int->long

ChangeLog

@@ -1,5 +1,6 @@
 2019-05-04  Benjamin Jean-Marie Tremblay  <[email protected]
 
+* converted a lot of int variables to long
 * faster char to int conversion in klets.cpp
 * countlets is much faster when providing alphabet (using unordered_map)
 * countlets version bumped to 1.3

src/countlets.cpp

@@ -49,15 +49,15 @@ void usage() {
   );
 }
 
-unordered_map<string, unsigned int> count_stream(istream &input, vector<string> klets,
+unordered_map<string, unsigned long> count_stream(istream &input, vector<string> klets,
     unsigned int k) {
 
   char l;
 
   string let;
   let.reserve(k + 1);
 
-  unordered_map<string, unsigned int> counts;
+  unordered_map<string, unsigned long> counts;
   counts.reserve(klets.size());
   for (size_t i = 0; i < klets.size(); ++i) {
     counts[klets[i]] = 0;
@@ -156,7 +156,7 @@ int main(int argc, char **argv) {
 
     /* this version loads the entire sequence into memory */
 
-    vector<unsigned int> counts;
+    vector<unsigned long> counts;
     vector<char> letters;
     size_t seqlen;
     char l;
@@ -199,7 +199,7 @@ int main(int argc, char **argv) {
 
     /* this version only keeps k+1 characters in memory */
 
-    unordered_map<string, unsigned int> counts;
+    unordered_map<string, unsigned long> counts;
 
     if (alph.length() < 1) {
       cerr << "Error: could not parse -a option" << '\n';

src/countwin.cpp

@@ -49,7 +49,7 @@ void usage() {
   );
 }
 
-string make_row(string START, string STOP, vector<unsigned int> counts,
+string make_row(string START, string STOP, vector<unsigned long> counts,
     vector<string> klets, bool nozero) {
 
   string out;
@@ -63,11 +63,11 @@ string make_row(string START, string STOP, vector<unsigned int> counts,
 
 }
 
-string extract_window(istream &input, unsigned int window) {
+string extract_window(istream &input, unsigned long window) {
 
   string out;
   char l;
-  unsigned int counter{0};
+  unsigned long counter{0};
   out.reserve(window);
 
   while (input >> l) {
@@ -84,9 +84,9 @@ int main(int argc, char **argv) {
 
   int ku{1};
   unsigned int k;
-  unsigned int START{1};
+  unsigned long START{1};
   int opt;
-  unsigned int window, STOP, step;
+  unsigned long window, step, STOP;
   size_t alphlen;
   string alph, seq;
   ifstream infile;
@@ -95,7 +95,7 @@ int main(int argc, char **argv) {
   vector<string> klets;
   set<unsigned int> lets_set;
   vector<char> lets_uniq;
-  vector<unsigned int> counts;
+  vector<unsigned long> counts;
 
   while ((opt = getopt(argc, argv, "i:o:a:k:w:s:nh")) != -1) {
     switch (opt) {
@@ -129,13 +129,13 @@ int main(int argc, char **argv) {
                 break;
 
       case 'w': if (optarg) {
-                  window = atoi(optarg);
+                  window = atol(optarg);
                   has_win = true;
                 }
                 break;
 
       case 's': if (optarg) {
-                  step = atoi(optarg);
+                  step = atol(optarg);
                   has_step = true;
                 }
                 break;

src/klets.cpp

@@ -35,8 +35,8 @@ using Clock = chrono::high_resolution_clock;
 vector<string> make_klets(vector<char> lets_uniq, unsigned int k) {
 
   size_t alphlen = lets_uniq.size();
-  unsigned int nlets = pow(alphlen, k);
-  unsigned int let_i, counter, step;
+  unsigned long nlets = pow(alphlen, k);
+  unsigned long let_i, counter, step;
   vector<string> klets(nlets, "");
 
