p,t,[],h,x,y,tol,Hmax,Hgrad);
> >
> > Error in ==> mhp at 63
> > [p,e,t] = initmesh(g);
> > 
> > I examined my geometry and found that there was a gap. Theoretically the two points must overlap since they are the same. Is this a visualization problem or is there really a gap in my boundary? How can I close this gap? Why does it work in the command space but not when I use the matrix in a .m file. Thanks. Below is my geometry matrix:
> >
> > g =[
> >
> > 2 2 2 2 2 2 2 4
> > 1 0 0.2242 0.7154 0 0.0950 0.4698 1
> > 1 0.2242 0.7154 1 0.0950 0.4698 0.8447 0.8447
> > 0 0 0 0 0 0.0950 0.4698 0.4698
> > 0.4698 0 0 0 0.0950 0.4698 0.8447 0.8447
> > 1 1 1 1 0 0 0 1
> > 0 0 0 0 1 1 1 0
> > 0 0 0 0 0 0 0 0.4698
> > 0 0 0 0 0 0 0 0.4698
> > 0 0 0 0 0 0 0 0.5302
> > 0 0 0 0 0 0 0 0.5302
> > 0 0 0 0 0 0 0 0
> > ];
> >
> > Thanks,
> >  Sai
>
>
> Hi,
>
> You said above:
> >export the decomposed geometry matrix g into the command space.
>
> How exactly did you do this? pdetool has an option to export gd (geometry
> description matrix), ns and sf into the the command space but I'm not
> aware of a way to export the decomposed geometry matrix.
>
> If you export gd, ns, and sf, you can then call decsg to *compute* the
> decomposed geometry matrix
>
> g = decsg(gd, sf, ns);
>
> Then the g matrix will have full numerical precision so that there are no
> gaps between the edges. The reason for the gaps in your
> decomposed geometry matrix may be that it is represented only
> to four significant digits. If you want to save a text form for
> the decomposed geometry matrix (like you show in your post),
> print the entries with 16 significant digits.
>
> Bill
