@Override
public Double set(int index, Double element) {
if (index < n) stateIn.setEntry(index, element);
else if (index < n + h) stateHidden.setEntry(index - n, element);
else stateOut.setEntry(index - n - h, element);
return element;
}
@Override
public void step() {
stateHidden.setSubVector(0, weights0.operate(stateIn));
for (int i : series(h)) stateHidden.setEntry(i, activation.function(stateHidden.getEntry(i)));
stateOut.setSubVector(0, weights1.operate(stateHidden));
}
public static RealVector randVector(int dim) {
Random r = new Random();
r.setSeed(System.currentTimeMillis());
RealVector res = MathFactory.createRealVector(dim);
for (int i = 0; i < dim; i++) {
res.setEntry(i, r.nextDouble());
}
return res;
}
public RepeatingLSH(List<LSH> lshList) throws MathException {
super(lshList.get(0).getDim(), lshList.get(0).getRandomGenerator());
this.lshList = lshList;
RandomGenerator rg = lshList.get(0).getRandomGenerator();
RandomData rd = new RandomDataImpl(rg);
/*
* Compute a random vector of lshList.size() with each component taken from U(0,10)
*/
randomVec = new ArrayRealVector(lshList.size());
for (int i = 0; i < randomVec.getDimension(); ++i) {
randomVec.setEntry(i, rd.nextUniform(0, 10.0));
}
}
public static void main(String[] args) {
RealMatrix coefficients2 =
new Array2DRowRealMatrix(
new double[][] {
{0.0D, 1.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D},
{0.0D, 0.0D, 0.857D, 0.0D, 0.054D, 0.018D, 0.0D, 0.071D, 0.0D, 0.0D, 0.0D},
{0.0D, 0.0D, 0.0D, 1.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D},
{0.0D, 0.0D, 0.857D, 0.0D, 0.054D, 0.018D, 0.0D, 0.071D, 0.0D, 0.0D, 0.0D},
{0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 1.0D, 0.0D, 0.0D, 0.0D, 0.0D},
{0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 1.0D, 0.0D, 0.0D, 0.0D, 0.0D},
{0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 1.0D, 0.0D, 0.0D},
{0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.6D, 0.4D},
{0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 1.0D},
{0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 1.0D, 0.0D, 0.0D, 1.0D},
{0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D, 0.0D}
},
false);
for (int i = 0; i < 11; i++) {
coefficients2.setEntry(i, i, -1d);
}
coefficients2 = coefficients2.transpose();
DecompositionSolver solver = new LUDecompositionImpl(coefficients2).getSolver();
System.out.println("1 method my Value :");
RealVector constants =
new ArrayRealVector(new double[] {-1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}, false);
RealVector solution = solver.solve(constants);
double[] data = solution.getData();
DecimalFormat df = new DecimalFormat();
df.setRoundingMode(RoundingMode.DOWN);
System.out.println("Корни уравнения:");
for (double dd : data) {
System.out.print(df.format(dd) + " ");
}
System.out.println();
System.out.println(
"Среднее число процессорных операций, выполняемых при одном прогоне алгоритма: "
+ operationsByProcess(data, arr));
System.out.println("Среднее число обращений к файлам:");
for (int i = 1; i < 4; i++) {
System.out.println(" Файл " + i + " : " + fileMiddleRequest(data, arr, i));
}
System.out.println("Среднее количество информации передаваемой при одном обращении к файлам:");
for (int i = 1; i < 4; i++) {
System.out.println(" Файл " + i + " : " + bitsPerFileTransfer(data, arr, i));
}
System.out.println(
"Сумма среднего числа обращений к основным операторам: " + operatorExecute(data, arr));
System.out.println("Средняя трудоемкость этапа: " + middleWork(data, arr));
}
/**
* Compute the aggregated hash of a vector. This is done by taking the output of the k hashes as a
* vector and computing the dot product with a vector of uniformly distributed components between
* 0 and 10.
*
* <p>So, consider a_{0,k} such that a_i ~ U(0,10) and hash functions h_{0,k} treated as 2
* k-dimensional vectors, we return the dot product of h(v) and a.
