Cleaning up the PC model.

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@236 ec762483-ff6d-05da-a07a-a48fb63a330f

6 files changed, 93 insertions, 88 deletions

diff --git a/gi/posterior-regularisation/prjava/src/arr/F.java b/gi/posterior-regularisation/prjava/src/arr/F.java
index 54dadeac..79de5d1a 100644
--- a/gi/posterior-regularisation/prjava/src/arr/F.java
+++ b/gi/posterior-regularisation/prjava/src/arr/F.java
@@ -56,6 +56,7 @@ public class F {
 	}

 	

 	public static double l1norm(double a[]){

+		// FIXME: this isn't the l1 norm for a < 0

 		double norm=0;

 		for(int i=0;i<a.length;i++){

 			norm += a[i];

@@ -63,6 +64,14 @@ public class F {
 		return norm;

 	}

 	

+	public static double l2norm(double a[]){

+		double norm=0;

+		for(int i=0;i<a.length;i++){

+			norm += a[i]*a[i];

+		}

+		return Math.sqrt(norm);

+	}

+	

 	public static int argmax(double probs[])

 	{

 		double m = Double.NEGATIVE_INFINITY;

diff --git a/gi/posterior-regularisation/prjava/src/optimization/gradientBasedMethods/AbstractGradientBaseMethod.java b/gi/posterior-regularisation/prjava/src/optimization/gradientBasedMethods/AbstractGradientBaseMethod.java
index 0a4a5445..2fcb7990 100644
--- a/gi/posterior-regularisation/prjava/src/optimization/gradientBasedMethods/AbstractGradientBaseMethod.java
+++ b/gi/posterior-regularisation/prjava/src/optimization/gradientBasedMethods/AbstractGradientBaseMethod.java
@@ -56,9 +56,10 @@ public abstract class AbstractGradientBaseMethod implements Optimizer{
 		stats.collectInitStats(this, o);
 		direction = new double[o.getNumParameters()];
 		initializeStructures(o, stats, stop);
-		for (currentProjectionIteration = 1; currentProjectionIteration < maxNumberOfIterations; currentProjectionIteration++){		
-//			System.out.println("starting iterations: parameters:" );
-//			o.printParameters();
+		for (currentProjectionIteration = 1; currentProjectionIteration < maxNumberOfIterations; currentProjectionIteration++){
+			//System.out.println("\tgradient descent iteration " + currentProjectionIteration);
+			//System.out.print("\tparameters:" );
+			//o.printParameters();
 			previousValue = currValue;
 			currValue = o.getValue();
 			gradient = o.getGradient();
@@ -76,7 +77,7 @@ public abstract class AbstractGradientBaseMethod implements Optimizer{
 			updateStructuresBeforeStep(o, stats, stop);
 			lso.reset(direction);
 			step = lineSearch.getStepSize(lso);
-//			System.out.println("Leave with step: " + step);
+			//System.out.println("\t\tLeave with step: " + step);
 			if(step==-1){
 				System.out.println("Failed to find step");
 				stats.collectFinalStats(this, o);
diff --git a/gi/posterior-regularisation/prjava/src/optimization/projections/SimplexProjection.java b/gi/posterior-regularisation/prjava/src/optimization/projections/SimplexProjection.java
index eec11bcf..f22afcaf 100644
--- a/gi/posterior-regularisation/prjava/src/optimization/projections/SimplexProjection.java
+++ b/gi/posterior-regularisation/prjava/src/optimization/projections/SimplexProjection.java
@@ -40,7 +40,7 @@ public class SimplexProjection extends Projection{
 		for (int i = 0; i < ds.length; i++) {
 			currentSum+=ds[i];
 			theta = (currentSum-scale)/(i+1);
-			if(ds[i]-theta <= 0){
+			if(ds[i]-theta < -1e-10){
 				break;
 			}
 			previousTheta = theta;
diff --git a/gi/posterior-regularisation/prjava/src/phrase/PhraseCluster.java b/gi/posterior-regularisation/prjava/src/phrase/PhraseCluster.java
index abd868c4..68148248 100644
--- a/gi/posterior-regularisation/prjava/src/phrase/PhraseCluster.java
+++ b/gi/posterior-regularisation/prjava/src/phrase/PhraseCluster.java
@@ -373,12 +373,13 @@ public class PhraseCluster {
 		

 		return primal;

 	}

+	

+	double[] lambda;

 

 	public double PREM_phrase_context_constraints(double scalePT, double scaleCT)

 	{	

 		double[][][] exp_emit = new double [K][n_positions][n_words];

 		double[][] exp_pi = new double[n_phrases][K];

-		double[] lambda = null;

 

 		//E step

 		PhraseContextObjective pco = new PhraseContextObjective(this, lambda, pool, scalePT, scaleCT);

diff --git a/gi/posterior-regularisation/prjava/src/phrase/PhraseContextObjective.java b/gi/posterior-regularisation/prjava/src/phrase/PhraseContextObjective.java
index ff135a3d..a9d3529c 100644
--- a/gi/posterior-regularisation/prjava/src/phrase/PhraseContextObjective.java
+++ b/gi/posterior-regularisation/prjava/src/phrase/PhraseContextObjective.java
@@ -32,7 +32,7 @@ public class PhraseContextObjective extends ProjectedObjective
 	

 	private PhraseCluster c;