   /* perhaps a bit primitive, but it works */
@@ -49,7 +49,7 @@ vector<string> make_klets(vector<char> lets_uniq, unsigned int k) {
 
     while (counter < nlets) {
 
-      for (unsigned int j = 0; j < step; ++j) {
+      for (unsigned long j = 0; j < step; ++j) {
         klets[counter] += lets_uniq[let_i];
         ++counter;
       }
@@ -67,7 +67,7 @@ vector<string> make_klets(vector<char> lets_uniq, unsigned int k) {
 
 }
 
-vector<unsigned int> count_klets(vector<char> letters, vector<char> lets_uniq,
+vector<unsigned long> count_klets(vector<char> letters, vector<char> lets_uniq,
     unsigned int k, size_t alphlen) {
 
   /* Scales very well with increasing k, but requires having the entire
@@ -80,9 +80,9 @@ vector<unsigned int> count_klets(vector<char> letters, vector<char> lets_uniq,
   #endif
 
   size_t seqlen = letters.size();
-  unsigned int nlets = pow(alphlen, k);
-  unsigned int l, counter;
-  vector<unsigned int> let_counts(nlets, 0);
+  unsigned long nlets = pow(alphlen, k);
+  unsigned long l, counter;
+  vector<unsigned long> let_counts(nlets, 0);
   vector<unsigned int> intletters;
   intletters.reserve(seqlen);
   unordered_map<char, unsigned int> let2int;

src/klets.hpp

@@ -26,7 +26,7 @@
 
 std::vector<std::string> make_klets(std::vector<char> lets_uniq, unsigned int k);
 
-std::vector<unsigned int> count_klets(std::vector<char> letters,
+std::vector<unsigned long> count_klets(std::vector<char> letters,
     std::vector<char> lets_uniq, unsigned int k, size_t alphlen);
 
 #endif

src/shuffle_euler.cpp

@@ -35,19 +35,19 @@ using namespace std;
 using Clock = chrono::high_resolution_clock;
 #endif
 
-vector<vector<unsigned int>> make_edgelist(vector<unsigned int> let_counts,
-    unsigned int nletsm1, size_t alphlen) {
+vector<vector<unsigned long>> make_edgelist(vector<unsigned long> let_counts,
+    unsigned long nletsm1, size_t alphlen) {
 
   /* 1D vector<int> --> 2D vector<vector<int>>
    * The first layer elements are vertices, second layer are the edges.
    */
 
   /* TODO: find a cheaper alternative */
 
-  vector<vector<unsigned int>> edgelist(nletsm1, vector<unsigned int>(alphlen));
-  unsigned int counter{0};
+  vector<vector<unsigned long>> edgelist(nletsm1, vector<unsigned long>(alphlen));
+  unsigned long counter{0};
 
-  for (unsigned int i = 0; i < nletsm1; ++i) {
+  for (unsigned long i = 0; i < nletsm1; ++i) {
 
     for (size_t j = 0; j < alphlen; ++j) {
       edgelist[i][j] = let_counts[counter];
@@ -60,16 +60,16 @@ vector<vector<unsigned int>> make_edgelist(vector<unsigned int> let_counts,
 
 }
 
-vector<unsigned int> find_euler(vector<vector<unsigned int>> edgelist, unsigned int lasti,
-    unsigned int nletsm1, default_random_engine gen, size_t alphlen, unsigned int k,
-    vector<bool> empty_vertices, bool verbose) {
+vector<unsigned long> find_euler(vector<vector<unsigned long>> edgelist,
+    unsigned long lasti, unsigned long nletsm1, default_random_engine gen,
+    size_t alphlen, unsigned int k, vector<bool> empty_vertices, bool verbose) {
 
-  unsigned int u;
-  unsigned int nletsm2 = pow(alphlen, k - 2);
-  unsigned int good_v{0}, counter{0};
+  unsigned long u;
+  unsigned long nletsm2 = pow(alphlen, k - 2);
+  unsigned long good_v{0}, counter{0};
   vector<bool> vertices(nletsm1, false);
-  vector<unsigned int> last_letsi(nletsm1, 0);
-  vector<unsigned int> next_let_i;
+  vector<unsigned long> last_letsi(nletsm1, 0);
+  vector<unsigned long> next_let_i;
   next_let_i.reserve(nletsm1);
 
