| 1 |
This is the ARPACK++ include directory. |
| 2 |
|
| 3 |
1) Files included in this directory: |
| 4 |
|
| 5 |
a) Files that contain ARPACK++ classes definitions: |
| 6 |
|
| 7 |
i) Base classes: |
| 8 |
|
| 9 |
file class |
| 10 |
---------- ---------------- |
| 11 |
arrseig.h ARrcStdEig |
| 12 |
arrgeig.h ARrcGenEig |
| 13 |
arseig.h ARStdEig |
| 14 |
argeig.h ARGenEig |
| 15 |
armat.h ARMatrix |
| 16 |
|
| 17 |
|
| 18 |
ii) Classes that require matrix-vector product functions: |
| 19 |
|
| 20 |
file class |
| 21 |
---------- ---------------- |
| 22 |
arssym.h ARSymStdEig |
| 23 |
arsnsym.h ARNonSymStdEig |
| 24 |
arscomp.h ARCompStdEig |
| 25 |
argsym.h ARSymGenEig |
| 26 |
argnsym.h ARNonSymGenEig |
| 27 |
argcomp.g ARCompGenEig |
| 28 |
|
| 29 |
|
| 30 |
iii) Classes that require matrices in CSC format (SuperLU version): |
| 31 |
|
| 32 |
file class |
| 33 |
---------- ---------------- |
| 34 |
arlssym.h ARluSymStdEig |
| 35 |
arlsnsym.h ARluNonSymStdEig |
| 36 |
arlscomp.h ARluCompStdEig |
| 37 |
arlgsym.h ARluSymGenEig |
| 38 |
arlgnsym.h ARluNonSymGenEig |
| 39 |
arlgcomp.h ARluCompGenEig |
| 40 |
|
| 41 |
|
| 42 |
iv) Classes that require matrices in CSC format (UMFPACK version): |
| 43 |
|
| 44 |
file class |
| 45 |
---------- ---------------- |
| 46 |
arussym.h ARluSymStdEig |
| 47 |
arusnsym.h ARluNonSymStdEig |
| 48 |
aruscomp.h ARluCompStdEig |
| 49 |
arugsym.h ARluSymGenEig |
| 50 |
arugnsym.h ARluNonSymGenEig |
| 51 |
arugcomp.h ARluCompGenEig |
| 52 |
|
| 53 |
|
| 54 |
v) Classes that require matrices in band format: |
| 55 |
|
| 56 |
file class |
| 57 |
---------- ---------------- |
| 58 |
arbssym.h ARluSymStdEig |
| 59 |
arbsnsym.h ARluNonSymStdEig |
| 60 |
arbscomp.h ARluCompStdEig |
| 61 |
arbgsym.h ARluSymGenEig |
| 62 |
arbgnsym.h ARluNonSymGenEig |
| 63 |
arbgcomp.h ARluCompGenEig |
| 64 |
|
| 65 |
|
| 66 |
vi) Reverse communication classes: |
| 67 |
|
| 68 |
file class |
| 69 |
---------- ---------------- |
| 70 |
arrssym.h ARrcSymStdEig |
| 71 |
arrsnsym.h ARrcNonSymStdEig |
| 72 |
arrscomp.h ARrcCompStdEig |
| 73 |
arrgsym.h ARrcSymGenEig |
| 74 |
arrgnsym.h ARrcNonSymGenEig |
| 75 |
arrgcomp.h ARrcCompGenEig |
| 76 |
|
| 77 |
|
| 78 |
vii) Matrix classes: |
| 79 |
|
| 80 |
file class |
| 81 |
---------- ---------------- |
| 82 |
arlsmat.h ARluSymMatrix |
| 83 |
arlspen.h ARluSymPencil |
| 84 |
arlnsmat.h ARluNonSymMatrix |
| 85 |
arlnspen.h ARluNonSymPencil |
| 86 |
arusmat.h ARumSymMatrix |
| 87 |
aruspen.h ARumSymPencil |
| 88 |
arunsmat.h ARumNonSymMatrix |
| 89 |
arunspen.h ARumNonSymPencil |
| 90 |
arbsmat.h ARbdSymMatrix |
| 91 |
arbspen.h ARbdSymPencil |
| 92 |
arbnsmat.h ARbdNonSymMatrix |
| 93 |
arbnspen.h ARbdNonSymPencil |
| 94 |
arhbmat.h ARhbMatrix |
| 95 |
|
| 96 |
|
| 97 |
b) Package interface files: |
| 98 |
|
| 99 |
i) ARPACK FORTRAN interface: |
| 100 |
|
| 101 |
file Contents |
| 102 |
---------- ----------------------------------------------- |
| 103 |
saupp.h Interface with dsaupd and ssaupd subroutines. |
| 104 |
seupp.h Interface with dseupd and sseupd subroutines. |
| 105 |
naupp.h Interface with dnaupd and snaupd subroutines. |
| 106 |
neupp.h Interface with dneupd and sneupd subroutines. |
| 107 |
caupp.h Interface with znaupd and cnaupd subroutines. |
| 108 |
ceupp.h Interface with zneupd and cneupd subroutines. |
| 109 |
debug.h Interface with ARPACK debugging variables. |
| 110 |
arpackf.h Fortran to C function prototypes convertion. |
| 111 |
|
| 112 |
|
| 113 |
ii) LAPACK and BLAS1 interface: |
| 114 |
|
| 115 |
file Contents |
| 116 |
---------- ----------------------------------------------- |
| 117 |
lapackc.h Various LAPACK function declarations. |
| 118 |
lapackf.h Fortran to C function prototypes convertion. |
| 119 |
blas1c.h Various BLAS1 function declarations. |
| 120 |
blas1f.h Fortran to C function prototypes convertion. |
| 121 |
|
| 122 |
|
| 123 |
iii) SuperLU interface: |
| 124 |
|
| 125 |
file Contents |
| 126 |
---------- ----------------------------------------------- |
| 127 |
superluc.h Various SuperLU function declarations. |
| 128 |
arlspdef.h Altered version of ssp_defs.h, dsp_defs.h, |
| 129 |
csp_defs.h and zsp_defs.h header files. |
| 130 |
arlsupm.h Unaltered copy of supermatrix.h header file. |
| 131 |
arlnames.h Unaltered copy of Cnames.h header file. |
| 132 |
arlutil.h Unaltered copy of util.h header file. |
| 133 |
arlcomp.h Unaltered copy of dcomplex.h and scomplex.h. |
| 134 |
|
| 135 |
|
| 136 |
iv) UMFPACK interface: |
| 137 |
|
| 138 |
file Contents |
| 139 |
---------- ----------------------------------------------- |