*/
public long apply(RealVector vector) {
long res = 0;
for (int i = 0; i < lshList.size(); ++i) {
res += randomVec.getEntry(i) * lshList.get(i).apply(vector);
}
return res;
}
/**
* Build a constraint involving two linear equations.
*
* <p>A linear constraint with two linear equation has one of the forms:
*
* <ul>
* <li>l<sub>1</sub>x<sub>1</sub> + ... l<sub>n</sub>x<sub>n</sub> + l<sub>cst</sub> =
* r<sub>1</sub>x<sub>1</sub> + ... r<sub>n</sub>x<sub>n</sub> + r<sub>cst</sub>
* <li>l<sub>1</sub>x<sub>1</sub> + ... l<sub>n</sub>x<sub>n</sub> + l<sub>cst</sub> <=
* r<sub>1</sub>x<sub>1</sub> + ... r<sub>n</sub>x<sub>n</sub> + r<sub>cst</sub>
* <li>l<sub>1</sub>x<sub>1</sub> + ... l<sub>n</sub>x<sub>n</sub> + l<sub>cst</sub> >=
* r<sub>1</sub>x<sub>1</sub> + ... r<sub>n</sub>x<sub>n</sub> + r<sub>cst</sub>
* </ul>
*
* @param lhsCoefficients The coefficients of the linear expression on the left hand side of the
* constraint
* @param lhsConstant The constant term of the linear expression on the left hand side of the
* constraint
* @param relationship The type of (in)equality used in the constraint
* @param rhsCoefficients The coefficients of the linear expression on the right hand side of the
* constraint
* @param rhsConstant The constant term of the linear expression on the right hand side of the
* constraint
*/
public LinearConstraint(
final RealVector lhsCoefficients,
final double lhsConstant,
final Relationship relationship,
final RealVector rhsCoefficients,
final double rhsConstant) {
this.coefficients = lhsCoefficients.subtract(rhsCoefficients);
this.relationship = relationship;
this.value = rhsConstant - lhsConstant;
}
/** {@inheritDoc} */
@Override
public boolean equals(Object other) {
if (this == other) {
return true;
}
if (other instanceof LinearConstraint) {
LinearConstraint rhs = (LinearConstraint) other;
return (relationship == rhs.relationship)
&& (value == rhs.value)
&& coefficients.equals(rhs.coefficients);
}
return false;
}
/**
* Calculates the variance on the y by GLS.
*
* <pre>
* Var(y)=Tr(u' Omega^-1 u)/(n-k)
* </pre>
*
* @return The Y variance
*/
@Override
protected double calculateYVariance() {
RealVector residuals = calculateResiduals();
double t = residuals.dotProduct(getOmegaInverse().operate(residuals));
return t / (X.getRowDimension() - X.getColumnDimension());
}
public void test4() throws Throwable {
java.lang.Double var0 = new java.lang.Double((-1.0d));
double[] var1 = new double[] {var0};
org.apache.commons.math.linear.OpenMapRealVector var2 =
new org.apache.commons.math.linear.OpenMapRealVector(var1);
int var3 = var2.getDimension();
java.lang.Double var4 = new java.lang.Double((-1.0d));
java.lang.Double var5 = new java.lang.Double(0.0d);
java.lang.Double var6 = new java.lang.Double((-1.0d));
double[] var7 = new double[] {var4, var5, var6};
org.apache.commons.math.linear.ArrayRealVector var8 =
new org.apache.commons.math.linear.ArrayRealVector(var7);
java.lang.Double var9 = new java.lang.Double((-1.0d));
java.lang.Double var10 = new java.lang.Double(0.0d);
java.lang.Double var11 = new java.lang.Double((-1.0d));
double[] var12 = new double[] {var9, var10, var11};
org.apache.commons.math.linear.ArrayRealVector var13 =
new org.apache.commons.math.linear.ArrayRealVector(var12);
org.apache.commons.math.linear.RealVector var14 =
var8.ebeDivide((org.apache.commons.math.linear.RealVector) var13);
