inheritance - constructor

**CornedBee** · 01-17-2008

All very valid questions. You won't find much code where vectors are annotated with their orientation. A vector is an n-tuple of numbers, period. Everything else is typically up to interpretation.

**hk_mp5kpdw** · 01-17-2008

Separate the class declarations (and only the declarations) into a header file, include a forward declaration for Vector1_2:

Code:

#ifndef FOO_H
#define FOO_H

class Vector1_2;

class Vector2_1
{
protected:
	double a, b;
public:
	Vector2_1();
	Vector2_1(double vec1, double vec2);
	void SetSpecificValue(double value, int pos);
	double GetSpecificValue(int pos) const;
	virtual void print();
	Vector1_2 transpose();
	Vector2_1 operator / (const double& C);
};



class Vector1_2:public Vector2_1
{
public:
	Vector1_2();
	//Vector1_2(double vec1, double vec2) {Vector2_1(vec1,vec2);}
	Vector1_2(double vec1, double vec2);
	//Vector1_2 operator * (const Matrix2_2&) const;
	void print();
	Vector2_1 transpose();
};
#endif

Then just include the header in your source file and write the code for the classes in that source file:

Code:

#include "foo.h"

Vector2_1::Vector2_1() {a=0; b=0;}

Vector2_1::Vector2_1(double vec1, double vec2):a(vec1), b(vec2) {}

void Vector2_1::print()
{
    std::cout<<std::endl<<"["<<a<<" "<<b<<"]T"<<std::endl;
}

Vector1_2 Vector2_1::transpose()
{
    return Vector1_2(a,b);
}  //the problem


Vector1_2::Vector1_2() {Vector2_1();}

Vector1_2::Vector1_2(double vec1, double vec2) {a=vec1; b=vec2;}

void Vector1_2::print()
{
    std::cout<<std::endl<<"["<<a<<" "<<b<<"]"<<std::endl;
}

Vector2_1 Vector1_2::transpose()
{
    return Vector2_1(a,b);
}

I've commented out some stuff related to things that you did not provide code for, e.g. the Matrix object, so I could test that it would at least compile under this new design. Something like the above should at least compile. Not commenting on any other issues which may or may not be present...

**misterowakka** · 01-17-2008

Originally Posted by anon

I'm sorry for not being very familiar with these things but...

What is the difference between Vector1_2 and Vector2_1 and why is one inherited from the other and not, perhaps, the other way round.? (Especially seeing that transpose should return the other for both.)

As I see the only difference is that one prints "T" in addition to the common things, and methods are rather randomly distributed between the two. Couldn't you put a bool in there to show that it is or isn't transposed? (Does it even matter if it is transposed? Wouldn't it be up to the user to decide what the vector represents?)

I had to do so because multiplication of vector to a matrix is by the proper dimensions. Therefore, in order to overload operators correctly, I assume that I have to have two different classes. The problem is that I can't find a way to create a function that will give me the other type each time.

I tried your offer, and built a base class vector2D that the two others vector2_1 and vector1_2 inherit from. It created a huge mass...And didn't solve the problem yet. Thank you anyway.

Anyone?

**anon** · 01-18-2008

I think rather than trying to make transpose return a different object with the exact same internal data, you should just have a single Vector class.

If you think that this data is so important you could add a member variable which shows how the vector should be multipltied with the matrix. Your transpose function might toggle this variable (return a copy with the value toggled).

Or you might give up the attempt to overload operator *. If it is a regular (member) function you can simply pass another parameter telling it how to multiply.

If I'm not mistaken vectors can be multiplied with several things and in several ways and if you used * for everything the user code may not become very clear at all.

**CornedBee** · 01-18-2008

In all vector * matrix multiplications, the vector is assumed to be vertical. Otherwise, the matrix would have to be one column wide, i.e. a vertical vector. Since that isn't useful, this case is disregarded.

**misterowakka** · 01-18-2008

Originally Posted by CornedBee

In all vector * matrix multiplications, the vector is assumed to be vertical. Otherwise, the matrix would have to be one column wide, i.e. a vertical vector. Since that isn't useful, this case is disregarded.

But what about vector*vector?
Either you'll get a scalar or a matrix n*n...

**misterowakka** · 01-18-2008

Another question. If I have some class, and I want to do

Class something(yada, yada, yada);
int i=5;
double b = i/something;

How do I say in the operator overloading that the input of the operator is supposed to be behind it?

Thanks.

