## Re: [Libmesh-users] Multiple equatiosn with different orders of interpolation

 Re: [Libmesh-users] Multiple equatiosn with different orders of interpolation From: David Knezevic - 2013-02-15 16:54:37 ```On 02/15/2013 11:48 AM, Kirk, Benjamin (JSC-EG311) wrote: > > On Feb 15, 2013, at 10:41 AM, "David Knezevic" wrote: > >> On 02/15/2013 11:39 AM, Subramanya Gautam Sadasiva wrote: >>> Hi, >>> I am trying to solve a problem with multiple system objects as part of the same equation syztems object. One of them ( A navier stokes solver ) needs quad9 elements and the other one a cahn hilliard solver, I want to solve using QUAD4 elements. Is this possible on a single mesh?. The code does not work when I try to do this. >> Sure, that should be no problem. >> >> David > The trick is the difference between geometric elements and finite elements. Your mesh will need Quad9 geometric elements, but you can add bilinear, biquadratic, hierarchic, whatever finite element approximations. Yes, sorry my post didn't actually point out the answer! ```

 [Libmesh-users] Multiple equatiosn with different orders of interpolation From: Subramanya Gautam Sadasiva - 2013-02-15 16:39:49 ```Hi, I am trying to solve a problem with multiple system objects as part of the same equation syztems object. One of them ( A navier stokes solver ) needs quad9 elements and the other one a cahn hilliard solver, I want to solve using QUAD4 elements. Is this possible on a single mesh?. The code does not work when I try to do this. Thanks, Subramanya Sadasiva "But memory eventually fades. Turbulences damp out, internal strains yield to plastic flow, concentration inhomogeneities diffuse to uniformity. Systems tend to subside to very simple states,independent of their specific history." Herbert Callen . Thermodynamics and an Introduction to Thermostatics. ```
 Re: [Libmesh-users] Multiple equatiosn with different orders of interpolation From: David Knezevic - 2013-02-15 16:41:31 ```On 02/15/2013 11:39 AM, Subramanya Gautam Sadasiva wrote: > Hi, > I am trying to solve a problem with multiple system objects as part of the same equation syztems object. One of them ( A navier stokes solver ) needs quad9 elements and the other one a cahn hilliard solver, I want to solve using QUAD4 elements. Is this possible on a single mesh?. The code does not work when I try to do this. Sure, that should be no problem. David ```
 Re: [Libmesh-users] Multiple equatiosn with different orders of interpolation From: Kirk, Benjamin (JSC-EG311) - 2013-02-15 16:49:02 ``` On Feb 15, 2013, at 10:41 AM, "David Knezevic" wrote: > > On 02/15/2013 11:39 AM, Subramanya Gautam Sadasiva wrote: >> Hi, >> I am trying to solve a problem with multiple system objects as part of the same equation syztems object. One of them ( A navier stokes solver ) needs quad9 elements and the other one a cahn hilliard solver, I want to solve using QUAD4 elements. Is this possible on a single mesh?. The code does not work when I try to do this. > > Sure, that should be no problem. > > David The trick is the difference between geometric elements and finite elements. Your mesh will need Quad9 geometric elements, but you can add bilinear, biquadratic, hierarchic, whatever finite element approximations. -Ben ```
 Re: [Libmesh-users] Multiple equatiosn with different orders of interpolation From: David Knezevic - 2013-02-15 16:54:37 ```On 02/15/2013 11:48 AM, Kirk, Benjamin (JSC-EG311) wrote: > > On Feb 15, 2013, at 10:41 AM, "David Knezevic" wrote: > >> On 02/15/2013 11:39 AM, Subramanya Gautam Sadasiva wrote: >>> Hi, >>> I am trying to solve a problem with multiple system objects as part of the same equation syztems object. One of them ( A navier stokes solver ) needs quad9 elements and the other one a cahn hilliard solver, I want to solve using QUAD4 elements. Is this possible on a single mesh?. The code does not work when I try to do this. >> Sure, that should be no problem. >> >> David > The trick is the difference between geometric elements and finite elements. Your mesh will need Quad9 geometric elements, but you can add bilinear, biquadratic, hierarchic, whatever finite element approximations. Yes, sorry my post didn't actually point out the answer! ```
 Re: [Libmesh-users] Multiple equatiosn with different orders of interpolation From: Subramanya Gautam Sadasiva - 2013-02-15 17:01:57 ```Hi , I knew that that was the way things were implemented. The main problem seems to be that I am using the snes-vi solver and to apply the bounds , I do MeshBase::const_node_iterator node_it = _solbub._mesh->local_nodes_begin(); const MeshBase::const_node_iterator node_it_end = _solbub._mesh->local_nodes_end(); for (;node_it!