java.lang.Double var15 = new java.lang.Double((-1.0d));
java.lang.Double var16 = new java.lang.Double(10.0d);
java.lang.Double var17 = new java.lang.Double(100.0d);
int var18 =
org.apache.commons.math.util.MathUtils.compareTo(
(double) var15, (double) var16, (double) var17);
org.apache.commons.math.linear.RealVector var19 = var13.mapPowToSelf((double) var17);
org.apache.commons.math.linear.ArrayRealVector var20 =
new org.apache.commons.math.linear.ArrayRealVector(var3, var17);
java.lang.Double var21 = new java.lang.Double((-1.0d));
double[] var22 = new double[] {var21};
org.apache.commons.math.linear.OpenMapRealVector var23 =
new org.apache.commons.math.linear.OpenMapRealVector(var22);
java.lang.Double var24 = new java.lang.Double((-1.0d));
double[] var25 = new double[] {var24};
org.apache.commons.math.linear.OpenMapRealVector var26 =
new org.apache.commons.math.linear.OpenMapRealVector(var25);
org.apache.commons.math.linear.OpenMapRealVector var27 = var23.add(var26);
org.apache.commons.math.linear.OpenMapRealVector var28 =
new org.apache.commons.math.linear.OpenMapRealVector(var27);
java.lang.Double var29 = new java.lang.Double((-1.0d));
java.lang.Double var30 = new java.lang.Double(0.0d);
java.lang.Double var31 = new java.lang.Double((-1.0d));
double[] var32 = new double[] {var29, var30, var31};
org.apache.commons.math.linear.ArrayRealVector var33 =
new org.apache.commons.math.linear.ArrayRealVector(var32);
java.lang.Double var34 = new java.lang.Double((-1.0d));
java.lang.Double var35 = new java.lang.Double(0.0d);
java.lang.Double var36 = new java.lang.Double((-1.0d));
double[] var37 = new double[] {var34, var35, var36};
org.apache.commons.math.linear.ArrayRealVector var38 =
new org.apache.commons.math.linear.ArrayRealVector(var37);
org.apache.commons.math.linear.RealVector var39 =
var33.ebeDivide((org.apache.commons.math.linear.RealVector) var38);
java.lang.Double var40 = new java.lang.Double(0.0d);
java.lang.Double var41 = new java.lang.Double((-1.0d));
java.lang.Double var42 = new java.lang.Double((-1.0d));
int var43 =
org.apache.commons.math.util.MathUtils.compareTo(
(double) var40, (double) var41, (double) var42);
org.apache.commons.math.linear.RealVector var44 = var33.mapDivide((double) var41);
org.apache.commons.math.linear.RealVector var45 = var28.mapSubtractToSelf((double) var41);
org.apache.commons.math.linear.RealVector var46 = var20.projection(var45);
// Checks the contract: equals-hashcode on var19 and var44
assertTrue(
"Contract failed: equals-hashcode on var19 and var44",
var19.equals(var44) ? var19.hashCode() == var44.hashCode() : true);
// Checks the contract: equals-hashcode on var44 and var19
assertTrue(
"Contract failed: equals-hashcode on var44 and var19",
var44.equals(var19) ? var44.hashCode() == var19.hashCode() : true);
}
public void test11() throws Throwable {
java.lang.Double var0 = new java.lang.Double((-1.0d));
java.lang.Double var1 = new java.lang.Double(0.0d);
java.lang.Double var2 = new java.lang.Double((-1.0d));
double[] var3 = new double[] {var0, var1, var2};
org.apache.commons.math.linear.ArrayRealVector var4 =
new org.apache.commons.math.linear.ArrayRealVector(var3);
java.lang.Double var5 = new java.lang.Double((-1.0d));
java.lang.Double var6 = new java.lang.Double(0.0d);
java.lang.Double var7 = new java.lang.Double((-1.0d));
double[] var8 = new double[] {var5, var6, var7};