**CornedBee** · 01-18-2008

Regarding your overloading question, make the operator a free function and you can order the argument however you want.

Regarding vector*vector, usually the distinction is made between dot product and cross product. One results in a scalar, the other in a vector. I've never heard of vector*vector -> matrix being actually used.
Which product to use for operator * differs between libraries.

**misterowakka** · 01-18-2008

Originally Posted by CornedBee

Regarding your overloading question, make the operator a free function and you can order the argument however you want.

Regarding vector*vector, usually the distinction is made between dot product and cross product. One results in a scalar, the other in a vector. I've never heard of vector*vector -> matrix being actually used.
Which product to use for operator * differs between libraries.

If you have two 2D vectors, one is a(2X1) the other b(1X2).

axb=matrix(2x2)
bxa=number...

My problem is as follows. Consider the following situation

b x matrix x a

b is a(transpose).
I have a class called vector2D that contains vector<double> and a bool for transpose or not. (As default it's a column).
The problem is, if I try to do

b x matrix x b.transpose()

inside the method transpose() I need to return a vector2D with its boolian for transposed converted to true. Because this flag is private I can't access it from instance b...

Header

Code:

class Vector2D
{
private:
	vector<double> vec;
	bool transposeState;
public:
	Vector2D();
	Vector2D(double vec1, double vec2);
	~Vector2D();
	void SetVectorItems(double vec1, double vec2);
	void SetSpecificValue(double value, int pos);
	double GetSpecificValue(int pos) const;
	virtual void print();
	Vector2D transpose(); //problem 1
	Vector2D operator / (const double& C);
	Matrix2_2 operator * (const Vector2D& vec); //problem 2 here
	double operator * (const Vector2D& vec); //and here
};

CPP

Code:

Vector2D Vector2D::transpose() 
{
	vector<double> vec2(*this);
	//now I need to change vec2 flag to true
	return vec2;
}

The second problem is that if I do multiplication between two vectors I either need to return matrix or double...How can I do that?! The compiler can't tell which of the two to call!

**misterowakka** · 01-18-2008

For whoever might need it. I had to create this matrix class for dealing with kalman filter. Since you helped me so much in this forum I soppuse that the final working code belongs to everyone.

Header file

Code:

#include <vector>
#include <iostream>
using namespace std;

class Matrix2_2;
class Vector1_2;

class Vector2_1
{
protected:
	double a, b;
public:
	Vector2_1();
	Vector2_1(double vec1, double vec2);
	void SetSpecificValue(double value, int pos);
	double GetSpecificValue(int pos) const;
	virtual void print();
	Vector1_2 transpose() const;
	Vector2_1 operator / (const double& C);
	Matrix2_2 operator * (const Vector1_2& vec);
	Vector2_1 operator * (const double& num);
	Vector2_1 operator + (const Vector2_1&) const;
};

class Vector1_2:public Vector2_1
{
public:
	Vector1_2():Vector2_1() {}
	Vector1_2(double vec1, double vec2):Vector2_1(vec1, vec2) {}
	Vector1_2 operator * (const Matrix2_2&) const;
	double operator * (const Vector2_1& vec) const;
	void print();
	Vector2_1 transpose() const;
};

class Matrix2_2
{
private:
	double M11, M12, M21, M22;
public:
	Matrix2_2();
	Matrix2_2(double UPLEFT, double UPRIGHT, double DOWNLEFT, double DOWNRIGHT):M11(UPLEFT), M12(UPRIGHT), M21(DOWNLEFT), M22(DOWNRIGHT) {}
	Matrix2_2(const Matrix2_2& M2);
	void SetAllValues(double UPLEFT, double UPRIGHT, double DOWNLEFT, double DOWNRIGHT);
	void SetSpecificValue(double value, int row, int column);
	double Getspecificvalue(int row, int column) const;
	Matrix2_2 Transpose() const;
	void print();

	Matrix2_2 operator * (const Matrix2_2&) const;
	Vector2_1 operator * (const Vector2_1&) const;
	Matrix2_2 operator + (const Matrix2_2&) const;
	Matrix2_2 operator - (const Matrix2_2&) const;
};

CPP

Code:

#include "windows.h"
#include "Matrix.h"


