Add check for number of rewritten links.

src/main/java/nl/peterbloem/motive/MotifModel.java

@@ -72,6 +72,7 @@
 public class MotifModel
 {
 	private static ExecutorService executor = null;
+	private static int maxRW = -1; 
 
 	/**
 	 * Sets the threadpool to use (for the beta model). If not set, the beta 
@@ -83,6 +84,11 @@ public static void setExecutor(ExecutorService executor)
 	{
 		MotifModel.executor = executor;
 	}
+
+	public static void setMaxRW(int maxRW)
+	{
+		MotifModel.maxRW = maxRW;
+	}
 
 
 	public static <L> double size(Graph<L> graph, Graph<L> sub,
@@ -833,7 +839,14 @@ public static double sizeEL(DGraph<?> graph, List<D> degrees, DGraph<?> sub,
 		FrequencyModel<Pair<Integer, Integer>> multiEdges = new FrequencyModel<Pair<Integer, Integer>>();
 		List<List<Integer>> rewiring = new LinkedList<List<Integer>>();
 
-		List<D> sDegrees = subbedDegrees(graph, degrees, occurrences, multiEdges, rewiring);
+		List<D> sDegrees = null;
+		try {
+			sDegrees = subbedDegrees(graph, degrees, occurrences, multiEdges, rewiring);
+		} catch(TooManyRWLinksException e)
+		{
+			Global.log().info("Number of links rewritten too high (with "+occurrences.size()+" instances). Returning Double.POSTIVE_INFINITY.");
+			return Double.POSITIVE_INFINITY;
+		}
 
 		// * store the template graph (as a simple graph) 
 		bits.add("subbed", EdgeListModel.directed(sDegrees, Prior.COMPLETE));
@@ -1344,11 +1357,9 @@ public static List<D> subbedDegrees(
 			multiEdges.add(Pair.p(a, b));
 
 			size ++;
-			if(size % 10000 == 0)
-				System.out.println(size + " rewritten links processed");
+			if(maxRW > 0 && size > maxRW)
+				throw new TooManyRWLinksException();
 		}
-
-		System.out.println(".");
 
 		// * Add each rewritten link _once_
 		for(Pair<Integer, Integer> link : multiEdges.tokens())
@@ -1548,4 +1559,11 @@ private static Pair<Integer, Integer> ordered(int i1, int i2)
 			return new Pair<Integer, Integer>(i1, i2);
 		return new Pair<Integer, Integer>(i2, i1);
 	}
+
+	private static class TooManyRWLinksException extends RuntimeException
+	{
+		private static final long serialVersionUID = -2882656149269609884L;
+
+
+	}
 }

src/main/java/nl/peterbloem/motive/exec/CompareLarge.java

@@ -85,7 +85,12 @@
 public class CompareLarge
 {	
 	private static final int BS_SAMPLES = 10000;
-
+
+	/** 
+	 * Maximum amount of rewritten links.
+	 */
+	public int maxRW = -1;
+
 	/** 
 	 * Number of samples to take to find potential motifs 
 	 */
@@ -137,6 +142,8 @@ public void main() throws IOException
 	{		
 		nl.peterbloem.kit.Global.secureRandom(42);
 
+		MotifModel.setMaxRW(maxRW);
+
 		Global.log().info("Computing motif code lengths");
 
 		final List<D> degrees = graphLoop ? null : DSequenceEstimator.sequence(data);

src/main/java/nl/peterbloem/motive/exec/Run.java

@@ -31,6 +31,11 @@ public class Run
 			usage="Output file.")
 	private static File outFile = new File("./graph.db");
 
+	@Option(
+			name="--fast.max-rw",
+			usage="Maximum number of rewritten links allowed. If the number is higher for a given motif, the motif is ignored. This is useful to limit the memory use for disk-backed graphs. -1 for no limit.")
+	private static int maxRW = -1;
+
 	@Option(
 			name="--samples",
 			usage="Number of samples to take.")
@@ -241,6 +246,7 @@ public static void main(String[] args)
 
     		CompareLarge large = new CompareLarge();
 
+    		large.maxRW = maxRW;
     		large.dataName = file.getName();
     		large.data = data;
     		large.motifMinSize = minSize;