org.apache.commons.math.linear.ArrayRealVector var9 =
new org.apache.commons.math.linear.ArrayRealVector(var8);
org.apache.commons.math.linear.RealVector var10 =
var4.ebeDivide((org.apache.commons.math.linear.RealVector) var9);
java.lang.Double var11 = new java.lang.Double((-1.0d));
java.lang.Double var12 = new java.lang.Double(10.0d);
java.lang.Double var13 = new java.lang.Double(100.0d);
int var14 =
org.apache.commons.math.util.MathUtils.compareTo(
(double) var11, (double) var12, (double) var13);
org.apache.commons.math.linear.RealVector var15 = var9.mapPowToSelf((double) var13);
java.lang.Double[] var16 = new java.lang.Double[] {var13};
org.apache.commons.math.linear.ArrayRealVector var17 =
new org.apache.commons.math.linear.ArrayRealVector(var16);
org.apache.commons.math.linear.RealVector var18 = var17.mapCbrtToSelf();
org.apache.commons.math.linear.RealVector var19 = var17.mapFloorToSelf();
org.apache.commons.math.linear.RealVector var20 = var17.mapCosh();
org.apache.commons.math.linear.RealVector var21 = var17.mapAcos();
java.lang.Double var22 = new java.lang.Double((-1.0d));
double[] var23 = new double[] {var22};
org.apache.commons.math.linear.OpenMapRealVector var24 =
new org.apache.commons.math.linear.OpenMapRealVector(var23);
java.lang.Double var25 = new java.lang.Double((-1.0d));
double[] var26 = new double[] {var25};
org.apache.commons.math.linear.OpenMapRealVector var27 =
new org.apache.commons.math.linear.OpenMapRealVector(var26);
org.apache.commons.math.linear.OpenMapRealVector var28 = var24.add(var27);
org.apache.commons.math.linear.OpenMapRealVector var29 =
new org.apache.commons.math.linear.OpenMapRealVector(var28);
java.lang.Double var30 = new java.lang.Double((-1.0d));
java.lang.Double var31 = new java.lang.Double(0.0d);
java.lang.Double var32 = new java.lang.Double((-1.0d));
double[] var33 = new double[] {var30, var31, var32};
org.apache.commons.math.linear.ArrayRealVector var34 =
new org.apache.commons.math.linear.ArrayRealVector(var33);
java.lang.Double var35 = new java.lang.Double((-1.0d));
java.lang.Double var36 = new java.lang.Double(0.0d);
java.lang.Double var37 = new java.lang.Double((-1.0d));
double[] var38 = new double[] {var35, var36, var37};
org.apache.commons.math.linear.ArrayRealVector var39 =
new org.apache.commons.math.linear.ArrayRealVector(var38);
org.apache.commons.math.linear.RealVector var40 =
var34.ebeDivide((org.apache.commons.math.linear.RealVector) var39);
java.lang.Double var41 = new java.lang.Double(0.0d);
java.lang.Double var42 = new java.lang.Double((-1.0d));
java.lang.Double var43 = new java.lang.Double((-1.0d));
int var44 =
org.apache.commons.math.util.MathUtils.compareTo(
(double) var41, (double) var42, (double) var43);
org.apache.commons.math.linear.RealVector var45 = var34.mapDivide((double) var42);
org.apache.commons.math.linear.RealVector var46 = var29.mapSubtractToSelf((double) var42);
boolean var47 = var17.equals((java.lang.Object) var42);
// Checks the contract: equals-hashcode on var15 and var45
assertTrue(
"Contract failed: equals-hashcode on var15 and var45",
var15.equals(var45) ? var15.hashCode() == var45.hashCode() : true);
// Checks the contract: equals-hashcode on var45 and var15
assertTrue(
"Contract failed: equals-hashcode on var45 and var15",
var45.equals(var15) ? var45.hashCode() == var15.hashCode() : true);
}
/**
* Returns the sum of squared residuals.