 	

-	// un-regularized  unnormalized posterior, p[edge][tag]

+	// un-regularized unnormalized posterior, p[edge][tag]

 	// P(tag|edge) \propto P(tag|phrase)P(context|tag)

 	private double p[][];

 

@@ -144,7 +144,7 @@ public class PhraseContextObjective extends ProjectedObjective
 				gradient[ic]=-q[e][tag];

 			}

 		}

-		//System.out.println("objective " + loglikelihood + " gradient: " + Arrays.toString(gradient));		

+		//System.out.println("objective " + loglikelihood + " ||gradient||_2: " + arr.F.l2norm(gradient));		

 		objectiveTime += System.currentTimeMillis() - begin;

 	}

 	

@@ -154,106 +154,100 @@ public class PhraseContextObjective extends ProjectedObjective
 		long begin = System.currentTimeMillis();

 		List<Future<?>> tasks = new ArrayList<Future<?>>();

 

-		//System.out.println("projectPoint: " + Arrays.toString(point));

+		//System.out.println("\t\tprojectPoint: " + Arrays.toString(point));

 		Arrays.fill(newPoint, 0, newPoint.length, 0);

 		

-		if (scalePT > 0)

+		// first project using the phrase-tag constraints,

+		// for all p,t: sum_c lambda_ptc < scaleP 

+		if (pool == null)

 		{

-			// first project using the phrase-tag constraints,

-			// for all p,t: sum_c lambda_ptc < scaleP 

-			if (pool == null)

+			for (int p = 0; p < c.c.getNumPhrases(); ++p)

 			{

-				for (int p = 0; p < c.c.getNumPhrases(); ++p)

+				List<Edge> edges = c.c.getEdgesForPhrase(p);

+				double[] toProject = new double[edges.size()];

+				for(int tag=0;tag<c.K;tag++)

 				{

-					List<Edge> edges = c.c.getEdgesForPhrase(p);

-					double[] toProject = new double[edges.size()];

-					for(int tag=0;tag<c.K;tag++)

-					{

-						for(int e=0; e<edges.size(); e++)

-							toProject[e] = point[index(edges.get(e), tag, true)];

-						long lbegin = System.currentTimeMillis();

-						projectionPhrase.project(toProject);

-						actualProjectionTime += System.currentTimeMillis() - lbegin;

-						for(int e=0; e<edges.size(); e++)

-							newPoint[index(edges.get(e), tag, true)] = toProject[e];

-					}

+					for(int e=0; e<edges.size(); e++)

+						toProject[e] = point[index(edges.get(e), tag, true)];

+					long lbegin = System.currentTimeMillis();

+					projectionPhrase.project(toProject);

+					actualProjectionTime += System.currentTimeMillis() - lbegin;

+					for(int e=0; e<edges.size(); e++)

+						newPoint[index(edges.get(e), tag, true)] = toProject[e];

 				}

 			}

-			else // do above in parallel using thread pool

-			{	

-				for (int p = 0; p < c.c.getNumPhrases(); ++p)

+		}

+		else // do above in parallel using thread pool

+		{	

+			for (int p = 0; p < c.c.getNumPhrases(); ++p)

+			{

+				final int phrase = p;

+				final double[] inPoint = point;

+				Runnable task = new Runnable()

 				{

-					final int phrase = p;

-					final double[] inPoint = point;

-					Runnable task = new Runnable()

+					public void run()

 					{

-						public void run()

+						List<Edge> edges = c.c.getEdgesForPhrase(phrase);

+						double toProject[] = new double[edges.size()];

+						for(int tag=0;tag<c.K;tag++)

 						{

-							List<Edge> edges = c.c.getEdgesForPhrase(phrase);

-							double toProject[] = new double[edges.size()];

-							for(int tag=0;tag<c.K;tag++)

-							{

-								for(int e=0; e<edges.size(); e++)

-									toProject[e] = inPoint[index(edges.get(e), tag, true)];

-								projectionPhrase.project(toProject);

-								for(int e=0; e<edges.size(); e++)

-									newPoint[index(edges.get(e), tag, true)] = toProject[e];

-							}

-						}		

-					};

-					tasks.add(pool.submit(task));

-				}

+							for(int e=0; e<edges.size(); e++)

+								toProject[e] = inPoint[index(edges.get(e), tag, true)];

+							projectionPhrase.project(toProject);

+							for(int e=0; e<edges.size(); e++)

+								newPoint[index(edges.get(e), tag, true)] = toProject[e];

+						}

+					}		

+				};

+				tasks.add(pool.submit(task));

 			}

 		}

 		//System.out.println("after PT " + Arrays.toString(newPoint));

 	

-		if (scaleCT > 1e-6)

+		// now project using the context-tag constraints,

+		// for all c,t: sum_p omega_pct < scaleC

+		if (pool == null)

