Some additional manual formatting.

Change-Id: Ie5590535d013aa2f747dd034fa2fcd2ae5c3956b Reviewed-on: https://gerrit.libreoffice.org/72226 Tested-by: Jenkins Reviewed-by: Samuel Mehrbrodt <Samuel.Mehrbrodt@cib.de>
2019-05-17 08:45:46 +02:00
parent acd4625f2f
commit 5ee453818c
13 changed files with 74 additions and 55 deletions
@@ -46,6 +46,7 @@ public class DEPSAgent implements ILibEngine {

  // Describes the problem to be solved
  private ProblemEncoder problemEncoder;
+
  // Forms the goodness landscape
  private IGoodnessCompareEngine qualityComparator;

@@ -57,8 +58,10 @@ public class DEPSAgent implements ILibEngine {

  // the own memory: store the point that generated in old learning cycle
  private BasicPoint pold_t;
+
  // the own memory: store the point that generated in last learning cycle
  private BasicPoint pcurrent_t;
+
  // the own memory: store the personal best point
  private SearchPoint pbest_t;

@@ -109,11 +112,12 @@ public class DEPSAgent implements ILibEngine {
  }

  public void generatePoint() {
-// generates a new point in the search space (S) based on
-// its memory and the library
+    // generates a new point in the search space (S) based on
+    // its memory and the library
    selectGTBehavior = this.getGTBehavior();
    selectGTBehavior.generateBehavior(trailPoint, problemEncoder);
-// evaluate into goodness information
+
+    // evaluate into goodness information
    problemEncoder.evaluate(trailPoint);
  }

@@ -37,9 +37,14 @@ import net.adaptivebox.problem.ProblemEncoder;
 import net.adaptivebox.space.BasicPoint;

 public class DEGTBehavior extends AbsGTBehavior implements ILibEngine {
-  private static final int DVNum = 2; // Number of differential vectors, normally be 1 or 2
-  public double FACTOR = 0.5; // scale constant: (0, 1.2], normally be 0.5
-  public double CR = 0.9; // crossover constant: [0, 1], normally be 0.1 or 0.9
+  //Number of differential vectors, normally be 1 or 2
+  private static final int DVNum = 2;
+
+  //scale constant: (0, 1.2], normally be 0.5
+  public double FACTOR = 0.5;
+
+  //crossover constant: [0, 1], normally be 0.1 or 0.9
+  public double CR = 0.9;

  // the own memory: store the point that generated in last learning cycle
  private SearchPoint pbest_t;
@@ -64,20 +64,23 @@ import net.adaptivebox.space.DesignSpace;

 public class PSGTBehavior extends AbsGTBehavior {
  // Two normally choices for (c1, c2, weight), i.e., (2, 2, 0.4), or (1.494,
-  // 1.494, 0.729)
-  // The first is used in dissipative PSO (cf. [4]) as CL>0, and the second is
-  // achieved by using
-  // constriction factors (cf. [3])
+  // 1.494, 0.729) The first is used in dissipative PSO (cf. [4]) as CL>0, and
+  // the second is achieved by using constriction factors (cf. [3])
  public double c1 = 2;
  public double c2 = 2;
-  public double weight = 0.4; // inertia weight

-  public double CL = 0; // See ref[4], normally be 0.001~0.005
+  //inertia weight
+  public double weight = 0.4;
+
+  //See ref[4], normally be 0.001~0.005
+  public double CL = 0;

  // the own memory: store the point that generated in old learning cycle
  private BasicPoint pold_t;
+
  // the own memory: store the point that generated in last learning cycle
  private BasicPoint pcurrent_t;
+
  // the own memory: store the personal best point
  private SearchPoint pbest_t;

