Do not compute histogram when using LOG scaler

Author: soerenreichardt

algo-common/src/main/java/org/neo4j/gds/result/AbstractCentralityResultBuilder.java

@@ -25,20 +25,26 @@
 import org.neo4j.gds.core.concurrency.Pools;
 import org.neo4j.gds.core.utils.ProgressTimer;
 import org.neo4j.gds.core.utils.statistics.CentralityStatistics;
+import org.neo4j.gds.scaling.ScalarScaler;
 import org.neo4j.internal.kernel.api.procs.ProcedureCallContext;
 
+import java.util.HashMap;
 import java.util.Map;
 import java.util.Optional;
 import java.util.function.LongToDoubleFunction;
 
 public abstract class AbstractCentralityResultBuilder<WRITE_RESULT> extends AbstractResultBuilder<WRITE_RESULT> {
 
+    private static final String HISTOGRAM_ERROR_KEY = "Error";
+
     private final int concurrency;
     protected boolean buildHistogram;
 
     protected long postProcessingMillis = -1L;
     protected Optional<DoubleHistogram> maybeCentralityHistogram = Optional.empty();
 
+    protected Map<String, Object> histogramError;
+
     protected @Nullable Map<String, Object> centralityHistogramOrNull() {
         return maybeCentralityHistogram.map(histogram -> MapUtil.map(
             "min", histogram.getMinValue(),
@@ -50,10 +56,11 @@ public abstract class AbstractCentralityResultBuilder<WRITE_RESULT> extends Abst
             "p95", histogram.getValueAtPercentile(95),
             "p99", histogram.getValueAtPercentile(99),
             "p999", histogram.getValueAtPercentile(99.9)
-        )).orElse(null);
+        )).orElse(histogramError);
     }
 
     protected LongToDoubleFunction centralityFunction = null;
+    protected ScalarScaler.Variant scaler;
 
     protected AbstractCentralityResultBuilder(
         ProcedureCallContext callContext,
@@ -63,6 +70,7 @@ protected AbstractCentralityResultBuilder(
             .outputFields()
             .anyMatch(s -> s.equalsIgnoreCase("centralityDistribution"));
         this.concurrency = concurrency;
+        this.histogramError = new HashMap<>();
     }
 
     protected abstract WRITE_RESULT buildResult();
@@ -72,17 +80,36 @@ public AbstractCentralityResultBuilder<WRITE_RESULT> withCentralityFunction(Long
         return this;
     }
 
+    public AbstractCentralityResultBuilder<WRITE_RESULT> withScalerVariant(ScalarScaler.Variant scaler) {
+        this.scaler = scaler;
+        return this;
+    }
+
     @Override
     public WRITE_RESULT build() {
         final ProgressTimer timer = ProgressTimer.start();
 
+        var logScaler = scaler == ScalarScaler.Variant.LOG;
         if (buildHistogram && centralityFunction != null) {
-            maybeCentralityHistogram = Optional.of(CentralityStatistics.histogram(nodeCount, centralityFunction,Pools.DEFAULT, concurrency));
+            if (logScaler) {
+                logScalerHistogramError();
+            } else {
+                maybeCentralityHistogram = Optional.of(CentralityStatistics.histogram(
+                    nodeCount,
+                    centralityFunction,
+                    Pools.DEFAULT,
+                    concurrency
+                ));
+            }
         }
         timer.stop();
         this.postProcessingMillis = timer.getDuration();
 
         return buildResult();
     }
 
+    private void logScalerHistogramError() {
+        this.histogramError.put(HISTOGRAM_ERROR_KEY, "Unable to create histogram when using scaler of type " + ScalarScaler.Variant.LOG);
+    }
+
 }

proc/centrality/src/main/java/org/neo4j/gds/pagerank/PageRankProc.java

@@ -19,9 +19,9 @@
  */
 package org.neo4j.gds.pagerank;
 
-import org.neo4j.gds.result.AbstractCentralityResultBuilder;
 import org.neo4j.gds.AlgoBaseProc;
 import org.neo4j.gds.api.NodeProperties;
+import org.neo4j.gds.result.AbstractCentralityResultBuilder;
 import org.neo4j.internal.kernel.api.procs.ProcedureCallContext;
 import org.neo4j.logging.Log;
 
@@ -47,7 +47,8 @@ static <PROC_RESULT, CONFIG extends PageRankConfig> PageRankResultBuilder<PROC_R
         procResultBuilder
             .withDidConverge(!computeResult.isGraphEmpty() && result.didConverge())
             .withRanIterations(!computeResult.isGraphEmpty() ? result.iterations() : 0)
-            .withCentralityFunction(!computeResult.isGraphEmpty() ? computeResult.result().scores()::get : null);
+            .withCentralityFunction(!computeResult.isGraphEmpty() ? computeResult.result().scores()::get : null)
+            .withScalerVariant(computeResult.config().scaler());
 
         return procResultBuilder;
     }

proc/centrality/src/test/java/org/neo4j/gds/pagerank/PageRankWriteProcTest.java

@@ -23,11 +23,12 @@
 import org.junit.jupiter.api.TestFactory;
 import org.junit.jupiter.params.ParameterizedTest;
 import org.junit.jupiter.params.provider.MethodSource;
+import org.neo4j.gds.AlgoBaseProc;
+import org.neo4j.gds.GdsCypher.ModeBuildStage;
 import org.neo4j.gds.WritePropertyConfigProcTest;
 import org.neo4j.gds.compat.MapUtil;
 import org.neo4j.gds.core.CypherMapWrapper;
-import org.neo4j.gds.AlgoBaseProc;
-import org.neo4j.gds.GdsCypher.ModeBuildStage;
+import org.neo4j.gds.scaling.ScalarScaler;
 
 import java.util.Collection;
 import java.util.List;
@@ -119,6 +120,23 @@ void testWriteYields(ModeBuildStage queryBuilder, String testCaseName) {
         )));
     }
 
+    @ParameterizedTest(name = "{1}")
+    @MethodSource("org.neo4j.gds.pagerank.PageRankProcTest#graphVariations")
+    void shouldNotComputeCentralityDistributionOnLogScaler(ModeBuildStage queryBuilder, String testCaseName) {
+        var writeProp = "writeProp";
+        var query = queryBuilder
+            .writeMode()
+            .addParameter("scaler", ScalarScaler.Variant.LOG)
+            .addParameter("writeProperty", writeProp)
+            .yields("centralityDistribution");
+
+        assertCypherResult(query, List.of(
+            Map.of(
+                "centralityDistribution", Map.of("Error", "Unable to create histogram when using scaler of type LOG")
+            )
+        ));
+    }
+
     @Override
     public PageRankWriteConfig createConfig(CypherMapWrapper mapWrapper) {
         return PageRankWriteConfig.of(