/////////////////VECTOR2_1//////////////////

Vector2_1::Vector2_1() 
{a=0; b=0;}

Vector2_1::Vector2_1(double vec1, double vec2)
{a=vec1; b=vec2;}

void Vector2_1::SetSpecificValue(double value, int pos)
{
	if(pos==1)
		a=value;
	else if(pos==2)
		b=value;
	else
		MessageBox(NULL, "wrong way of dealing with Vector2_1 class", "Located in: void Vector2_1::SetSpecificValue", NULL);
}

double Vector2_1::GetSpecificValue(int pos) const
{
	if(pos==1)
		return a;
	else if(pos==2)
		return b;
	else
		MessageBox(NULL, "wrong way of dealing with Vector2_1 class", "Located in: void Vector2_1::SetSpecificValue", NULL);
	return -99999; //error
}


void Vector2_1::print() 
{cout<<endl<<"["<<a<<" "<<b<<"]T"<<endl;}

Vector1_2 Vector2_1::transpose() const
{
	return Vector1_2(a,b);
}

Vector2_1 Vector2_1::operator / (const double& C)
{
	return Vector2_1(a/C, b/C);
}

Matrix2_2 Vector2_1::operator * (const Vector1_2& vec)
{
	return Matrix2_2(a*vec.GetSpecificValue(1), a*vec.GetSpecificValue(2), b*vec.GetSpecificValue(1), b*vec.GetSpecificValue(2));
}


Vector2_1 Vector2_1::operator * (const double& num)
{
	return Vector2_1(a*num, b*num);
}


Vector2_1 Vector2_1::operator + (const Vector2_1& vec2) const
{
	return Vector2_1(a+vec2.GetSpecificValue(1), b+vec2.GetSpecificValue(2));
}


/////////////////VECTOR1_2//////////////////


Vector1_2 Vector1_2::operator * (const Matrix2_2& M) const
{
	Vector1_2 vec2;
	vec2.SetSpecificValue(a*M.Getspecificvalue(1,1)+b*M.Getspecificvalue(2,1),1);
	vec2.SetSpecificValue(a*M.Getspecificvalue(1,2)+b*M.Getspecificvalue(2,2),2);
	return vec2;
}


double Vector1_2::operator * (const Vector2_1& vec) const
{
	return a*vec.GetSpecificValue(1)+b*vec.GetSpecificValue(2);
}


void Vector1_2::print() 
{cout<<endl<<"["<<a<<" "<<b<<"]"<<endl;}

Vector2_1 Vector1_2::transpose() const
{
	return Vector2_1(a,b);
}







/////////////////MATRIX2_2//////////////////

Matrix2_2::Matrix2_2()
{
	M11=0; M12=0; M21=0; M22=0;
}

Matrix2_2::Matrix2_2(const Matrix2_2& M2) 
{
	*this=M2;
}

void Matrix2_2::SetAllValues(double UPLEFT, double UPRIGHT, double DOWNLEFT, double DOWNRIGHT)
{
	M11=UPLEFT; M12=UPRIGHT; M21=DOWNLEFT; M22=DOWNRIGHT;
}

void Matrix2_2::SetSpecificValue(double value, int row, int column)
{
	if(row==1)
		if(column==1)
			M11=value;
		else if(column==2)
			M12=value;
		else
			MessageBox(NULL, "wrong way of dealing with Matrix2_2 class", "Located in: void Matrix2_2::SetSpecificValue", NULL);
	else if(row==2)
		if(column==1)
			M21=value;
		else if(column==2)
			M22=value;
		else
			MessageBox(NULL, "wrong way of dealing with Matrix2_2 class", "Located in: void Matrix2_2::SetSpecificValue", NULL);
	else
		MessageBox(NULL, "wrong way of dealing with Matrix2_2 class", "Located in: void Matrix2_2::SetSpecificValue", NULL);
}


double Matrix2_2::Getspecificvalue(int row, int column) const
{
	if(row==1)
		if(column==1)
			return M11;
		else if(column==2)
			return M12;
		else
			MessageBox(NULL, "wrong way of dealing with Matrix2_2 class", "Located in: void Matrix2_2::SetSpecificValue", NULL);
	else if(row==2)
		if(column==1)
			return M21;
		else if(column==2)
			return M22;
		else
			MessageBox(NULL, "wrong way of dealing with Matrix2_2 class", "Located in: void Matrix2_2::SetSpecificValue", NULL);
	else
		MessageBox(NULL, "wrong way of dealing with Matrix2_2 class", "Located in: void Matrix2_2::SetSpecificValue", NULL);
	return -999999;//for error
}


