lower the default resolution from 18 back down to 12

(18 was more than needed for most runs),
also add more comments on performance vs resolution


git-svn-id: http://svn.einsteintoolkit.org/cactus/EinsteinAnalysis/AHFinderDirect/trunk@940 f88db872-0e4f-0410-b76b-b9085cfa78c5

1 files changed, 12 insertions, 8 deletions

diff --git a/param.ccl b/param.ccl
index d908172..f5ce1e5 100644
--- a/param.ccl
+++ b/param.ccl
@@ -133,8 +133,10 @@ keyword Jacobian_compute_method "how do we compute the Jacobian matrix?"
 keyword Jacobian_store_solve_method \
   "how do we store/linear-solve the Jacobian matrix?"
 {
+# good for debugging, but very slow for large  N_zones_per_right_angle
 "dense matrix/LAPACK"	:: \
   "store as (Fortran) dense matrix, solve with LAPACK routines"
+# use this for normal runs
 "row-oriented sparse matrix/ILUCG" :: \
   "store as sparse matrix (row-oriented storage format), \
    solve with ILUCG (incomplete LU decomposition - conjugate gradient) method"
@@ -456,8 +458,9 @@ keyword patch_system_type[5] "what type of patch system should we use?"
 # for better performance, without seriously affecting the accuracy
 # with which we can locate the horizon.
 #
-# For any patch system type other than "full sphere", there are patches
-# with 45 degree widths, so this parameter must be even.
+# *** IMPORTANT
+# *** For any patch system type other than "full sphere", there are
+# *** patches with 45 degree widths, so this parameter must be even.
 #
 # Normally we use Simpson's rule (in a variant which doesn't require the
 # number of zones to be even) for angular integrations over the patch
@@ -467,16 +470,17 @@ keyword patch_system_type[5] "what type of patch system should we use?"
 # N_zones_per_right_angle = 3 or 5 for a full sphere patch system,
 # or N_zones_per_right_angle = 6 or 10 for any other patch system type.
 #
-# If you are thinking of setting this to a large value (high resolution),
-# note also that with the current dense-matrix storage of the Jacobian,
-# the memory/running time of the LAPACK linear system solve scales as
-# the 4th/6th power of this parameter!  For example, doubling the resolution
-# takes 16 times as much memory, and 64 times as long to run!
+# For large values of this parameter, this thorn takes CPU time which
+# scales roughly as the ...
+# ... 6th power of this parameter
+#     if Jacobian_store_solve_method = "dense matrix/LAPACK"
+# ... 3rd power of this parameter
+#     if Jacobian_store_solve_method = "row-oriented sparse matrix/ILUCG"
 #
 int N_zones_per_right_angle[5] "sets angular resolution of patch systems"
 {
 1:* :: "any integer >= 1; must be even for patch systems other than full-sphere"
-} 18
+} 12
 
 #
 # This parameter sets the width of the interpatch ghost zones in the