 		{

-			// now project using the context-tag constraints,

-			// for all c,t: sum_p omega_pct < scaleC

-			if (pool == null)

+			for (int ctx = 0; ctx < c.c.getNumContexts(); ++ctx)

 			{

-				for (int ctx = 0; ctx < c.c.getNumContexts(); ++ctx)

+				List<Edge> edges = c.c.getEdgesForContext(ctx);

+				double toProject[] = new double[edges.size()];

+				for(int tag=0;tag<c.K;tag++)

 				{

-					List<Edge> edges = c.c.getEdgesForContext(ctx);

-					double toProject[] = new double[edges.size()];

-					for(int tag=0;tag<c.K;tag++)

-					{

-						for(int e=0; e<edges.size(); e++)

-							toProject[e] = point[index(edges.get(e), tag, false)];

-						long lbegin = System.currentTimeMillis();

-						projectionContext.project(toProject);

-						actualProjectionTime += System.currentTimeMillis() - lbegin;

-						for(int e=0; e<edges.size(); e++)

-							newPoint[index(edges.get(e), tag, false)] = toProject[e];

-					}

+					for(int e=0; e<edges.size(); e++)

+						toProject[e] = point[index(edges.get(e), tag, false)];

+					long lbegin = System.currentTimeMillis();

+					projectionContext.project(toProject);

+					actualProjectionTime += System.currentTimeMillis() - lbegin;

+					for(int e=0; e<edges.size(); e++)

+						newPoint[index(edges.get(e), tag, false)] = toProject[e];

 				}

 			}

-			else

+		}

+		else

+		{

+			// do above in parallel using thread pool

+			for (int ctx = 0; ctx < c.c.getNumContexts(); ++ctx)

 			{

-				// do above in parallel using thread pool

-				for (int ctx = 0; ctx < c.c.getNumContexts(); ++ctx)

+				final int context = ctx;

+				final double[] inPoint = point;

+				Runnable task = new Runnable()

 				{

-					final int context = ctx;

-					final double[] inPoint = point;

-					Runnable task = new Runnable()

+					public void run()

 					{

-						public void run()

+						List<Edge> edges = c.c.getEdgesForContext(context);

+						double toProject[] = new double[edges.size()];

+						for(int tag=0;tag<c.K;tag++)

 						{

-							List<Edge> edges = c.c.getEdgesForContext(context);

-							double toProject[] = new double[edges.size()];

-							for(int tag=0;tag<c.K;tag++)

-							{

-								for(int e=0; e<edges.size(); e++)

-									toProject[e] = inPoint[index(edges.get(e), tag, false)];

-								projectionContext.project(toProject);

-								for(int e=0; e<edges.size(); e++)

-									newPoint[index(edges.get(e), tag, false)] = toProject[e];

-							}

+							for(int e=0; e<edges.size(); e++)

+								toProject[e] = inPoint[index(edges.get(e), tag, false)];

+							projectionContext.project(toProject);

+							for(int e=0; e<edges.size(); e++)

+								newPoint[index(edges.get(e), tag, false)] = toProject[e];

 						}

-					};

-					tasks.add(pool.submit(task));

-				}

+					}

+				};

+				tasks.add(pool.submit(task));

 			}

 		}

 		

@@ -283,9 +277,8 @@ public class PhraseContextObjective extends ProjectedObjective
 		double[] tmp = newPoint;

 		newPoint = point;

 		projectionTime += System.currentTimeMillis() - begin;

-

 		

-		//System.out.println("\treturning " + Arrays.toString(tmp));

+		//System.out.println("\t\treturning " + Arrays.toString(tmp));

 		return tmp;

 	}

 	

@@ -405,6 +398,6 @@ public class PhraseContextObjective extends ProjectedObjective
 	// L - KL(q||p) - scalePT * l1lmax_phrase - scaleCT * l1lmax_context

 	public double primal()

 	{

-		return loglikelihood() - KL_divergence() - scalePT * phrase_l1lmax() - scalePT * context_l1lmax();

+		return loglikelihood() - KL_divergence() - scalePT * phrase_l1lmax() - scaleCT * context_l1lmax();

 	}

 }
\ No newline at end of file
diff --git a/gi/posterior-regularisation/prjava/src/phrase/PhraseObjective.java b/gi/posterior-regularisation/prjava/src/phrase/PhraseObjective.java
index 33167c20..0e2ab4b9 100644
--- a/gi/posterior-regularisation/prjava/src/phrase/PhraseObjective.java
+++ b/gi/posterior-regularisation/prjava/src/phrase/PhraseObjective.java
@@ -1,5 +1,6 @@
 package phrase;

 

+import java.util.Arrays;

 import java.util.List;

 

 import optimization.gradientBasedMethods.ProjectedGradientDescent;

@@ -155,7 +156,7 @@ public class PhraseObjective extends ProjectedObjective
 

 	@Override

 	public String toString() {

-		return "No need for pointless toString";

+		return Arrays.toString(parameters);

 	}

 

 	public double [][]posterior(){