Matrix2_2 Matrix2_2::operator + (const Matrix2_2& M2) const
{
	Matrix2_2 M3;
	M3.SetSpecificValue(M11+M2.Getspecificvalue(1,1),1,1);
	M3.SetSpecificValue(M12+M2.Getspecificvalue(1,2),1,2);
	M3.SetSpecificValue(M21+M2.Getspecificvalue(2,1),2,1);
	M3.SetSpecificValue(M22+M2.Getspecificvalue(2,2),2,2);
	return M3;
}


Matrix2_2 Matrix2_2::operator - (const Matrix2_2& M2) const
{
	Matrix2_2 M3;
	M3.SetSpecificValue(M11-M2.Getspecificvalue(1,1),1,1);
	M3.SetSpecificValue(M12-M2.Getspecificvalue(1,2),1,2);
	M3.SetSpecificValue(M21-M2.Getspecificvalue(2,1),2,1);
	M3.SetSpecificValue(M22-M2.Getspecificvalue(2,2),2,2);
	return M3;
}

Matrix2_2 Matrix2_2::operator * (const Matrix2_2& M2) const
{
	Matrix2_2 mat;
	mat.SetSpecificValue(M11*M2.Getspecificvalue(1,1)+M12*M2.Getspecificvalue(2,1), 1,1);
	mat.SetSpecificValue(M11*M2.Getspecificvalue(1,2)+M12*M2.Getspecificvalue(2,2), 1,2);
	mat.SetSpecificValue(M21*M2.Getspecificvalue(1,1)+M22*M2.Getspecificvalue(2,1), 2,1);
	mat.SetSpecificValue(M21*M2.Getspecificvalue(1,2)+M22*M2.Getspecificvalue(2,2), 2,2);
	return mat;
}

Vector2_1 Matrix2_2::operator * (const Vector2_1& vec) const
{
	Vector2_1 vec2;
	vec2.SetSpecificValue(M11*vec.GetSpecificValue(1)+M12*vec.GetSpecificValue(2), 1);
	vec2.SetSpecificValue(M21*vec.GetSpecificValue(1)+M22*vec.GetSpecificValue(2), 2);
	return vec2;
}



void Matrix2_2::print()
{
	cout<<endl<<M11<<" "<<M12<<endl<<M21<<" "<<M22<<endl<<endl;
}

Matrix2_2 Matrix2_2::Transpose() const
{
	return Matrix2_2(M11,M21,M12,M22);
}

Kalman filter implementation using these classes

Code:

#include "Matrix.h"
#include <iostream>
using namespace std;

void main()
{

	double vel[8]={1,2,3,4,5,6,7,8};
	for (int counter=0; counter<8; counter++)
	{
		static Vector2_1 Xhat_previous(2,2);
		const Matrix2_2 A(1,1,0,1);
		
		Vector2_1 XhatPre=A*Xhat_previous;
		//XhatPre.print();
		
		static Matrix2_2 P_previous(1,2,3,4);
		const Matrix2_2 Q(5,6,7,8);
		const double R=0.1666666;
		
		Matrix2_2 Ppre=A*P_previous*A.Transpose()+Q;
		//Ppre.print();
		
		const Vector1_2 H(0,1);
		
		Vector2_1 K = (Ppre*H.transpose())/((H*Ppre*H.transpose())+R);
		//K.print();

		Vector2_1 Xhat=XhatPre+K*(vel[counter]-H*XhatPre);
		Xhat.print();

		Xhat = Xhat_previous;

		const Matrix2_2 I(1,0,0,1);
		Matrix2_2 P=(I-K*H)*Ppre;
		P.print();

		P = P_previous;
	}
	//return Xhat.GetSpecificValue(1); //this is the calculated velocity from kalman filter. Hhat[2] is the acceleration
}

**Elysia** · 01-18-2008

Now that I see you're including Windows.h, you could use precompiled headers.
Put windows.h in stdafx.h and enable PCH in project options under compiler (create PCH first, then use PCH).
It will speed up compile time significantly.

And I see you're using void main too >_<
Undefined! Use int main!

**robwhit** · 01-18-2008

http://faq.cprogramming.com/cgi-bin/...&id=1043284376

Thread: inheritance - constructor

Thread Tools

Search Thread

Display

Kalman Filter

Similar Threads

Can you call a constructor within a destructor, or vise vesa?

C++ have a constructor call another constructor

How do I override constructor in inheritance?

Constructor inheritance

Constructor inheritance