*
* @return residual sum of squares
* @since 2.2
*/
public double calculateResidualSumOfSquares() {
final RealVector residuals = calculateResiduals();
return residuals.dotProduct(residuals);
}
public void test13() throws Throwable {
java.lang.Double var0 = new java.lang.Double((-1.0d));
double[] var1 = new double[] {var0};
org.apache.commons.math.linear.OpenMapRealVector var2 =
new org.apache.commons.math.linear.OpenMapRealVector(var1);
java.lang.Double var3 = new java.lang.Double((-1.0d));
double[] var4 = new double[] {var3};
org.apache.commons.math.linear.OpenMapRealVector var5 =
new org.apache.commons.math.linear.OpenMapRealVector(var4);
org.apache.commons.math.linear.OpenMapRealVector var6 = var2.add(var5);
org.apache.commons.math.linear.OpenMapRealVector var7 =
new org.apache.commons.math.linear.OpenMapRealVector(var6);
java.lang.Double var8 = new java.lang.Double((-1.0d));
java.lang.Double var9 = new java.lang.Double(0.0d);
java.lang.Double var10 = new java.lang.Double((-1.0d));
double[] var11 = new double[] {var8, var9, var10};
org.apache.commons.math.linear.ArrayRealVector var12 =
new org.apache.commons.math.linear.ArrayRealVector(var11);
java.lang.Double var13 = new java.lang.Double((-1.0d));
java.lang.Double var14 = new java.lang.Double(0.0d);
java.lang.Double var15 = new java.lang.Double((-1.0d));
double[] var16 = new double[] {var13, var14, var15};
org.apache.commons.math.linear.ArrayRealVector var17 =
new org.apache.commons.math.linear.ArrayRealVector(var16);
org.apache.commons.math.linear.RealVector var18 =
var12.ebeDivide((org.apache.commons.math.linear.RealVector) var17);
java.lang.Double var19 = new java.lang.Double(0.0d);
java.lang.Double var20 = new java.lang.Double((-1.0d));
java.lang.Double var21 = new java.lang.Double((-1.0d));
int var22 =
org.apache.commons.math.util.MathUtils.compareTo(
(double) var19, (double) var20, (double) var21);
org.apache.commons.math.linear.RealVector var23 = var12.mapDivide((double) var20);
org.apache.commons.math.linear.RealVector var24 = var7.mapSubtractToSelf((double) var20);
double var25 = var7.getSparcity();
java.lang.Double var26 = new java.lang.Double((-1.0d));
double[] var27 = new double[] {var26};
org.apache.commons.math.linear.OpenMapRealVector var28 =
new org.apache.commons.math.linear.OpenMapRealVector(var27);
int var29 = var28.getDimension();
java.lang.Double var30 = new java.lang.Double((-1.0d));
java.lang.Double var31 = new java.lang.Double(0.0d);
java.lang.Double var32 = new java.lang.Double((-1.0d));
double[] var33 = new double[] {var30, var31, var32};
org.apache.commons.math.linear.ArrayRealVector var34 =
new org.apache.commons.math.linear.ArrayRealVector(var33);
java.lang.Double var35 = new java.lang.Double((-1.0d));
java.lang.Double var36 = new java.lang.Double(0.0d);
java.lang.Double var37 = new java.lang.Double((-1.0d));
double[] var38 = new double[] {var35, var36, var37};
org.apache.commons.math.linear.ArrayRealVector var39 =
new org.apache.commons.math.linear.ArrayRealVector(var38);
org.apache.commons.math.linear.RealVector var40 =
var34.ebeDivide((org.apache.commons.math.linear.RealVector) var39);
java.lang.Double var41 = new java.lang.Double((-1.0d));
java.lang.Double var42 = new java.lang.Double(10.0d);
java.lang.Double var43 = new java.lang.Double(100.0d);
int var44 =
org.apache.commons.math.util.MathUtils.compareTo(
(double) var41, (double) var42, (double) var43);
org.apache.commons.math.linear.RealVector var45 = var39.mapPowToSelf((double) var43);
org.apache.commons.math.linear.ArrayRealVector var46 =
new org.apache.commons.math.linear.ArrayRealVector(var29, var43);
java.lang.Double var47 = new java.lang.Double((-1.0d));