| 140 |
umfpackc.h Various UMFPACK function declarations. |
| 141 |
umfpackf.h Fortran to C function prototypes convertion. |
| 142 |
|
| 143 |
|
| 144 |
c) Other auxiliary files: |
| 145 |
|
| 146 |
file Contents |
| 147 |
---------- ----------------------------------------------- |
| 148 |
arch.h Machine dependent functions and variable types. |
| 149 |
arcomp.h "arcomplex" complex type definition. |
| 150 |
arerror.h "ArpackError" class definition. |
| 151 |
|
| 152 |
|
| 153 |
|
| 154 |
2) Compiler-dependent instructions. |
| 155 |
|
| 156 |
Some compiler-dependent functions and data types used by arpack++ are |
| 157 |
grouped in the file arch.h. This file should be changed to reflect the |
| 158 |
characteristics of your system. Another file, arcomp.h, contains the |
| 159 |
definition of a class template called arcomplex, created to emulate |
| 160 |
the g++ complex class when another compiler is being used. This file |
| 161 |
must also be changed if g++ (or CC) is not being used. |
| 162 |
|
| 163 |
a) Changing ARPACK++ parameters and definitions included in arch.h: |
| 164 |
|
| 165 |
All ARPACK++ parameters that are not intended to be changed frequently |
| 166 |
were included in the arch.h file. Are defined in this file |
| 167 |
|
| 168 |
i) Some machine and problem-dependent umfpack parameters. |
| 169 |
|
| 170 |
If the umfpack is to be used, the user can modify some of its |
| 171 |
parameters to correctly reflect the environment and the class |
| 172 |
of problems being solved. The constants included in arch.h |
| 173 |
correspond to a subset of the parameters generated by the um21i |
| 174 |
umfpack function. Other relevant parameters can also be passed |
| 175 |
to the ARumNonSymMatrix class constructor. |
| 176 |
|
| 177 |
ii) Some fortran to c conversion functions. |
| 178 |
|
| 179 |
Because fortran and c++ functions tend to have different |
| 180 |
representations in different platforms, a function that |
| 181 |
converts a fortran function name to the c++ format is |
| 182 |
defined in arch.h. This function can be altered by the user |
| 183 |
if the environment being used was not included in arch.h. |
| 184 |
|
| 185 |
iii) Some fortran to c type conversion rules. |
| 186 |
|
| 187 |
arch.h also includes the definition of some rules required |
| 188 |
to convert INTEGER and LOGICAL FORTRAN types to c++. |
| 189 |
|
| 190 |
iv) The c++ bool type. |
| 191 |
|
| 192 |
If the c++ compiler being used does not include a bool type, |
| 193 |
this type can also be defined in arch.h. |
| 194 |
|
| 195 |
|
| 196 |
b) Redefining arcomplex class in arcomp.h: |
| 197 |
|
| 198 |
ARPACK++ uses a self-defined complex class called arcomplex. |
| 199 |
Actually, arcomplex is a class template used to represent |
| 200 |
both single and double precision complex numbers. It was created |
| 201 |
in an effort to permit ARPACK++ to deal with different compilers, |
| 202 |
since c++ does not define a unique complex type. |
| 203 |
arcomplex is intended to emulate the gnu g++ complex class when |
| 204 |
other compilers are being used (when g++ is used, ARPACK++ simply |
| 205 |
declares arcomplex to be the standard complex type). arcomp.h |
| 206 |
includes a complex class definition for the CC compiler only. At |
| 207 |
the present time, no other compiler was used to generate ARPACK++ |
| 208 |
programs, so further work must be done to permit the use of the |
| 209 |
library with other compilers. |
| 210 |
To define a new complex type, the user must create a class |
| 211 |
template (called arcomplex) that contains at least three members: |
| 212 |
|
| 213 |
i) A default constructor; |
| 214 |
ii) A copy constructor; and |
| 215 |
iii) A constructor that takes two real numbers as parameters (one |
| 216 |
is the real and other the imaginary part of the complex number). |
| 217 |
|
| 218 |
Naturally, all usual mathematical operations on complex numbers, |
| 219 |
such as addition, multiplication, multiplication by a real number, |
| 220 |
etc, should also be defined. But because most compilers include a |
| 221 |
complex data type, the simplest way of defining arcomplex is to use |
| 222 |
only the three constructors mentioned above to establish a relation |
| 223 |
between the actual complex class and the gnu g++ standard. |
| 224 |
|
| 225 |
|
| 226 |
5) ARPACK (fortran) authors: |
| 227 |
|
| 228 |
Danny Sorensen ([email protected]) |
| 229 |
Richard Lehoucq ([email protected]) |
| 230 |
Chao Yang ([email protected]) |
| 231 |
Kristi Maschhoff ([email protected]) |
| 232 |
|