ulrikey
343709585a
changed relax_weight to relax_weight[level] in line 433
1999-06-10 15:40:17 +00:00
ulrikey
585f0d4846
changed relax_weight from double to double * in HYPRE_ParAMGSetRelaxWeight
1999-06-08 16:10:03 +00:00
ulrikey
0539bcf95f
changed relax_weight to be double * so that relax_Weight can be varied on each
...
level
1999-06-03 23:39:46 +00:00
ulrikey
75303dcb1f
changed relax_weight to be double * so that relax_weight can be varied on each
...
level
1999-06-03 23:35:52 +00:00
ulrikey
9312d86f3a
changed relax_weight to be double * so that relax_Weight can be varied on each
...
level
1999-06-03 23:34:14 +00:00
ulrikey
7cba13b6d4
changed relax_weight to be double * so that relax_Weight can be varied on each
...
level
relax_weight is now printed for each level
1999-06-03 23:32:02 +00:00
ulrikey
963720c48d
if relax_weight is zero , relax_weight will be evaluated as 1/|D^-1/2AD^-1/2|
1999-06-03 23:30:53 +00:00
ulrikey
1dfc34b84b
changed relax_weight to be double * so that relax_weight can be varied on each
...
level
1999-06-03 23:24:51 +00:00
ulrikey
e6b2389288
computes the norm of D^-1/2AD^-1/2 where D is the diagonal matrix with
...
the diagonal of A
1999-06-03 23:22:27 +00:00
ulrikey
f15c1f17c6
added par_scaled_matnorm.c
1999-06-03 23:20:46 +00:00
ulrikey
9cd8614493
changed relax_weight to be double * so that relax_weight can be varied on each
...
level
added hypre_ParScaledMatNorm
1999-06-03 23:19:22 +00:00
ulrikey
4e138c5c2f
changed relax_weight to be double * so that relax_Weight can be varied on each level
1999-06-03 23:12:18 +00:00
vhenson
e8a4203712
Set PCG and GMRES to work with -tol
1999-05-28 17:27:12 +00:00
vhenson
5c78284a1b
Added coarsen_type and Measure_type to PCG and GMRES routines
1999-05-15 00:34:24 +00:00
cleary
4408b791ba
Changed MPI_INTEGER to MPI_INT in C program.
1999-05-14 21:04:12 +00:00
ulrikey
68a4655c2b
added options -rlx 1 (sequential Gauss-Seidel, performs one processor after
...
another) and
-rlx 4 (Gauss-Seidel, evaluates interior points in parallel, points on
processor boundaries sequentially)
1999-05-14 18:08:52 +00:00
vhenson
c99a0c26ba
replaced ranlib with @RANLIB@
1999-05-12 18:17:24 +00:00
cleary
2deb2be62a
Fixed memory bug in parcsr_pilut.
1999-05-10 23:52:19 +00:00
falgout
7fab661e0b
Removed __STDC__ symbol generated by mkproto by introducing a
...
new 'protos' script in the utilities directory.
1999-05-08 00:26:20 +00:00
falgout
4175304e4f
Added debug_flag argument to ParAMGCoarsen call.
1999-05-08 00:22:36 +00:00
ulrikey
da359d1c62
added check for rhs = 0 , if rhs = 0 the routine sets x = 0 and returns
1999-05-07 18:25:34 +00:00
ulrikey
dcd91537e0
added parameter debug_flag to hypre_ParAMGCOarsen
1999-05-07 17:35:03 +00:00
ulrikey
a9b3da8a31
replaced serial search by binary search
1999-05-07 17:33:15 +00:00
ulrikey
ad5fe2a19e
replaced serial search through binary search
...
added debug_flag in parameter list of hypre_ParCoarsen for more detailed
timings for -dbg 3
1999-05-07 17:32:39 +00:00
cleary
ff1b6b1e10
Added automatic finding of BLAS libraries.
1999-05-06 18:36:11 +00:00
cleary
a7186e7303
Wrapped pilut for use with gmres, etc.
...
Added "GetRow" implementation for parcsr_matrix.
1999-05-05 22:04:43 +00:00
ulrikey
282b527748
added computation of operator and grid complexity here
1999-04-30 22:09:19 +00:00
ulrikey
b9e94cc94f
changed order to relax point for option -rugerlx
1999-04-29 19:21:33 +00:00
vhenson
3467516f23
Corrected error in coarsen_type fo relaxation of ruge3rlx
1999-04-29 18:10:21 +00:00
vhenson
d8761b913d
Performed some general clean-up regarding command-line arguments.
...
Made hybrid Jacobi/Gauss-Seidel the default. Interpolation truncation
is now activated by -tr X.X where X.X is any non-zero real number
0 < X.X < 1.0 (0.0 means no truncation is performed, and is the
default).
1999-04-28 23:07:39 +00:00
vhenson
c2d14796a4
fixed hiny problem
1999-04-28 20:15:29 +00:00
vhenson
f003a8c7b8
Removed useless final communication
1999-04-28 06:16:59 +00:00
ulrikey
327a67326e
added breaks
1999-04-27 22:55:05 +00:00
vhenson
6d322c002e
Put 'break' in loop over A_ext, to shorten search.
1999-04-27 22:03:18 +00:00
vhenson
bcd3b58410
changed hx,hy,hz to hinx,hiny,hinz to avoid the Blue predefined hz problem
1999-04-27 21:40:59 +00:00
vhenson
b533bf557c
changed internal debug instrumentation to debug_flag==4
1999-04-27 19:27:03 +00:00
vhenson
380f19d713
Well, this time the /* problem really is fixed
1999-04-27 15:46:08 +00:00
vhenson
37f93038a9
removed an extraneous /* within a comment
1999-04-27 00:18:53 +00:00
vhenson
e58a47511f
corrected erroneous comment statement
1999-04-26 20:53:07 +00:00
vhenson
6841e51a23
added difconv
1999-04-26 20:51:11 +00:00
vhenson
1390784869
par_difconv.c diffusion-convection operator
1999-04-26 20:43:55 +00:00
vhenson
0c5759e5d5
added convection-diffusion operator -difconv
1999-04-26 20:42:45 +00:00
vhenson
d641b0b0e3
added difconv.c
1999-04-26 20:41:02 +00:00
vhenson
bd9012a77f
added -dbg 3 for instrumenting par_interp.c
1999-04-26 19:21:09 +00:00
vhenson
af7dd24a98
Added -dbg 3 for instrumenting par_interp.c
1999-04-26 19:20:40 +00:00
ulrikey
1835cdf5ce
added option ruge3c which treats boundary entries into CF_marker in the
...
following way: it only accepts values from processors with proc_id > my_id,
but does not overwrite corase points
1999-04-23 18:22:02 +00:00
ulrikey
42e1cdff4a
changed the coarsen_types 2 (ruge2b) and 3 (ruge3) to use the strategy
...
largest processor wins at the boundaries, i.e. new entries in CF_marker
for points at the boundary are only accepted if they are from a processor
with a larger id than my_id
also added coarsen_type 4 (ruge3c) which is similar to above with the
modification that coarse points will not be overwritten
1999-04-23 18:19:27 +00:00
ulrikey
d183a24462
fixed bug in hypre_ParCoarsenFalgout
1999-04-01 17:57:17 +00:00
dwalker
b4b7826145
removed -DHYPRE_TIMING as a default
1999-03-31 18:27:28 +00:00
dwalker
ed5096a397
changed min max to hypre_min hypre_max
1999-03-30 22:25:41 +00:00