java.lang.Double var48 = new java.lang.Double(10.0d);
java.lang.Double var49 = new java.lang.Double(1.0d);
java.lang.Double var50 = new java.lang.Double(1.0d);
boolean var51 = org.apache.commons.math.util.MathUtils.equals((double) var49, (double) var50);
int var52 =
org.apache.commons.math.util.MathUtils.compareTo(
(double) var47, (double) var48, (double) var49);
java.lang.Double var53 = new java.lang.Double((-1.0d));
java.lang.Double var54 = new java.lang.Double(0.0d);
java.lang.Double var55 = new java.lang.Double((-1.0d));
double[] var56 = new double[] {var53, var54, var55};
org.apache.commons.math.linear.ArrayRealVector var57 =
new org.apache.commons.math.linear.ArrayRealVector(var56);
java.lang.Double var58 = new java.lang.Double((-1.0d));
java.lang.Double var59 = new java.lang.Double(0.0d);
java.lang.Double var60 = new java.lang.Double((-1.0d));
double[] var61 = new double[] {var58, var59, var60};
org.apache.commons.math.linear.ArrayRealVector var62 =
new org.apache.commons.math.linear.ArrayRealVector(var61);
org.apache.commons.math.linear.RealVector var63 =
var57.ebeDivide((org.apache.commons.math.linear.RealVector) var62);
org.apache.commons.math.linear.RealVector var64 = var57.mapLog10ToSelf();
java.lang.Double var65 = new java.lang.Double(1.0d);
org.apache.commons.math.linear.RealVector var66 = var57.mapMultiplyToSelf((double) var65);
double var67 = org.apache.commons.math.util.MathUtils.indicator((double) var65);
java.lang.Double var68 = new java.lang.Double((-1.0d));
java.lang.Double var69 = new java.lang.Double(10.0d);
java.lang.Double var70 = new java.lang.Double(100.0d);
int var71 =
org.apache.commons.math.util.MathUtils.compareTo(
(double) var68, (double) var69, (double) var70);
double var72 =
org.apache.commons.math.util.MathUtils.normalizeAngle((double) var65, (double) var69);
java.lang.Integer var73 = new java.lang.Integer(10);
java.lang.Integer var74 = new java.lang.Integer(10);
double var75 =
org.apache.commons.math.util.MathUtils.binomialCoefficientLog((int) var73, (int) var74);
java.lang.Integer var76 = new java.lang.Integer(0);
java.lang.Long var77 = new java.lang.Long(100L);
int var78 = org.apache.commons.math.util.MathUtils.pow((int) var76, (long) var77);
double var79 =
org.apache.commons.math.util.MathUtils.binomialCoefficientLog((int) var74, (int) var76);
boolean var80 =
org.apache.commons.math.util.MathUtils.equalsIncludingNaN(
(double) var49, (double) var69, (int) var76);
org.apache.commons.math.linear.RealVector var81 = var46.mapMultiplyToSelf((double) var49);
org.apache.commons.math.linear.RealVector var82 = var7.mapPow((double) var49);
// Checks the contract: equals-hashcode on var23 and var45
assertTrue(
"Contract failed: equals-hashcode on var23 and var45",
var23.equals(var45) ? var23.hashCode() == var45.hashCode() : true);
// Checks the contract: equals-hashcode on var45 and var23
assertTrue(
"Contract failed: equals-hashcode on var45 and var23",
var45.equals(var23) ? var45.hashCode() == var23.hashCode() : true);
}
public static String realVecToString(RealVector r) {
String res = "RealVector(" + r.getDimension() + ") of type '" + r.getClass().getName() + "'\n";
return res += Arrays.toString(r.getData()) + "\n";
}
@Override
public Double get(int index) {
if (index < n) return stateIn.getEntry(index);
if (index < n + h) return stateHidden.getEntry(index - n);
return stateOut.getEntry(index - n - h);
}
/** {@inheritDoc} */
@Override
public int hashCode() {
return relationship.hashCode() ^ Double.valueOf(value).hashCode() ^ coefficients.hashCode();
}
/**
* Solve an estimation problem using a least squares criterion.
