math — Commands

airy
Length Scale Default is microns which can be changed with the option command
Name            Type     Default               Information
x               Float    0.000000e+00         airy (Ai) function input

densesolve
Length Scale Default is microns which can be changed with the option command
Name            Type     Default               Information
A               String                        Row-major matrix as flat Tcl list (N*N doubles)
b               String                        RHS vector as Tcl list (N doubles)

math
Length Scale Default is microns which can be changed with the option command
Name            Type     Default               Information
analysis        Switch                        set parameters for this analysis type
	diffuse         Boolean                      
	flow            Boolean                      
	device          Boolean                      
	ac.device       Boolean                      
	level           Boolean                      
	smooth          Boolean                      
	stress          Boolean                      
model.flow      Switch                        flow models
	viscoelastic    Boolean                      
	viscous         Boolean                      
model.err       Switch                        error method
	thickness       Boolean                      
	velocity        Boolean                      
model.bc        Switch                        boundary conditions
	reflecting      Boolean                      
	periodic        Boolean                      
	curvature       Boolean                      
splitve         Boolean  0                    split elements
timescale       Boolean  0                    time scale
pressure        Boolean  0                    pressure calc
nonlingauss     Boolean  0                    gauss iters for nonlin
nonlinnewton    Boolean  0                    newton iters for nonlin
nonlin.error    Float    0.000000e+00         nonlinear error tolerance
nonlin.maxiter  Float    0.000000e+00         nonlinear max iterations
nonlin.wplim    Float    0.000000e+00        
nonlin.wpdec    Float    0.000000e+00        
nonlin.wpinc    Float    0.000000e+00        
ox.fin.vol      Boolean  0                    finite volumes
ox.fin.ele      Boolean  0                    finite elements
split.mid       Boolean                      
nl.ks           Boolean                      
nl.diff         Boolean                      
nl.visc         Boolean                      
taulim          Float    0.000000e+00        
vel.acc         Float    0.000000e+00        
max.oxstep      Float    0.000000e+00        
dimension       Int      1                    dimension to apply params
matpar          Switch                        matrix format
	triplet         Boolean  0                    triplet ordering
	row             Boolean  0                    row ordering
	col             Boolean  0                    column ordering
	eigencol        Boolean  0                    eigen-based column ordering (ColMajor SpMat64). Supports all esolve* methods. Eigen's sparse_time_dense_product_impl is SERIAL for ColMajor, so the inner matvec in esolve.bicgstab / esolve.gmres does not parallelise under OpenMP
	eigenrow        Boolean  0                    eigen-based row ordering (RowMajor SpMat64R). Eigen's sparse_time_dense_product_impl HAS an OpenMP region for RowMajor, so eigenrow + esolve.bicgstab / esolve.gmres is the only combination whose matvec actually multi-threads in flooxs.openmp. ILUT preconditioner triangular solves remain serial regardless
ordpar          Switch                        ordering options
	natural         Boolean                      
	qmd             Boolean                      
	rcm             Boolean                      
	one1wd          Boolean                      
	nd              Boolean                      
methpar         Switch                        preconditioner options
	lu              Boolean  0                    Real LU factorization (UMFPACK)
	umf             Boolean  0                    Real LU factorization (UMFPACK)
	complex         Boolean  0                    Complex LU factorization (UMFPACK zi)
	ilu             Boolean                      
	diag            Boolean                      
	jacobi          Boolean                      
	block           Boolean                      
	epre            Boolean                      
itpar           Switch                        iterative methods
	none            Boolean                      
	bcgs            Boolean                      
	blasbcgs        Boolean                      
	gmres           Boolean                      
	esolve          Boolean  0                    Eigen UmfPackLU solver (direct LU through Eigen's UMFPACK wrapper). Serial in flooxs.openmp -- UMFPACK has no #pragma omp on its LU path
	esolve.sparselu Boolean  0                    Eigen SparseLU solver (direct LU, Eigen's own implementation). Serial -- SparseLU has no OpenMP regions
	esolve.bicgstab Boolean  0                    Eigen BiCGSTAB iterative solver with IncompleteLUT preconditioner. Inner matvec is OpenMP-parallel ONLY when combined with eigenrow (RowMajor matrix); eigencol stays serial. ILUT triangular solves are serial in all cases
	esolve.gmres    Boolean  0                    Eigen GMRES iterative solver (currently aliases to BiCGSTAB). Same OpenMP behaviour as esolve.bicgstab: parallel matvec only with eigenrow
tolerance       Float    0.000000e+00         solution tolerance
max.iter        Int      0                    maximum iteration count
scale           Boolean  0                    scale the matrix?
skipmap         Boolean  0                    skip building map for matrix 
cudss           Boolean  0                    use cudss for solve, only works of eigen crm 
usennzperrow    Boolean  0                    use nnz per row only works for eigen now 
difference      Boolean  0                    difference method for time step error
milne           Boolean  0                    Milne's device for time step error
expmilne        Boolean  0                    Explicit Milne's device for time step error
arpack          Boolean  0                    use ARPACK iterative eigensolver
iterLimit       Int      20                   Newton iteration limit
updateLimit     Float    1.000000e-03         Newton update convergence limit
rhsLimit        Float    1.000000e-08         Newton RHS convergence limit
rhsMax          Float    1.000000e-12         Newton minimum RHS for convergence
residCheck      Boolean  1                    check residual during Newton iteration