@@ -100,7 +103,8 @@ public class PSGTBehavior extends AbsGTBehavior {
        deltaxb = weight * (pcurrent_t.getLocation()[b] - pold_t.getLocation()[b])
            + c1 * Math.random() * (pbest_t.getLocation()[b] - pcurrent_t.getLocation()[b])
            + c2 * Math.random() * (gbest_t.getLocation()[b] - pcurrent_t.getLocation()[b]);
-// limitation for delta_x
+
+        // limitation for delta_x
        deltaxbm = 0.5 * designSpace.getMagnitudeIn(b);
        if (deltaxb < -deltaxbm) {
          deltaxb = -deltaxbm;
@@ -118,5 +122,4 @@ public class PSGTBehavior extends AbsGTBehavior {
    pold_t.importLocation(pcurrent_t);
    pcurrent_t.importLocation(trailPoint);
  }
-
 }
@@ -42,13 +42,13 @@ public class EvalStruct {
    evalRes[0] = evalRes[1] = 0;
    for (int i = 0; i < evalElems.length; i++) {
      if (evalElems[i].isOptType()) {
-// The objectives (OPTIM type)
-// The multi-objective will be translated into single-objective
+        // The objectives (OPTIM type)
+        // The multi-objective will be translated into single-objective
        evalRes[1] += evalElems[i].evaluateOPTIM(targetValues[i]);
      } else {
-// The constraints (CONS type)
-// If evalRes[0] equals to 0, then be a feasible point, i.e. satisfies
-// all the constraints
+        // The constraints (CONS type)
+        // If evalRes[0] equals to 0, then be a feasible point, i.e. satisfies
+        // all the constraints
        evalRes[0] += evalElems[i].evaluateCONS(targetValues[i]);
      }
    }
@@ -51,8 +51,8 @@ public class ACRComparator implements IGoodnessCompareEngine, IUpdateCycleEngine
  public ACRComparator(Library lib, int T) {
    socialPool = lib;
    this.T = T;
-// set the (epsilon_t|t=0) as the maximum CONS value among the SearchPoints in
-// the library
+
+    // set the (epsilon_t|t=0) as the maximum CONS value among the SearchPoints in the library
    epsilon_t = lib.getExtremalVcon(true);
  }

@@ -74,8 +74,9 @@ public class ACRComparator implements IGoodnessCompareEngine, IUpdateCycleEngine
  }

  public void updateCycle(int t) {
-// calculates the ratio
+    // calculates the ratio
    double rn = (double) socialPool.getVconThanNum(epsilon_t) / (double) socialPool.getPopSize();
+
    if (t > TthR * T && T != -1) { // Forcing sub-rule
      epsilon_t *= BETAF;
    } else { // Ratio-keeping sub-rules
@@ -28,9 +28,7 @@ package net.adaptivebox.goodness;

 public class BCHComparator implements IGoodnessCompareEngine {

-  /*
-   * check the magnitude of two array, the frontal is more important
-   **/
+  /* check the magnitude of two array, the frontal is more important */
  private static int compareArray(double[] fit1, double[] fit2) {
    for (int i = 0; i < fit1.length; i++) {
      if (fit1[i] > fit2[i]) {
@@ -29,9 +29,13 @@ public abstract interface IGoodnessCompareEngine {
  int LESS_THAN = 0;

  /**
-   * check the magnitude of two IEncodeEngine LARGER_THAN: goodness1 is worse than
-   * goodness2 LESS_THAN: goodness1 is better than goodness2 EQUAL_TO : goodness1
-   * is equal to goodness2
-   **/
+   * check the magnitude of two IEncodeEngine
+   *
+   * LARGER_THAN: goodness1 is worse than goodness2
+   *
+   * LESS_THAN: goodness1 is better than goodness2
+   *
+   * EQUAL_TO : goodness1 is equal to goodness2
+   */
  int compare(double[] goodness1, double[] goodness2);
 }
@@ -99,5 +99,4 @@ public class Library {
    }
    return num;
  }
-
 }
@@ -25,8 +25,7 @@ import net.adaptivebox.space.BasicPoint;

 public class SearchPoint extends BasicPoint implements IEncodeEngine {
  // store the encode information for goodness evaluation
-  // encodeInfo[0]: the sum of constraints (if it equals to 0, then be a feasible
-  // point)
+  // encodeInfo[0]: the sum of constraints (if it equals to 0, then be a feasible point)
  // encodeInfo[1]: the value of objective function
  private final double[] encodeInfo = new double[2];
  private double objectiveValue;
@@ -68,5 +67,4 @@ public class SearchPoint extends BasicPoint implements IEncodeEngine {
  public boolean isFeasible() {
    return encodeInfo[0] == 0; // no constraint violations
  }
-
-}
+}
@@ -85,11 +85,13 @@ public abstract class ProblemEncoder {

