Work at SourceForge, help us to make it a better place! We have an immediate need for a Support Technician in our San Francisco or Denver office.

## Re: [Libmesh-users] Optimizing matrix storage for a lumped mass matrix.

 Re: [Libmesh-users] Optimizing matrix storage for a lumped mass matrix. From: Kirk, Benjamin (JSC-EG311) - 2014-01-17 16:28:51 ```On Jan 17, 2014, at 10:10 AM, subramanya sadasiva wrote: > Thanks.I wanted to try and keep it simple to alternate between lumped and full mass matrices. I had a doubt about using them with adaptivity. will solving , followed by applying constraints exactly give me the correct results? > Subramanya I'll see if I can find a reference, but I recall reading that since the mass matrix is so nice it can be iteratively solved via Jacobi or Gauss-Siedel iteration as Mu = f (D + OD)u = f Du = f - ODu Du^{l+1} = f - ODu^l where D is diagonal and OD is the rest… Now, I'd love a patch that basically implements that if you are feeling ambitious… For the lumped case of course OD=0 and everything converges in one iteration. -Ben ```

 [Libmesh-users] Optimizing matrix storage for a lumped mass matrix. From: subramanya sadasiva - 2014-01-17 14:54:23 ```I am trying to compute the L2 projection of some gauss point quantities to the nodes, and I'd like to use a trapezoid rule to integrate the mass matrix. Is there anyway to optimize the allocation of the matrix, so that it is a strictly diagonal matrix? Thanks, Subramanya ```
 Re: [Libmesh-users] Optimizing matrix storage for a lumped mass matrix. From: Kirk, Benjamin (JSC-EG311) - 2014-01-17 15:40:09 ```Not through the dof map, but you can manually assemble into a 'diagonal' matrix you create - e.g. a vector. -Ben > On Jan 17, 2014, at 8:54 AM, "subramanya sadasiva" wrote: > > I am trying to compute the L2 projection of some gauss point quantities to the nodes, and I'd like to use a trapezoid rule to integrate the mass matrix. Is there anyway to optimize the allocation of the matrix, so that it is a strictly diagonal matrix? Thanks, Subramanya > ------------------------------------------------------------------------------ > CenturyLink Cloud: The Leader in Enterprise Cloud Services. > Learn Why More Businesses Are Choosing CenturyLink Cloud For > Critical Workloads, Development Environments & Everything In Between. > Get a Quote or Start a Free Trial Today. > http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk > _______________________________________________ > Libmesh-users mailing list > Libmesh-users@... > https://lists.sourceforge.net/lists/listinfo/libmesh-users ```
 Re: [Libmesh-users] Optimizing matrix storage for a lumped mass matrix. From: Kirk, Benjamin (JSC-EG311) - 2014-01-17 16:28:51 ```On Jan 17, 2014, at 10:10 AM, subramanya sadasiva wrote: > Thanks.I wanted to try and keep it simple to alternate between lumped and full mass matrices. I had a doubt about using them with adaptivity. will solving , followed by applying constraints exactly give me the correct results? > Subramanya I'll see if I can find a reference, but I recall reading that since the mass matrix is so nice it can be iteratively solved via Jacobi or Gauss-Siedel iteration as Mu = f (D + OD)u = f Du = f - ODu Du^{l+1} = f - ODu^l where D is diagonal and OD is the rest… Now, I'd love a patch that basically implements that if you are feeling ambitious… For the lumped case of course OD=0 and everything converges in one iteration. -Ben ```