commit 3d060c348fdfff074a9b902d56f539664789d831
parent 67c35b638c04ab7a860aac0c166936e9481633c9
Author: Sebastiano Tronto <sebastiano@tronto.net>
Date: Mon, 15 Jul 2024 11:15:35 +0200
Solved bug in cocsep generation
Diffstat:
4 files changed, 24 insertions(+), 61 deletions(-)
diff --git a/TODO.txt b/TODO.txt
@@ -1,30 +1,14 @@
-Bug in cocsepdata
- - Add tests for ttrep
- - check that ttrep indeed brings to representative
- - Fix?
- - Once fixed, fix other tests
- - maybe add longtest, e.g. as a tool?
- - Clean up mixed bfs fromdone / fromnew, use as new for h0k4
+Bug in esep table generation
- Re-do stats
-
-Check stats for all tables using H48stats solver
+ - Add long-running test for h0k4 (maybe as a tool?)
- try DFS for h0 solver
- - compare results, the bfs method could be wrong
- - if faster: remove bfs
- - if slower: why do I get different results with the new bfs?
+ - use dfs for computing big table, save distance %3 until the last two steps,
+ then clean the table and double loop over moves to fill the value
+ - dfs for tables with h=1 to 10?
-Bug in esep table generation
- Fails for UFRUFU, try command
./run solve -solver H48 -options "2;20" -n 1 -M 10 -cube \
"$(./run frommoves -moves "UFRUFU")"
- - Fundamental error in how tables are generated, each coordinate has too
- many representative. I need to use the big table with the full coordinate
- first (h=11, ~241 billion positions, 60Gb with k=2). From this the smaller
- tables can be easily deduced.
- - Investigate the possibility of computing smaller tables directly in some
- other way, even if slow.
- - use dfs for computing big table, save distance %3 until the last two steps,
- then clean the table and double loop over moves to fill the value
Solver
- cleanup h48 solver
@@ -118,6 +102,10 @@ switch. Here NISS may be useful.
* see if vcube's method to flip all corners is better
* find a better way for computing the inverse?
* Transform with big table: make static cube actually static (how?)
+* Use selfsim: in generating some tables, it is in thery possible to only check
+ the few transformations that give self-similarity instead of all 48.
+ The performance drop is almost insignificant, but I would like to figure out
+ the mistake I made previously.
## Improvements and other things
diff --git a/src/solve_h48.h b/src/solve_h48.h
@@ -422,16 +422,10 @@ gendata_h48h0k4_return_size:
_static int64_t
gendata_h48h0k4_bfs(bfsarg_esep_t *arg)
{
-/*
-TODO: the new method gives a slightly different answer. If the new
-method is correct, then the old bfs method is wrong. Which one is it?
-Try also DFS and compare results (it could be faster).
-*/
if (2 * arg->done < (int64_t)ESEP_MAX(0))
return gendata_h48h0k4_bfs_fromdone(arg);
else
return gendata_h48h0k4_bfs_fromnew(arg);
-// return gendata_h48h0k4_bfs_fromnew(arg);
}
_static int64_t
@@ -464,22 +458,7 @@ gendata_h48h0k4_bfs_fromdone(bfsarg_esep_t *arg)
cocsep_coord = j / H48_ESIZE(0);
sim = arg->selfsim[cocsep_coord] >> UINT64_C(1);
for (t = 1; t < 48 && sim; t++, sim >>= UINT64_C(1)) {
- if (!(sim & UINT64_C(1))) {
- transd = transform(moved, t);
- k = coord_h48(transd, arg->cocsepdata, 0);
- if (k != j) {
-/*
-LOG("t=%" PRId64 ", tinv=%" PRIu8 "\n", t, inverse_trans(t));
-int64_t ccm = coord_cocsep(moved);
-int64_t repm = coord_cocsep(arg->crep[j/H48_ESIZE(0)]);
-LOG("moved: full %" PRId64 ", cocsep %" PRId64 ", rep %" PRId64 ", ttrep %" PRId32 "\n", j, ccm, repm, TTREP(arg->cocsepdata[ccm]));
-int64_t cct = coord_cocsep(transd);
-int64_t rept = coord_cocsep(arg->crep[k/H48_ESIZE(0)]);
-LOG("moved: full %" PRId64 ", cocsep %" PRId64 ", rep %" PRId64 ", ttrep %" PRId32 "\n", j, cct, rept, TTREP(arg->cocsepdata[cct]));
-*/
- }
- continue;
- }
+ /* TODO: use only selfsim */
transd = transform(moved, t);
k = coord_h48(transd, arg->cocsepdata, 0);
x = get_esep_pval(arg->buf32, k);
@@ -512,10 +491,7 @@ gendata_h48h0k4_bfs_fromnew(bfsarg_esep_t *arg)
j = coord_h48(moved, arg->cocsepdata, 0);
x = get_esep_pval(arg->buf32, j);
if (x < arg->depth)
-{
-if (x < arg->depth -1) LOG("WAT %" PRIu8 " while scanning %" PRIu8 "\n",x, arg->depth);
goto neighbor_found;
-}
}
continue;
neighbor_found:
@@ -524,8 +500,7 @@ neighbor_found:
cocsep_coord = i / H48_ESIZE(0);
sim = arg->selfsim[cocsep_coord] >> 1;
for (t = 1; t < 48 && sim; t++, sim >>= 1) {
- if (!(sim & 1))
- continue;
+ /* TODO: use only selfsim */
transd = transform(cube, t);
j = coord_h48(transd, arg->cocsepdata, 0);
x = get_esep_pval(arg->buf32, j);
diff --git a/test/103_cocsep_selfsim_distribution/00_all.out b/test/103_cocsep_selfsim_distribution/00_all.out
@@ -1,12 +1,12 @@
373 self-similar positions out of 3393
-Size number of groups
-1 3020
-2 311
-3 8
-4 35
-6 9
-8 6
-12 1
-16 1
-24 1
-48 1
+Size number of groups
+1 3020
+2 311
+3 8
+4 35
+6 9
+8 6
+12 1
+16 1
+24 1
+48 1
diff --git a/test/111_gendata_h48_h0/00_h_0.out b/test/111_gendata_h48_h0/00_h_0.out
@@ -19,5 +19,5 @@ h48:
1: 1
2: 4
3: 34
-4: 329
-5: 3587
+4: 331
+5: 3612