*
* <p>This method set the unbound parameters of the given problem starting from their current
* values through several iterations. At each step, the unbound parameters are changed in order to
* minimize a weighted least square criterion based on the measurements of the problem.
*
* <p>The iterations are stopped either when the criterion goes below a physical threshold under
* which improvement are considered useless or when the algorithm is unable to improve it (even if
* it is still high). The first condition that is met stops the iterations. If the convergence it
* not reached before the maximum number of iterations, an {@link EstimationException} is thrown.
*
* @param problem estimation problem to solve
* @exception EstimationException if the problem cannot be solved
* @see EstimationProblem
*/
@Override
public void estimate(EstimationProblem problem) throws EstimationException {
initializeEstimate(problem);
// work matrices
double[] grad = new double[parameters.length];
ArrayRealVector bDecrement = new ArrayRealVector(parameters.length);
double[] bDecrementData = bDecrement.getDataRef();
RealMatrix wGradGradT = MatrixUtils.createRealMatrix(parameters.length, parameters.length);
// iterate until convergence is reached
double previous = Double.POSITIVE_INFINITY;
do {
// build the linear problem
incrementJacobianEvaluationsCounter();
RealVector b = new ArrayRealVector(parameters.length);
RealMatrix a = MatrixUtils.createRealMatrix(parameters.length, parameters.length);
for (int i = 0; i < measurements.length; ++i) {
if (!measurements[i].isIgnored()) {
double weight = measurements[i].getWeight();
double residual = measurements[i].getResidual();
// compute the normal equation
for (int j = 0; j < parameters.length; ++j) {
grad[j] = measurements[i].getPartial(parameters[j]);
bDecrementData[j] = weight * residual * grad[j];
}
// build the contribution matrix for measurement i
for (int k = 0; k < parameters.length; ++k) {
double gk = grad[k];
for (int l = 0; l < parameters.length; ++l) {
wGradGradT.setEntry(k, l, weight * gk * grad[l]);
}
}
// update the matrices
a = a.add(wGradGradT);
b = b.add(bDecrement);
}
}
try {
// solve the linearized least squares problem
RealVector dX = new LUDecompositionImpl(a).getSolver().solve(b);
// update the estimated parameters
for (int i = 0; i < parameters.length; ++i) {
parameters[i].setEstimate(parameters[i].getEstimate() + dX.getEntry(i));
}
} catch (InvalidMatrixException e) {
throw new EstimationException("unable to solve: singular problem");
}
previous = cost;
updateResidualsAndCost();
} while ((getCostEvaluations() < 2)
|| (Math.abs(previous - cost) > (cost * steadyStateThreshold)
&& (Math.abs(cost) > convergence)));
}
/**
* Modifies this map through a single backpropagation iteration using the given error values on
* the output nodes.
*
* @param error
*/
public void train(List<Double> error, double learningRate) {
RealVector eOut = new ArrayRealVector(error.size());
for (int i : series(error.size())) eOut.setEntry(i, error.get(i));
// * gHidden: delta for the non-bias nodes of the hidden layer
gHidden.setSubVector(0, stateHidden.getSubVector(0, n)); // optimize
for (int i : Series.series(gHidden.getDimension()))
gHidden.setEntry(i, activation.derivative(gHidden.getEntry(i)));
eHiddenL = weights1.transpose().operate(eOut);
eHidden.setSubVector(0, eHiddenL.getSubVector(0, h));
for (int i : series(h)) eHidden.setEntry(i, eHidden.getEntry(i) * gHidden.getEntry(i));
weights1Delta = MatrixTools.outer(eOut, stateHidden);
weights1Delta = weights1Delta.scalarMultiply(-1.0 * learningRate); // optimize
weights0Delta = MatrixTools.outer(eHidden, stateIn);
weights0Delta = weights0Delta.scalarMultiply(-1.0 * learningRate);
weights0 = weights0.add(weights0Delta);
weights1 = weights1.add(weights1Delta);
}