=node_it_end;node_it++) { Node* node = *node_it; const unsigned int dofnum_phi = node->dof_number(sys_num,phi_var,0); const unsigned int dofnum_mu = node->dof_number(sys_num,mu_var,0); } This causes a segmentation fault. I guess the node iterator goes over ALL the nodes that it sees. How do I check if dof_number is returning something meaningful? Sorry my first email was a bit hasty. I sent it before I localized the error. Thanks, Subramanya Sadasiva "But memory eventually fades. Turbulences damp out, internal strains yield to plastic flow, concentration inhomogeneities diffuse to uniformity. Systems tend to subside to very simple states,independent of their specific history." Herbert Callen . Thermodynamics and an Introduction to Thermostatics. ----- Original Message ----- From: "David Knezevic" Cc: libmesh-users@... Sent: Friday, February 15, 2013 11:54:28 AM Subject: Re: [Libmesh-users] Multiple equatiosn with different orders of interpolation On 02/15/2013 11:48 AM, Kirk, Benjamin (JSC-EG311) wrote: > > On Feb 15, 2013, at 10:41 AM, "David Knezevic" wrote: > >> On 02/15/2013 11:39 AM, Subramanya Gautam Sadasiva wrote: >>> Hi, >>> I am trying to solve a problem with multiple system objects as part of the same equation syztems object. One of them ( A navier stokes solver ) needs quad9 elements and the other one a cahn hilliard solver, I want to solve using QUAD4 elements. Is this possible on a single mesh?. The code does not work when I try to do this. >> Sure, that should be no problem. >> >> David > The trick is the difference between geometric elements and finite elements. Your mesh will need Quad9 geometric elements, but you can add bilinear, biquadratic, hierarchic, whatever finite element approximations. Yes, sorry my post didn't actually point out the answer! ------------------------------------------------------------------------------ Free Next-Gen Firewall Hardware Offer Buy your Sophos next-gen firewall before the end March 2013 and get the hardware for free! Learn more. http://p.sf.net/sfu/sophos-d2d-feb _______________________________________________ Libmesh-users mailing list Libmesh-users@... https://lists.sourceforge.net/lists/listinfo/libmesh-users ```
 Re: [Libmesh-users] Multiple equatiosn with different orders of interpolation From: Kirk, Benjamin (JSC-EG311) - 2013-02-15 17:42:05 ```I believe you need to first check DofObject::n_comp() for the node to make sure it's not 0. On Feb 15, 2013, at 11:02 AM, "Subramanya Gautam Sadasiva" wrote: > Hi , > I knew that that was the way things were implemented. The main problem seems to be that I am using the snes-vi solver and to apply the bounds , I do > > > > MeshBase::const_node_iterator node_it = _solbub._mesh->local_nodes_begin(); > const MeshBase::const_node_iterator node_it_end = _solbub._mesh->local_nodes_end(); > for (;node_it!=node_it_end;node_it++) > > { > > Node* node = *node_it; > > const unsigned int dofnum_phi = node->dof_number(sys_num,phi_var,0); > const unsigned int dofnum_mu = node->dof_number(sys_num,mu_var,0); > > > } > > This causes a segmentation fault. I guess the node iterator goes over ALL the nodes that it sees. How do I check if dof_number is returning something meaningful? Sorry my first email was a bit hasty. I sent it before I localized the error. > > Thanks, > > Subramanya Sadasiva > > "But memory eventually fades. Turbulences damp out, internal strains yield to plastic flow, concentration inhomogeneities diffuse to uniformity. Systems tend to subside to very simple states,independent of their specific history." > Herbert Callen . Thermodynamics and an Introduction to Thermostatics. > > ----- Original Message ----- > From: "David Knezevic" > Cc: libmesh-users@... > Sent: Friday, February 15, 2013 11:54:28 AM > Subject: Re: [Libmesh-users] Multiple equatiosn with different orders of interpolation > > > On 02/15/2013 11:48 AM, Kirk, Benjamin (JSC-EG311) wrote: >> >> On Feb 15, 2013, at 10:41 AM, "David Knezevic" wrote: >> >>> On 02/15/2013 11:39 AM, Subramanya Gautam Sadasiva wrote: >>>> Hi, >>>> I am trying to solve a problem with multiple system objects as part of the same equation syztems object. One of them ( A navier stokes solver ) needs quad9 elements and the other one a cahn hilliard solver, I want to solve using QUAD4 elements. Is this possible on a single mesh?. The code does not work when I try to do this. >>> Sure, that should be no problem. >>> >>> David >> The trick is the difference between geometric elements and finite elements. Your mesh will need Quad9 geometric elements, but you can add bilinear, biquadratic, hierarchic, whatever finite element approximations. > > Yes, sorry my post didn't actually point out the answer! > > > > > ------------------------------------------------------------------------------ > Free Next-Gen Firewall Hardware Offer > Buy your Sophos next-gen firewall before the end March 2013 > and get the hardware for free! Learn more. > http://p.sf.net/sfu/sophos-d2d-feb > _______________________________________________ > Libmesh-users mailing list > Libmesh-users@... > https://lists.sourceforge.net/lists/listinfo/libmesh-users > > ------------------------------------------------------------------------------ > Free Next-Gen Firewall Hardware Offer > Buy your Sophos next-gen firewall before the end March 2013 > and get the hardware for free! Learn more. > http://p.sf.net/sfu/sophos-d2d-feb > _______________________________________________ > Libmesh-users mailing list > Libmesh-users@... > https://lists.sourceforge.net/lists/listinfo/libmesh-users ```
 Re: [Libmesh-users] Multiple equatiosn with different orders of interpolation From: John Peterson - 2013-02-15 17:47:49 ```On Fri, Feb 15, 2013 at 10:41 AM, Kirk, Benjamin (JSC-EG311) < benjamin.kirk-1@...> wrote: > I believe you need to first check DofObject::n_comp() for the node to make > sure it's not 0. > Running in devel/debug mode will help you catch things like this too. We have libmesh_assert_less (comp, this->n_comp(s,var)); in DofObject::dof_number() -- John ```