GMS:LMG Package

From XMS Wiki
Revision as of 20:43, 1 July 2016 by Dharker (talk | contribs) (Included information about Parallel PEST)
Jump to navigationJump to search
MODFLOW
Pre-processing
MODFLOW Commands
Building a MODFLOW Model
Map to MODFLOW
Calibration
Packages Supported in GMS
Saving a MODFLOW Simulation
Importing MODFLOW Files
Unsupported MODFLOW Features
Run MODFLOW
Post-processing
MODFLOW Display Options
MODFLOW Post-Processing Viewing Options
Reading a MODFLOW Simulation
Tutorials
Packages
Flow: BCF6, HUF, LPF, UPW
Solvers:

DE4, GMG, NWT, PCG,

PCGN, LMG, SIP, SOR,

SMS
Other:

BAS6, BFH, CHD1, CLN,

DRN1, DRT1, EVT1, ETS1,

GAGE, GHB1, GNC, HFB1,

HUF, LAK3, MNW1, MNW2,

OUT1, RCH1, RIV1, SFR2,

STR1, SUB1, SWI2, WEL1,

UZF1

The MODFLOW LINK-AMG (LMG) Package dialog.

The MODFLOW Link-AMG (Algebraic Multi-Grid), or LMG (Link Multi-Grid), package solver is an interface linking the SAMG (Systems Algebraic Multi-Grid) library to MODFLOW. The LMG3 package offers some improvements over the LMG1 package for certain types of problems, and is backward compatible with LMG1 input files. For non-linear problems and simulations which use several time steps, subsequent matrices to be solved for often change only slightly in their algebraic properties. In these cases the LMG3 solver can accelerate the overall simulation run enormously. Using SAMG the user can run a specific GMS MODFLOW executable (mf2k5_..._parallel.exe) with parallel processing on a single machine (though not across multiple machines on a network).

The LMG1 package is only supported in GMS versions prior to 8.0.

MXITER. Maximum outer iterations

This value is the maximum iterations, or calls to the solver. For linear problems, this value can be 2. For non linear problems, this values should be larger, but generally less than 50.

MXCYC. Maximum inner iterations

This value is the maximum number of iterations per call to the solver. The default value is 50, and this is sufficient for most problems.

RCLOSE. Inner iteration closure criterion

Same as RCLOSE for the GMG solver package. The PCG algorithm computes the l2norm of the residual and compares it against RCLOSE. In LMG1 this was called Budget Closure Criterion. When the scaled norm of the matrix equations is less than this value then the solver iteration stops.

HCLOSE. Outer iteration closure criterion

Head change convergence criterion for nonlinear problems. After each linear solve (inner iteration), the maximum norm of the head change is compared against HCLOSE.

CONTROL. LMG version to use

This variable allows the user to select either LMG1 or LMG2 version of the LMG solver.

Damping method

Three different damping methods are available: constant, Adaptive damping using the Cooley method, or Adaptive RRR damping using the relative reduced residual. Refer to the official package documentation for more explanation on the different damping methods.

DAMP. Acceleration Parameter

This value is used if the acceleration method is set to constant. Usually, this value is 1.0, but for some problems lower values may help achieve convergence.

DUP. Max Acceleration

This is the maximum damping value when using the adaptive damping with relative reduced residual.

DLOW. Min Acceleration

This is the minimum damping value when using the adaptive damping with relative reduced residual.

Conjugate Gradient

Controls whether the conjugate gradient method is used to accelerate the solver. By default this option is on.

Stor1, Stor2, Stor3

These values are ignored in LMG2. In LMG1, these variables are used to control the amount of memory storage used by the solver. The defaults rarely need to be changed.

IOUTAMG. Output control

This controls the information printed to the SAMG output file while the solver is running.

Reset

This button will change all options to the default state.