/**
* Initiates the fit with function constructed weights and a JAMA matrix. N is the number of
* y-data points, K is the dimension of the fit function and M is the number of fit parameters.
* Call <code>this.fit()</code> to start the actual fitting.
*
* @param function The model function to be fitted. Input parameter sizes K and M.
* @param parameters The initial guess for the fit parameters, length M.
* @param dataPoints The data points in two dimensional array where each array, dataPoints[i],
* contains one y-value followed by the corresponding x-array values. I.e., the arrays should
* look like this:
* <p>dataPoints[0] = y0 x00 x01 x02 ... x0[K-1]<br>
* dataPoints[1] = y1 x10 x11 x12 ... x1[K-1]<br>
* . ..<br>
* dataPoints[N] = yN xN0 xN1 xN2 ... x[N-1][K-1]
*/
public LMA(LMAMultiDimFunction function, float[] parameters, float[][] dataPoints) {
this(
function,
ArrayConverter.asDoubleArray(parameters),
ArrayConverter.asDoubleArray(dataPoints),
function.constructWeights(ArrayConverter.asDoubleArray(dataPoints)),
new JAMAMatrix(parameters.length, parameters.length));
}
/**
* One dimensional convenience constructor for LMAFunction. You can also implement the same
* function using LMAMultiDimFunction.
*
* <p>Initiates the fit with function constructed weights and a JAMA matrix. N is the number of
* data points, M is the number of fit parameters. Call <code>fit()</code> to start the actual
* fitting.
*
* @param function The model function to be fitted. Must be able to take M input parameters.
* @param parameters The initial guess for the fit parameters, length M.
* @param dataPoints The data points in an * array, <code>float[0 = x, 1 = y][point index]</code>.
* Size must * be <code>float[2][N]</code>.
* @param weights The weights, normally given * as: <code>weights[i] = 1 / sigma_i^2</code>. If
* you have a bad data point, set its weight to zero. If the given array is null, a new array
* is created with all elements set to 1.
*/
public LMA(
final LMAFunction function, float[] parameters, float[][] dataPoints, float[] weights) {
this(
function,
ArrayConverter.asDoubleArray(parameters),
ArrayConverter.asDoubleArray(dataPoints),
ArrayConverter.asDoubleArray(weights));
}
/**
* Initiates the fit. N is the number of y-data points, K is the dimension of the fit function and
* M is the number of fit parameters. Call <code>this.fit()</code> to start the actual fitting.
*
* @param function The model function to be fitted. Input parameter sizes K and M.
* @param parameters The initial guess for the fit parameters, length M.
* @param dataPoints The data points in two dimensional array where each array, dataPoints[i],
* contains one y-value followed by the corresponding x-array values. I.e., the arrays should
* look like this:
* <p>dataPoints[0] = y0 x00 x01 x02 ... x0[K-1]<br>
* dataPoints[1] = y1 x10 x11 x12 ... x1[K-1]<br>
* . ..<br>
* dataPoints[N] = yN xN0 xN1 xN2 ... x[N-1][K-1]
* <p>
* @param weights The weights, normally given * as: <code>weights[i] = 1 / sigma_i^2</code>. If
* you have a bad data point, set its weight to zero. If the given array is null, a new array
* is created with all elements set to 1.
* @param alpha An LMAMatrix instance. Must be initiated to (M x M) size.
*/
public LMA(
LMAMultiDimFunction function,
float[] parameters,
float[][] dataPoints,
float[] weights,
LMAMatrix alpha) {
this(
function,
ArrayConverter.asDoubleArray(parameters),
ArrayConverter.asDoubleArray(dataPoints),
ArrayConverter.asDoubleArray(weights),
alpha);
}
/**
* Initiates the fit with function constructed weights and a JAMA matrix. N is the number of
* y-data points, K is the dimension of the fit function and M is the number of fit parameters.
* Call <code>this.fit()</code> to start the actual fitting.
