findIntervals is now O(n)

[germs.git] / src / genealgorithms.cpp

diff --git a/src/genealgorithms.cpp b/src/genealgorithms.cpp
index eecdf3fea92bd8048c6c47fd1ef466d5e4698c67..4315cb391e46bfad6c7d10073ee485f2962a4158 100644 (file)
--- a/src/genealgorithms.cpp
+++ b/src/genealgorithms.cpp
@@ -56,12 +56,37 @@ std::vector<std::vector<int> > robinsonSchensted(const GeneOrder& go){
        return v;
 }
 
+struct FindP{
+       size_t p;
+       FindP(size_t p) : p(p) {}
+       bool operator()(Interval i){
+               return (i.first == p || i.second == p);
+       }
+};
+
+
 int countCycles(const GeneOrder& go){
        int cycles = 0;
        set<size_t> marked;
-       for (size_t p = 1; p < go.size() - 1; ++p){
+       vector<Interval> intervals = findIntervals(go);
+       vector<Interval> points = findIntervalsAtPoints(intervals);
+       for (size_t p = 1; p < go.size(); ++p){
                if (marked.find(p) != marked.end())
                        continue;
+               Interval i = intervals[points[p].first];
+               while (marked.find(p) == marked.end()){
+                       marked.insert(p);
+                       if (i == intervals[points[p].first])
+                               i = intervals[points[p].second];
+                       else
+                               i = intervals[points[p].first];
+
+                       if (p == i.first)
+                               p = i.second;
+                       else
+                               p = i.first;
+               }
+               ++cycles;
        }
        return cycles;
 }
@@ -70,33 +95,56 @@ std::vector<Component> findComponents(const GeneOrder& go){
        return vector<Component>();
 }
 
-//TODO: Think of a better than O(n^2) implementation
+/**
+ * 
+ */
 std::vector<Interval> findIntervals(const GeneOrder& go){
-       vector<Interval> intervals;
-       for (size_t i = 0; i < go.size() - 1; ++i){
+       vector<Interval> intervals(go.size()-1,Interval(go.size(),go.size()));
+       size_t n = 0;
+       for (GeneOrder::iterator g = go.begin(); g != go.end(); ++g, ++n){
+                       size_t i = abs(*g);
+                       if (i < go.size() - 1)
+                               intervals[i].first = n + (*g >= 0 ? 1 : 0);
+                       if (i > 0)
+                               intervals[i-1].second = n + (*g < 0 ? 1 : 0);
+       }
+       return intervals;
+}
+
+/**
+ * TODO: Think of a better than O(n^2) implementation.
+ * Possibly move to cache result
+ */
+std::vector<Interval> findIntervalsAtPoints(const vector<Interval>& intervals){
+       vector<Interval> points;
+       points.push_back(Interval(intervals.size(),intervals.size())); //Dummy interval to match point and index
+       for (size_t p = 1; p <= intervals.size(); ++p){
                size_t f = 0;
                size_t s = 0;
                bool found = false;
                size_t n = 0;
-               for (GeneOrder::iterator g = go.begin(); g != go.end(); ++g, ++n){
-                       if (static_cast<size_t>(abs(*g)) == i){
-                               f = n;
-                               if (*g >= 0)
-                                       ++f;
-                               if (found)
+               for (vector<Interval>::const_iterator i = intervals.begin(); i != intervals.end(); ++i, ++n){
+                       if (i->first == p){
+                               if (!found){
+                                       f = n;
+                                       found = true;
+                               }else{
+                                       s = n;
                                        break;
-                               found = true;
+                               }
                        }
-                       if(static_cast<size_t>(abs(*g)) == i+1){
-                               s = n;
-                               if (*g < 0)
-                                       ++s;
-                               if (found)
+                       if (i->second == p){
+                               if (!found){
+                                       f = n;
+                                       found = true;
+                               }else{
+                                       s = n;
                                        break;
-                               found = true;
+                               }
                        }
                }
-               intervals.push_back(Interval(f,s));
+               points.push_back(Interval(f,s));
        }
-       return intervals;
+       return points;
+
 }