Math 2.0

Numerical analysis Iterative methods for linear systems Theorem

Balancing domain decomposition

Builds on Schur complement methodSchur complementPreconditionerCondition numberFinite element methodElliptic partial differential equation

Statement

Let Ω⊂ℝd\Omega \subset \R^d, d∈{2,3}d \in \set{2, 3}, be a polygonal or polyhedral domain. Divide it into NN non-overlapping subdomains Ωi\Omega_i of diameter about HH, triangulated by a quasi-uniform mesh of size hh. Let ρ>0\rho > 0 be constant on each Ωi\Omega_i. Consider the finite element discretization of −∇⋅(ρ∇u)=f-\nabla \cdot (\rho \nabla u) = f by continuous piecewise linear functions, with homogeneous Dirichlet conditions on part of ∂Ω\partial \Omega. Let S=∑iRiTSiRiS = \sum_i R_i^T S_i R_i be the interface Schur complement, where RiR_i restricts interface values to ∂Ωi\partial \Omega_i and SiS_i is the local Schur complement of Ωi\Omega_i. Let DiD_i be diagonal weight matrices with ∑iRiTDiRi=I\sum_i R_i^T D_i R_i = I. Let ZiZ_i have columns spanning the null space of SiS_i (the constants for a floating subdomain), and let R0TR_0^T have the columns RiTDiZiR_i^T D_i Z_i. With Q0=R0T(R0SR0T)−1R0Q_0 = R_0^T (R_0 S R_0^T)^{-1} R_0, the balancing domain decomposition preconditioner is M−1=Q0+(I−Q0S)(∑i=1NRiTDiSi†DiRi)(I−SQ0),M^{-1} = Q_0 + (I - Q_0 S) \Bigl( \sum_{i=1}^{N} R_i^T D_i S_i^{\dagger} D_i R_i \Bigr) (I - S Q_0), where Si†S_i^\dagger is a generalized inverse of SiS_i. Choose the weights DiD_i from the values of ρ\rho as in the standard theory. Then the condition number of M−1SM^{-1} S satisfies κ(M−1S)≤C(1+log⁡Hh)2,\kappa(M^{-1} S) \le C \Bigl( 1 + \log \frac{H}{h} \Bigr)^2, with CC independent of hh, HH, NN and the values of ρ\rho.

Remarks

The local operator ∑iRiTDiSi†DiRi\sum_i R_i^T D_i S_i^\dagger D_i R_i is the Neumann–Neumann preconditioner: each Si†S_i^\dagger solves a Neumann problem on Ωi\Omega_i. A Neumann problem on a subdomain that does not touch the Dirichlet boundary is solvable only if its right-hand side is orthogonal to the constants. The coarse correction Q0Q_0 "balances" the residual so that these compatibility conditions hold, and at the same time it passes global information between all subdomains. The proof bounds the energy of the weighted average of local functions by trace and discrete harmonic extension estimates, which bring in the factor (1+log⁡(H/h))2(1 + \log(H/h))^2.

The bound depends on H/hH/h only, so refining the mesh and adding subdomains in proportion keeps the number of conjugate gradient iterations nearly constant. The method was introduced by Jan Mandel in 1993.

Going deeper

For a student

The plain Neumann–Neumann method has no coarse space and scales poorly with the number of subdomains. BDD adds a coarse space, and the later BDDC method uses a different coarse space built from primal constraints. Its condition number bound is of the same polylogarithmic type as that of FETI methods, and in fact the two families are closely related. It is one of the domain decomposition methods that are robust with respect to large jumps of the coefficient across subdomains.

Reading path

To understand this page from the ground up, read these 9 pages in order. The foundations come first.

Show the path as a graph
Each page uses the pages above it. The colour of the bar on a box shows how strong the evidence is, and the box also gives it as text. The list has the same pages.
  • no evidence level
  1. Block matrix concept
  2. Condition number concept
  3. Elliptic partial differential equation concept
  4. Finite element method concept
  5. Positive definite matrix concept
  6. Preconditioner concept
  7. Schur complement concept
  8. Schur complement method statement
  9. Balancing domain decomposition statement

Further reading

  1. Web https://en.wikipedia.org/wiki/Balancing_domain_decomposition_method
About this page
TagX8MCVW56QP
URIm2://X8MCVW56QP
Kindstatement, theorem
WritingW0 stub
MSC65N55, 65F08, 65F10
StatusThis copy of the site has no build data. Run m2 status X8MCVW56QP.
AboutDomain decomposition methods, Preconditioner
UsesSchur complement method, Schur complement, Preconditioner, Condition number, Finite element method, Elliptic partial differential equation
Same as

Markdown · JSON · Edit on GitHub

What this page needs for W1

  • an inventory row: the row of the inventory that the node accounts for (0 of 1)
  • a source with a locator (0 of 1)

The writing levels say how complete the text of a page is. They do not measure its truth.

What links here

No page links to this page yet.