Contents
RATE OF CONVERGENCE OF BILINEAR FINITE ELEMENT METHOD
This example is to show the rate of convergence of bilinear finite element approximation of the Poisson equation on the unit square:
for the following boundary condition:
- Non-empty Dirichlet boundary condition. .
- Pure Neumann boundary condition. .
- Robin boundary condition.
clear variable
Setting
[node,elem] = squarequadmesh([0,1,0,1],1/2^3); mesh = struct('node',node,'elem',elem); option.L0 = 2; option.maxIt = 4; option.printlevel = 1; option.plotflag = 0; option.elemType = 'Q1';
Non-empty Dirichlet boundary condition.
pde = sincosdata; mesh.bdFlag = setboundary(node,elem,'Dirichlet','~(x==0)','Neumann','x==0'); femPoisson(mesh,pde,option);
Multigrid V-cycle Preconditioner with Conjugate Gradient Method #dof: 4225, #nnz: 35530, smoothing: (1,1), iter: 7, err = 3.84e-09, time = 0.017 s Multigrid V-cycle Preconditioner with Conjugate Gradient Method #dof: 16641, #nnz: 144778, smoothing: (1,1), iter: 7, err = 6.20e-09, time = 0.063 s Multigrid V-cycle Preconditioner with Conjugate Gradient Method #dof: 66049, #nnz: 584458, smoothing: (1,1), iter: 7, err = 9.62e-09, time = 0.21 s #Dof h ||u-u_h|| ||Du-Du_h|| ||DuI-Du_h|| ||uI-u_h||_{max} 1089 3.12e-02 7.57200e-04 4.44017e+00 8.53343e-04 2.71970e-04 4225 1.56e-02 1.89319e-04 4.44220e+00 2.13806e-04 6.79552e-05 16641 7.81e-03 4.73308e-05 4.44271e+00 5.34807e-05 1.69906e-05 66049 3.91e-03 1.18328e-05 4.44284e+00 1.33715e-05 4.24808e-06 #Dof Assemble Solve Error Mesh 1089 1.00e-02 1.92e-03 2.00e-02 1.00e-02 4225 7.00e-02 1.67e-02 1.00e-01 1.00e-02 16641 2.20e-01 6.32e-02 2.90e-01 5.00e-02 66049 8.10e-01 2.14e-01 9.20e-01 2.00e-01
Pure Neumann boundary condition.
pde = sincosNeumanndata;
mesh.bdFlag = setboundary(node,elem,'Neumann');
femPoisson(mesh,pde,option);
Multigrid V-cycle Preconditioner with Conjugate Gradient Method #dof: 4225, #nnz: 37242, smoothing: (1,1), iter: 9, err = 4.35e-10, time = 0.017 s Multigrid V-cycle Preconditioner with Conjugate Gradient Method #dof: 16641, #nnz: 148218, smoothing: (1,1), iter: 9, err = 1.68e-09, time = 0.051 s Multigrid V-cycle Preconditioner with Conjugate Gradient Method #dof: 66049, #nnz: 591354, smoothing: (1,1), iter: 9, err = 9.54e-09, time = 0.25 s #Dof h ||u-u_h|| ||Du-Du_h|| ||DuI-Du_h|| ||uI-u_h||_{max} 1089 3.12e-02 1.89964e-03 8.87506e+00 1.42236e-02 3.21831e-03 4225 1.56e-02 5.39236e-04 8.88372e+00 4.90169e-02 4.88870e-02 16641 7.81e-03 1.34875e-04 8.88525e+00 2.45349e-02 2.45187e-02 66049 3.91e-03 3.37228e-05 8.88564e+00 1.22707e-02 1.22687e-02 #Dof Assemble Solve Error Mesh 1089 1.00e-02 2.49e-03 1.00e-02 0.00e+00 4225 6.00e-02 1.68e-02 9.00e-02 1.00e-02 16641 2.30e-01 5.15e-02 2.30e-01 5.00e-02 66049 8.20e-01 2.46e-01 1.00e+00 2.00e-01
Pure Robin boundary condition.
option.plotflag = 0;
pde = sincosRobindata;
mesh.bdFlag = setboundary(node,elem,'Robin');
femPoisson(mesh,pde,option);
Multigrid V-cycle Preconditioner with Conjugate Gradient Method #dof: 4225, #nnz: 37249, smoothing: (1,1), iter: 7, err = 1.38e-09, time = 0.018 s Multigrid V-cycle Preconditioner with Conjugate Gradient Method #dof: 16641, #nnz: 148225, smoothing: (1,1), iter: 7, err = 2.31e-09, time = 0.056 s Multigrid V-cycle Preconditioner with Conjugate Gradient Method #dof: 66049, #nnz: 591361, smoothing: (1,1), iter: 7, err = 3.45e-09, time = 0.24 s #Dof h ||u-u_h|| ||Du-Du_h|| ||DuI-Du_h|| ||uI-u_h||_{max} 1089 3.12e-02 1.89975e-03 8.87506e+00 1.45807e-02 3.21823e-03 4225 1.56e-02 4.75115e-04 8.88309e+00 3.65461e-03 8.03533e-04 16641 7.81e-03 1.18790e-04 8.88510e+00 9.14242e-04 2.00819e-04 66049 3.91e-03 2.96981e-05 8.88560e+00 2.28597e-04 5.02008e-05 #Dof Assemble Solve Error Mesh 1089 1.00e-02 2.15e-03 1.00e-02 0.00e+00 4225 7.00e-02 1.75e-02 1.10e-01 1.00e-02 16641 2.20e-01 5.60e-02 2.70e-01 5.00e-02 66049 8.80e-01 2.38e-01 1.04e+00 2.50e-01
Conclusion
To do: 1. Fix the getH1errorQ1. The H1 error is computed wrong. 2. For pure Neuman boundary condition, the rate is not quite right.