Next:
Contents
Contents
Index
JT
PACK77
(LA-CC-93-5) - A Library / Toolkit of Linear Algebra Routines
(corresponds to code version 26.12)
John A. Turner
XTM
September 18, 1998
(DRAFT)
Contents
1. Introduction
2. Usage
2.1 Roadmap
2.2 Using the C Preprocessor
2.3 Solvers
2.3.1 Common Arguments
2.3.2 Stationary Methods
2.3.3 Krylov Subspace Solvers
2.3.3.1 Calling the Nonstationary Solvers Directly
2.3.3.2 Using the Black-Box Routine
2.3.4 Example using
cpp
variables
2.3.5 Example using hardwired values
2.3.6 Custom preconditioning
via
reverse communication
2.4
JTpack77
as a Toolkit
3. Design Goals and Implementation Details
3.1 Adherence to standards and use of extensions
3.2 Naming of routines, files,
etc
.
3.3 Use of parameter vectors and
cpp
variables
3.4 Automatic determination of environment
3.5 Automatic dependency generation
3.6 Dynamic memory
3.7 Input / Output
4. Storage Parameters
4.1
_JT_ storage_ unknown_
- unknown format
4.2
_JT_ storage_ full_
- full (conventional) format
4.3
_JT_ storage_ EISPACK_
- EISPACK format
4.4
_JT_ storage_ LAPACK_
- LAPACK format
4.5
_JT_ storage_ blocktridiaggeneral_
- general block-tridiagonal format
4.6
_JT_ storage_ blocktridiagsymmetric_
- symmetric block-tridiagonal format
4.7
_JT_ storage_ blocktridiagdiagoff_
- block-tridiagonal with diagonal off-diagonal blocks
4.8
_JT_ storage_ blocktridiagdiagoffsymmetric_
- symmetric, block-tridiagonal with diagonal off-diagonal blocks
4.9
_JT_ storage_ ELL_
- ELLPACK-ITPACK, or row-packed format
4.10
_JT_ storage_ ELLtranspose_
- ELLPACK-ITPACK transpose
4.11
_JT_ storage_ COO_
- coordinate format
4.12
_JT_ storage_ RSS_
- row sequential format
4.13
_JT_ storage_ CSS_
- column sequential format
5. Control Parameters
5.1 Integer Parameters
5.1.1
iparm(
_JT_ safety_
)
- toggles extra input checking
5.1.2
iparm(
_JT_ iter_
)
- number of iterations completed
5.1.3
iparm(
_JT_ out_
)
- output mode
5.1.4
iparm(
_JT_ luout_
)
- logical unit for output
5.1.5
iparm(
_JT_ luerr_
)
- logical unit for error messages
5.1.6
iparm(
_JT_ itmax_
)
- maximum number of iterations to allow
5.1.7
iparm(
_JT_ stop_
)
- stopping test
5.1.8
iparm(
_JT_ norm_
)
- norm
5.1.9
iparm(
_JT_ resid_
)
- determines how often to compute and/or use true residual (
b
-
Ax
) and how often to compute error estimate based on true residual (
errt
)
5.1.10
iparm(
_JT_ method_
)
- determines which method to use if multiple methods are implemented in one routine
5.1.11
iparm(
_JT_ prein_
)
- indicates whether or not a preconditioner will be passed in
5.1.12
iparm(
_JT_ pre_
)
- controls preconditioning
5.1.13
iparm(
_JT_ steps_
)
- number of Jacobi or SSOR steps to take
5.1.14
iparm(
_JT_ nold_
)
- number of previous vectors to use in truncated methods
5.2 Floating-point Parameters
5.2.1
rparm(
_JT_ eps_
)
- convergence criterion
5.2.2
rparm(
_JT_ epspre_
)
- convergence criterion for preconditioning systems
5.2.3
rparm(
_JT_ omega_
)
- relaxation parameter
5.2.4
rparm(
_JT_ anorm_
)
- norm of coefficient matrix
5.2.5
rparm(
_JT_ tiny_
)
- smallest floating-point value to worry about
Bibliography
A. Building and Testing JT
PACK77
A1. Requirements
A2. Building The Code
A2.1 Code Organization
A2.2 Building the Library
A2.3 Building the Test Code
A3. Running the Test Suite
A4. Cleanup
Index
About this document ...
Next:
Contents
Contents
Index
John A. Turner