  // evaluate the point into encoded information
  public void evaluate(SearchPoint point) {
-// copy to temp point
+    // copy to temp point
    System.arraycopy(point.getLocation(), 0, this.tempLocation, 0, tempLocation.length);
-// mapping the temp point to original search space S
+
+    // mapping the temp point to original search space S
    designSpace.getMappingPoint(tempLocation);
-// calculate based on the temp point
+
+    // calculate based on the temp point
    calcTargets(tempResponseSet, tempLocation);
    evalStruct.evaluate(point.getEncodeInfo(), tempResponseSet);
    point.setObjectiveValue(tempResponseSet[0]);
@@ -43,22 +43,24 @@ import net.adaptivebox.knowledge.SearchPoint;

 public class SCAgent {

-  // Describes the problem to be solved (encode the point into intermediate
-  // information)
+  // Describes the problem to be solved (encode the point into intermediate information)
  private ProblemEncoder problemEncoder;
+
  // Forms the goodness landscape
  private IGoodnessCompareEngine specComparator;

  // the coefficients of SCAgent
  private static final int TaoB = 2;
-  // The early version set TaoW as the size of external library (NL), but 4 is
-  // often enough
+
+  // The early version set TaoW as the size of external library (NL), but 4 is often enough
  private static final int TaoW = 4;

  // The referred external library
  private Library externalLib;
+
  // store the point that generated in current learning cycle
  private SearchPoint trailPoint;
+
  // the own memory: store the point that generated in last learning cycle
  private SearchPoint pcurrent_t;

@@ -77,9 +79,10 @@ public class SCAgent {
  }

  public SearchPoint generatePoint() {
-// generate a new point
+    // generate a new point
    generatePoint(trailPoint);
-// evaluate the generated point
+
+    // evaluate the generated point
    problemEncoder.evaluate(trailPoint);
    return trailPoint;
  }
@@ -87,13 +90,14 @@ public class SCAgent {
  private void generatePoint(ILocationEngine tempPoint) {
    SearchPoint Xmodel, Xrefer, libBPoint;

-// choose Selects a better point (libBPoint) from externalLib (L) based
-// on tournament selection
+    // choose Selects a better point (libBPoint) from externalLib (L) based
+    // on tournament selection
    int xb = externalLib.tournamentSelection(specComparator, TaoB, true);
    libBPoint = externalLib.getSelectedPoint(xb);
-// Compares pcurrent_t with libBPoint
-// The better one becomes model point (Xmodel)
-// The worse one becomes refer point (Xrefer)
+
+    // Compares pcurrent_t with libBPoint
+    // The better one becomes model point (Xmodel)
+    // The worse one becomes refer point (Xrefer)
    if (specComparator.compare(pcurrent_t.getEncodeInfo(),
        libBPoint.getEncodeInfo()) == IGoodnessCompareEngine.LARGER_THAN) {
      Xmodel = libBPoint;
@@ -102,19 +106,22 @@ public class SCAgent {
      Xmodel = pcurrent_t;
      Xrefer = libBPoint;
    }
-// observational learning: generates a new point near the model point, which
-// the variation range is decided by the difference of Xmodel and Xrefer
+
+    // observational learning: generates a new point near the model point, which
+    // the variation range is decided by the difference of Xmodel and Xrefer
    inferPoint(tempPoint, Xmodel, Xrefer, problemEncoder.getDesignSpace());
  }

  // 1. Update the current point into the external library
  // 2. Replace the current point by the generated point
  public void updateInfo() {
-// Selects a bad point kw from TaoW points in Library
+    // Selects a bad point kw from TaoW points in Library
    int xw = externalLib.tournamentSelection(specComparator, TaoW, false);
-// Replaces kw with pcurrent_t
+
+    // Replaces kw with pcurrent_t
    externalLib.getSelectedPoint(xw).importPoint(pcurrent_t);
-// Replaces pcurrent_t (x(t)) with trailPoint (x(t+1))
+
+    // Replaces pcurrent_t (x(t)) with trailPoint (x(t+1))
    pcurrent_t.importPoint(trailPoint);
  }

@@ -42,5 +42,4 @@ public class DesignDim {
      return paramBound.maxValue - Math.rint((paramBound.maxValue - value) / grain) * grain;
    }
  }
-
 }
@@ -70,5 +70,4 @@ public class DesignSpace {
      }
    }
  }
-
 }