*
* @param function The model function to be fitted. Input parameter sizes K and M.
* @param parameters The initial guess for the fit parameters, length M.
* @param yDataPoints The y-data points in an array.
* @param xDataPoints The x-data points for each y data point, double[y-index][x-index]
*/
public LMA(
LMAMultiDimFunction function,
double[] parameters,
float[] yDataPoints,
float[][] xDataPoints) {
this(
function,
parameters,
ArrayConverter.asDoubleArray(yDataPoints),
ArrayConverter.asDoubleArray(xDataPoints),
function.constructWeights(
ArrayConverter.combineMultiDimDataPoints(yDataPoints, xDataPoints)),
new JAMAMatrix(parameters.length, parameters.length));
}
/**
* Initiates the fit. N is the number of y-data points, K is the dimension of the fit function and
* M is the number of fit parameters. Call <code>this.fit()</code> to start the actual fitting.
*
* @param function The model function to be fitted. Input parameter sizes K and M.
* @param parameters The initial guess for the fit parameters, length M.
* @param dataPoints The data points in two dimensional array where each array, dataPoints[i],
* contains one y-value followed by the corresponding x-array values. I.e., the arrays should
* look like this:
* <p>dataPoints[0] = y0 x00 x01 x02 ... x0[K-1]<br>
* dataPoints[1] = y1 x10 x11 x12 ... x1[K-1]<br>
* . ..<br>
* dataPoints[N] = yN xN0 xN1 xN2 ... x[N-1][K-1]
* <p>
* @param weights The weights, normally given * as: <code>weights[i] = 1 / sigma_i^2</code>. If
* you have a bad data point, set its weight to zero. If the given array is null, a new array
* is created with all elements set to 1.
* @param alpha An LMAMatrix instance. Must be initiated to (M x M) size.
*/
public LMA(
LMAMultiDimFunction function,
double[] parameters,
double[][] dataPoints,
double[] weights,
LMAMatrix alpha) {
SeparatedData s = ArrayConverter.separateMultiDimDataToXY(dataPoints);
this.yDataPoints = s.yDataPoints;
this.xDataPoints = s.xDataPoints;
init(function, parameters, yDataPoints, xDataPoints, weights, alpha);
}
/**
* One dimensional convenience constructor for LMAFunction. You can also implement the same
* function using LMAMultiDimFunction.
*
* <p>Initiates the fit with function constructed weights and a JAMA matrix. N is the number of
* data points, M is the number of fit parameters. Call <code>fit()</code> to start the actual
* fitting.
*
* @param function The model function to be fitted. Must be able to take M input parameters.
* @param parameters The initial guess for the fit parameters, length M.
* @param dataPoints The data points in an * array, <code>double[0 = x, 1 = y][point index]</code>
* . Size must * be <code>double[2][N]</code>.
*/
public LMA(
final LMAFunction function, double[] parameters, double[][] dataPoints, double[] weights) {
this(
// convert LMAFunction to LMAMultiDimFunction
new LMAMultiDimFunction() {
private LMAFunction f = function;
@Override
public double getPartialDerivate(double[] x, double[] a, int parameterIndex) {
return f.getPartialDerivate(x[0], a, parameterIndex);
}
@Override
public double getY(double[] x, double[] a) {
return f.getY(x[0], a);
}
},
parameters,
dataPoints[1], // y-data
ArrayConverter.transpose(dataPoints[0]), // x-data
weights,
new JAMAMatrix(parameters.length, parameters.length));
}
/**
* Computes Chinese Reminder Theorem: x == congs[i] mod moduli[i]
*
* @param congs
* @param moduli
* @return
*/
public static BigInteger CRT(List<BigInteger> congs, List<BigInteger> moduli) {
BigInteger[] cs = ArrayConverter.convertListToArray(congs);
BigInteger[] ms = ArrayConverter.convertListToArray(moduli);
return CRT(cs, ms);
}
