diff options
author | Arnur Nigmetov <anigmetov@lbl.gov> | 2021-05-07 22:38:27 -0700 |
---|---|---|
committer | Arnur Nigmetov <anigmetov@lbl.gov> | 2021-05-07 22:40:37 -0700 |
commit | 2ed9afc052bee7956f6abb195947de1f80cb9d91 (patch) | |
tree | 15211941e0b14ffa2b80ce4de5b66bf89068ac20 /matching/include/matching_distance.hpp | |
parent | b528c4067a8aac346eb307d3c23b82d5953cfe2d (diff) |
Fix bugs in matching distance for modules.
1. Twist reduction should not be used for modules, clearing makes no
sense and creates erroneous points. Just use standard reduction.
2. Infinite points for modules (gens/relations case) are treated
slightly differently.
Diffstat (limited to 'matching/include/matching_distance.hpp')
-rw-r--r-- | matching/include/matching_distance.hpp | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/matching/include/matching_distance.hpp b/matching/include/matching_distance.hpp index f7f44a5..9beab1f 100644 --- a/matching/include/matching_distance.hpp +++ b/matching/include/matching_distance.hpp @@ -362,6 +362,7 @@ namespace md { // TODO: think about this - how to call Hera auto dgm_a = module_a_.weighted_slice_diagram(line); auto dgm_b = module_b_.weighted_slice_diagram(line); + R result; if (params_.hera_epsilon > static_cast<R>(0)) { result = hera::bottleneckDistApprox(dgm_a, dgm_b, params_.hera_epsilon) / ( params_.hera_epsilon + 1); |