Reputation:
As far as I understand, the default type or argument passing in c#
is by value. Therefore no statement is required. But when I try the run following code, my A
matrix in Main is being modified by the operations done to dMatrixU
in the Factorize()
method of class Decomposition
. I'm sure the problem is in the constructor of the Decomposition
when I just assing A
to dMatrixU
, the reference of A
is being assigned instead of the values. Therefore my question on how to avoid this, all I have found is how to pass the arguments by reference. Again, as I understand no modifier is needed for passing the argument by value. Where am I wrong?
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text.RegularExpressions;
using LinearEquations;
namespace Rextester
{
public class Program
{
public static void Main(string[] args)
{
double[,] A = new double[,]
{ { 1, 1, 1 } ,
{ 4, 3, -1 } ,
{ 3, 5, 3 } };
double[] B = new double[] {1,6,4};
Decomposition lu = new Decomposition(A,B);
lu.Factorize();
PrintMatrix(A,"A:");
PrintVector(B,"B:");
PrintMatrix(lu.L,"L:");
PrintMatrix(lu.U,"U:");
PrintVector(lu.D,"D:");
}
public static void PrintMatrix(double[,] M, String Title = "Matrix: ")
{
Console.WriteLine(Title);
for(int i = 0; i<M.GetLength(0); i++)
{
for(int j = 0; j<M.GetLength(1);j++)
{
Console.Write(M[i,j]+"\t");
}
Console.Write("\n");
}
Console.Write("\n");
}
public static void PrintVector(double[] V, String Title = "Vector: ",bool AsRow = true)
{
String str = (AsRow)? "\t" : "\n";
Console.WriteLine(Title);
for(int i = 0; i<V.GetLength(0); i++)
{
Console.Write(V[i]+str);
}
Console.WriteLine("\n");
}
}
}
namespace LinearEquations
{
public class Decomposition
{
// Fields
private double[,] dMatrixA; // Parameter in A*X=B
private double[] dVectorB; // Parameter in A*X=B
private double[] dVectorX; // Result wanted in A*X=B
private double[,] dMatrixU; // A splits into L and U
private double[,] dMatrixL; // L is used to calculate D in L*D=B
private double [] dVectorD; // D is used to calculate X in U*X=D
// Properties
public double[,] A
{
get { return dMatrixA; }
set { dMatrixA = value; }
}
public double[] B
{
get { return dVectorB; }
set { dVectorB = value; }
}
public double[] X
{
get { return dVectorX; }
set { dVectorX = value; }
}
public double[,] L
{
get { return dMatrixL; }
set { dMatrixL = value; }
}
public double[,] U
{
get { return dMatrixU; }
set { dMatrixU = value; }
}
public double[] D
{
get { return dVectorD; }
set { dVectorD = value; }
}
// Constructor
public Decomposition(double[,] A, double[] B)
{
dMatrixA = A;
dVectorB = B;
dVectorX = new double[B.Length];
dMatrixU = A;
dMatrixL = new double[A.GetLength(0),A.GetLength(1)];
dVectorD = new double[B.Length];
}
// Split A into L and U
public void Factorize()
{
// Iterate per each row
for(int i = 0; i<dMatrixU.GetLength(0); i++)
{
// For all the rows make element i equals 0
for(int j = i+1; j<dMatrixU.GetLength(0);j++)
{
// Factor that assures substraction makes 0
dMatrixL[1,1] = dMatrixU[j,i] / dMatrixU[i,i];
// Iterate per each column
for(int k = 0; k<dMatrixU.GetLength(1);k++)
{
dMatrixU[j,k] = dMatrixU[j,k] - dMatrixU[i,k]*dMatrixL[1,1];
}
}
}
}
}
}
Upvotes: 0
Views: 1725
Reputation: 9804
As far is i understand, the default type or argument passing in c# is by value.
Unfortunately it is a bit more complicated and also has some execptions:
Reference types like Decomposition
you hand in by making a copy of the reference. Unfortunately that means both still reference the same instance in memory. So despite a copy operation, it is call-by-Reference.
With value types like Int
or double
and their aliases, usually a copy is made. I do not know of any case where it does not, but I was wrong on those things before. So they are call by value.
Finally String
and a few other reference types are inmutable by design. That has the advantage that they behave kinda like value types in this area. You hand in a Reference, but the instance itself can not be changed. The code can only create a new instance in memory with a different value. So despite handing over literal references, it kinda works like call by value.
Your specific case
Arrays are very explicitly Reference types. Handing them into a function without side effects, requires proper cloning. If it is a array of reference types, the cloning must be deep.
In your case you have arrays of value types. If you want to avoid call-by-reference side effects, you those arrays must be cloned. However as double is a value type, this cloning can be shallow. No need for a deep clone.
Unlike Java there is not a dedicated Clone() Method. And I am not sure why exactly. However you can often use one Collection to initialize another through the constructor. Or they even have a function like Array.Copy()
, as TheBatman pointed out.
Upvotes: 2