   /* The idea is to go through and make sure that every last letter for each
@@ -83,26 +83,26 @@ vector<unsigned int> find_euler(vector<vector<unsigned int>> edgelist, unsigned
   vertices[lasti] = true;  /* tree root */
 
   /* I don't think there's a formula for this, so just prepare these beforehand */
-  for (unsigned int i = 0; i < nletsm1; ++i) {
+  for (unsigned long i = 0; i < nletsm1; ++i) {
     next_let_i.push_back(counter * alphlen);
     if (counter == nletsm2 - 1) counter = 0;
     else ++counter;
   }
 
-  for (unsigned int i = 0; i < nletsm1; ++i) {
+  for (unsigned long i = 0; i < nletsm1; ++i) {
     if (empty_vertices[i]) vertices[i] = true;  /* ignore unconnected vertices */
     else ++good_v;
   }
 
   if (verbose) cerr << "    Total vertices to travel: " << good_v << endl;
 
-  for (unsigned int i = 0; i < nletsm1; ++i) {
+  for (unsigned long i = 0; i < nletsm1; ++i) {
 
     u = i;
 
     while (!vertices[u]) {
       /* pick a random possible edge from the vertex */
-      discrete_distribution<unsigned int> next_let(edgelist[u].begin(), edgelist[u].end());
+      discrete_distribution<unsigned long> next_let(edgelist[u].begin(), edgelist[u].end());
       last_letsi[u] = next_let(gen);
       /* now follow the edge to the next vertex */
       if (k == 2)
@@ -130,20 +130,20 @@ vector<unsigned int> find_euler(vector<vector<unsigned int>> edgelist, unsigned
 
 }
 
-vector<vector<unsigned int>> fill_vertices(vector<vector<unsigned int>> edgelist,
-    vector<unsigned int> last_letsi, unsigned int nletsm1, size_t alphlen,
-    unsigned int lasti, default_random_engine gen, vector<bool> empty_vertices) {
+vector<vector<unsigned long>> fill_vertices(vector<vector<unsigned long>> edgelist,
+    vector<unsigned long> last_letsi, unsigned long nletsm1, size_t alphlen,
+    unsigned long lasti, default_random_engine gen, vector<bool> empty_vertices) {
 
   /* The incoming edgelist is just a set of counts for each letter. This
    * will actually create vectors of letter indices based on counts.
    */
 
   /* TODO: find a cheaper alternative */
 
-  vector<vector<unsigned int>> edgelist2(nletsm1);
-  unsigned int b;
+  vector<vector<unsigned long>> edgelist2(nletsm1);
+  unsigned long b;
 
-  for (unsigned int i = 0; i < nletsm1; ++i) {
+  for (unsigned long i = 0; i < nletsm1; ++i) {
 
     if (empty_vertices[i]) continue;
 
@@ -152,7 +152,7 @@ vector<vector<unsigned int>> fill_vertices(vector<vector<unsigned int>> edgelist
     for (size_t j = 0; j < alphlen; ++j) {
 
       b = edgelist[i][j];
-      for (unsigned int h = 0; h < b; ++h) {
+      for (unsigned long h = 0; h < b; ++h) {
         edgelist2[i].push_back(j);
       }
 
@@ -169,15 +169,15 @@ vector<vector<unsigned int>> fill_vertices(vector<vector<unsigned int>> edgelist
 
 }
 
-vector<unsigned int> walk_euler(vector<vector<unsigned int>> edgelist,
+vector<unsigned long> walk_euler(vector<vector<unsigned long>> edgelist,
     size_t seqlen, vector<char> lets_uniq, string firstl) {
 
   size_t alphlen = lets_uniq.size();
   size_t nletsm1 = edgelist.size();
-  unsigned int current{0};
+  unsigned long current{0};
   size_t n = firstl.length();
-  vector<unsigned int> edgelist_counter(nletsm1, 0);
-  vector<unsigned int> out_i;
+  vector<unsigned long> edgelist_counter(nletsm1, 0);
+  vector<unsigned long> out_i;
   out_i.reserve(seqlen);
 
