Introduction

In this cheatsheet, a rudimentary overview of a couple executables that can be compiled from the test folder will be provided.

The instructions included in this document targets LIS version 2.1.10. The sources for LIS can be downloaded here.

All the executables used in this document are obtained from the compilation of the sources in the test folder in the sources.

The following table lists all the executable and their respective sources:

Executable	Source
`test1`	`test1.c`
`etest1`	`etest1.c`

To compile the sources, the following command was used:

mpicc -DUSE_MPI -I</path/to/LIS/includes> -L</path/to/LIS/library> -llis </path/to/source.c> -o </path/to/output/executable>

The compiled executables can be both run standalone or with mpirun.

Linear system solution

The test1 executable is used to solve linear systems. It includes multiple iterative solvers that can be tuned with various parameters.

Syntax: ./test1 INPUT_FILE RHS_SETTINGS SOLUTION_FILE HISTORY_FILE [OPTIONS] where

INPUT_FILE is the A matrix as in Ax = b stored in matrix market format (mtx);
RHS_SETTINGS is the b of the system: it can be a number (1 to set b = (1, 1, \dots, 1), 2 to set b = A \times (1, 1, \dots, 1)) or a file containing the vector in mtx format;
SOLUTION_FILE is the file to which the solution is written;
HISTORY_FILE is the file to which all the intermediate solution approximation are written.

Multiple optional options can be passed to the executable:

Option	Description
`-i jacobi\|gs\|cg\|bicgstab\|gmres\|bicg`	Selects the method used to solve the system
`-tol <tolerance>`	Sets the tolerance for the iterative method
`-maxiter <number>`	Sets the maximum number of iterations for the iterative method
`-restart <number>`	Restart parameter for GMRES
`-p jacobi\|sainv\|ilu\|ssor\|ilut\|saamg`	Selects a preconditioner
`-ilu_fill <number>`	Treshold for ILU
`-ssor_omega <number>`
`-adds true`	Enables Additive Schwarz
`-adds_iter <number>`	Sets additive Schwarz iterations
`-saamg_unsym true\|false`	Tells saamg to use or not the symmetric version

The etest1 executable is used to solve eigenproblems. It can be tuned with various parameters.

Syntax: ./etest1 INPUT_FILE EVEC_FILE HISTORY_FILE [OPTIONS] where

INPUT_FILE is the input matrix in MTX format;
EVEC_FILE is the file in which the eigenvectors are written;
HISTORY_FILE is the file to which all the intermediate solution approximation are written.

Multiple options can be passed to the executable:

Option	Description
`-e pi\|ii\|cr\|rqi\|si\|li\|ai`	Selects the eigensolver
`-i`, `-p`	Options for the linear solver used by the method. Same options as here here
`-emaxiter <number>`	Sets the maximum number of iterations for the eigensolver
`-etol <number>`	Sets the tolerance for the eigensolver
`-shift <number>`	Sets the shift value for the inverse power method
`-ie cg\|ii\|rqi`	Sets the method for the error correction
`-ss <number>`	Sets the number of eigenvalues to approximate

Work in progress