Thank you.
Actually the model object was my first try, but the 'bad command name'
error made me change the plans. every time I defined something using
"gf_model_set ( modelM, ‘add nonlinear term’)" that error occurred. then I
saw the 'add nonlinear generic assembly brick', but in the matlab_interface
guide is defined as deprecated, so finally i'm trying with the assembly way.
So the plans now are to create the Sparse Matrix SM = gf_spmat('empty', m);
add the bricks one by one, gf_spmat_set(SM, 'add',...) , create the rhs
vector; and then solve = gf_linsolve('superlu', SM, vec_b). Can this work?
Thank you again, for your time and disponibility
Nicolò Cambiaso
Il giorno mer 2 ott 2019 alle ore 22:29 Yves Renard <
yves.ren...@insa-lyon.fr> ha scritto:
>
> Dear Nicolò,
>
> The command gf_asm just allows to compute some different matrices. Is you
> use this strategy, then this is your own responsability to make a
> resolution with the corresponding matrices. You can also use the more
> integrated model object which will agregate all the terms and propose
> directly a solve. For that, you can have a look to the tutorial, for
> instance http://getfem.org/tutorial/thermo_coupling.html
>
> For source term, you can still use gf_asm('generic', ...) yes, with the
> option gf_asm('generic', mim, 1, ...), the "1" indicating that this is a
> vector assembly, not a matrix one. You can also use gf_asm('volumic
> source', ...)
>
> Best regards,
>
> Yves
>
>
>
> ----- Mail original -----
> De: "Nicolò Cambiaso" <cambiaso.nic...@gmail.com>
> À: "yves renard" <yves.ren...@insa-lyon.fr>
> Cc: "getfem-users" <getfem-users@nongnu.org>
> Envoyé: Mercredi 2 Octobre 2019 15:58:16
> Objet: Re: [Getfem-users] Generic Bricks definition using gf_asm.
>
> I'm sorry to bother you again, but a couple of questions raised.
>
> Some of the blocks are symmetric, so in my matrix i'll have the transposed
> corresponding block. Do I need to create the transposed block too with the
> command gf_asm('generic' ...) or it's going to do it by himself?
>
> Also,
> From the equation, I have a term comprendig a test function and some static
> initial data defined in the whole mesh, but no variables ('A*test_pv').
> This term is gonig to be in the Right Hand Side of the linear system. Now,
> my question is: which one of the functions (gf_asm('function', ...)) is the
> correct one to use? Is again 'generic' going to understand it? Right now i
> bet on 'volumic source', but i defined nowhere which test function i'm
> using, so i'm not so sure about that.
>
> Thank you for your help and your time.
> Have a nice day,
> Nicolò Cambiaso
>
> Il giorno mer 2 ott 2019 alle ore 09:37 Nicolò Cambiaso <
> cambiaso.nic...@gmail.com> ha scritto:
>
> > Thank you very much, now it works.
> > to completeness to the future readers, I would like to correct a typo I
> > made when asking for the command.
> > K= --equation--
> > gf_model_set(md, 'add initialized data', 'K', [K]);
> > gf_asm('generic', mim3, 2, 'K*(Test_ut.*ut)', -1, 'ut', 1, mf3,
> > zeros(gf_mesh_fem_get(mf3, 'nbdof'),1), 'K' , 0, K);
> >
> > Again, thank you,
> > Have a nice day
> > Nicolò Cambiaso
> >
> > Il giorno lun 30 set 2019 alle ore 22:19 Yves Renard <
> > yves.ren...@insa-lyon.fr> ha scritto:
> >
> >>
> >> Dear Nicolo Cambiaso,
> >>
> >> You can use gf_asm to produce a mass matrix as you tried, but you have
> to
> >> specify a fem on which the variable is described. The correct syntax is
> for
> >> instance
> >>
> >> gf_asm('generic', mim3, 2, 'Grad_Test2_ut.Grad_Test_ut', -1, 'ut',1, mf,
> >> zero(gf_mesh_fem_get(mf, 'nbdof'),1));
> >>
> >> where mf is the mesh_fem structure describing the fem.
> >>
> >> Best regards,
> >>
> >> Y.R.
> >>
> >> ----- Mail original -----
> >> De: "Nicolò Cambiaso" <cambiaso.nic...@gmail.com>
> >> À: "getfem-users" <getfem-users@nongnu.org>
> >> Envoyé: Lundi 30 Septembre 2019 12:25:51
> >> Objet: [Getfem-users] Generic Bricks definition using gf_asm.
> >>
> >> Dear users,
> >> I would like to solve a system of equations using Getfemm++ with Matlab
> >> interface.
> >> So I created a block matrix, and now i cannot create the corresponding
> >> bricks.
> >> I'm using the command "gf_asm", hoping it's the correct one.
> >>
> >> the block is the following:
> >> Mtt=1/k u_t v_t. (v_t is the test function corresponding to u_t)
> >> in full despair, i also removed the 1/k therm, and still nothing worked.
> >> this is what i wrote:
> >>
> >> gf_asm('generic', mim3, 2, 'Grad_Test2_ut.Grad_Test_ut', -1, 'ut',1, U);
> >>
> >> And the error message is:
> >>
> >> "Error using gf_matlab
> >> Error in getfem_generic_assembly.cc, line 9519 void
> >> getfem::ga_node_analysis(const string&, getfem::ga_tree&, const
> >> getfem::ga_workspace&, getfem::pga_tree_node, bgeot::size_type,
> >> bgeot::size_type, bool, bool, int):
> >> Error in assembly string
> >>
> >> Error in gf_asm (line 338)
> >> [varargout{1:nargout}]=gf_matlab('asm', varargin{:});
> >>
> >> Error in Comeon (line 56)
> >> Mtt=gf_asm('generic', mim3, 2, 'Grad_Test2_ut.Grad_Test_ut', -1, 'ut',1,
> >> U); % M_tt, proviamo"
> >>
> >> with
> >> mim3= gf_mesh_im( mf3, gf_integ('IM_NC(3,1)') )
> >> mf3=gf_mesh_fem(OmegaM,3)
> >> OmegaM=gf_mesh('regular simplices',vectX, vectY, vectZ, 'degree', 1)
> >>
> >> at his point, i defined U as ones(2,2,2,3), having the vectX(/Y/Z) only
> 2
> >> points each one, for now, and because u_t is a 3D vector.
> >> I think that's not correct, but i have no idea of the correct dimension
> of
> >> U, and i tried everything that came to my mind ... and still nothing
> >> worked.
> >>
> >> I found a lot of material dealing with this on c++ interface, but
> >> unfortunately I cannot male it work in Matlab.
> >>
> >> thank you very much,
> >> Nicolo Cambiaso.
> >>
> >
>