   /* initialize shuffled sequence with starting vertex */
@@ -219,14 +219,14 @@ string shuffle_euler(vector<char> letters, default_random_engine gen, unsigned i
   #endif
 
   size_t seqlen = letters.size();
-  unsigned int nlets, nletsm1;
+  unsigned long nlets, nletsm1;
   size_t alphlen;
-  unsigned int lasti{0};
-  vector<unsigned int> last_letsi, out_i;
-  vector<unsigned int> let_counts;
+  unsigned long lasti{0};
+  vector<unsigned long> last_letsi, out_i;
+  vector<unsigned long> let_counts;
   vector<char> lets_uniq;
   set<unsigned int> lets_set;
-  vector<vector<unsigned int>> edgelist;
+  vector<vector<unsigned long>> edgelist;
   string firstl, out;
 
   /* the first and last letters remain unchanged; these are special vertices
@@ -275,7 +275,7 @@ string shuffle_euler(vector<char> letters, default_random_engine gen, unsigned i
    */
   vector<bool> empty_vertices;
   empty_vertices.reserve(nletsm1);
-  for (unsigned int i = 0; i < nletsm1; ++i) {
+  for (unsigned long i = 0; i < nletsm1; ++i) {
     empty_vertices.push_back(true);
     for (size_t j = 0; j < alphlen; ++j) {
       if (edgelist[i][j] > 0) {
@@ -303,7 +303,7 @@ string shuffle_euler(vector<char> letters, default_random_engine gen, unsigned i
   #endif
 
   /* delete last edges from edge pool */
-  vector<vector<unsigned int>> edgelist2;
+  vector<vector<unsigned long>> edgelist2;
   for (size_t i = 0; i < last_letsi.size(); ++i) {
     if (i != lasti) --edgelist[i][last_letsi[i]];
   }

src/shuffle_linear.cpp

@@ -29,10 +29,10 @@ string shuffle_linear(vector<char> letters, default_random_engine gen,
 
   /* variables */
 
-  unsigned int seqlen1 = letters.size();
-  unsigned int seqlen2{seqlen1 / k};
-  unsigned int seqrem{seqlen1 % k};
-  unsigned int seqremlen{seqlen1 - seqrem};
+  unsigned long seqlen1 = letters.size();
+  unsigned long seqlen2{seqlen1 / k};
+  unsigned long seqrem{seqlen1 % k};
+  unsigned long seqremlen{seqlen1 - seqrem};
 
   if (verbose) {
     cerr << "  Times split: " << seqlen2 << endl;
@@ -42,20 +42,20 @@ string shuffle_linear(vector<char> letters, default_random_engine gen,
   string out;
   out.reserve(seqlen1);
 
-  vector<unsigned int> seqindex;
+  vector<unsigned long> seqindex;
   seqindex.reserve(seqlen2);
 
   /* shuffle index */
 
-  for (unsigned int i = 0; i < seqlen2; ++i) {
+  for (unsigned long i = 0; i < seqlen2; ++i) {
     seqindex.push_back(i * k);
   }
 
   shuffle(seqindex.begin(), seqindex.end(), gen);
 
   /* build output string from shuffled index */
 
-  for (unsigned int i = 0; i < seqlen2; ++i) {
+  for (unsigned long i = 0; i < seqlen2; ++i) {
     for (unsigned int j = 0; j < k; ++j) {
       out += letters[seqindex[i] + j];
     }
@@ -64,7 +64,7 @@ string shuffle_linear(vector<char> letters, default_random_engine gen,
   /* add leftover letters */
 
   if (seqrem > 0) {
-    for (unsigned int i = seqremlen; i < seqlen1; ++i) {
+    for (unsigned long i = seqremlen; i < seqlen1; ++i) {
       out += letters[i];
     }
   }