solver Namespace Reference

Functions
	surrogate (G, k)
	gs_rate (A, b, n=200)
	run (ND, k, dtn)

Variables
str	SURR = "centroid"
	o = run(ND, k, dtn)
	gs = o["GS"]
	flush

Detailed Description

2-D prototype, round 5: can peclet's solvers handle the probe-flux operator?

A = probe-flux cut-cell FV (jc.py, probe 0.7h) on the disc, Dirichlet (worst case) and Robin.
Systems: steady (A u = b) and one backward-Euler step (kappa/dt + A) u = b at diffusion number
dt*D/h^2 = 1, 10.
Tested:
  GS     : lexicographic Gauss-Seidel directly on A (stands in for RB-GS): converges?
  DC     : defect correction  u += S^-1 (b - A u)  with S = symmetric compact surrogate
           (aperture FV + diagonal wall term A_w/(d + 1/k), d = centroid distance; SPD) -> rho(I - S^-1 A)
  BiCGStab / GMRES(20) preconditioned by S (exact S^-1 here = an ideal MG V-cycle): iterations to 1e-10
Also: does S itself suit MG (it is the existing operator family + a diagonal) - assumed, not tested here.

Function Documentation

surrogate()

solver.surrogate	(		G,
			k
	)

Definition at line 21 of file solver.py.

Referenced by run().

gs_rate()

solver.gs_rate	(		A,
			b,
			n = 200
	)

Definition at line 30 of file solver.py.

Referenced by run().

run()

solver.run	(		ND,
			k,
			dtn
	)

Definition at line 42 of file solver.py.

References gs_rate(), and surrogate().

Variable Documentation

SURR

int solver.SURR = "centroid"

Definition at line 18 of file solver.py.

o

solver.o = run(ND, k, dtn)

Definition at line 80 of file solver.py.

gs

solver.gs = o["GS"]

Definition at line 81 of file solver.py.

flush

solver.flush

Definition at line 84 of file solver.py.