NullPointerException

[Samuel Andermatt]

Hi
I have done some C++ during my studies and today I decided to learn a little about java. by making a simple Molecular dynamics simulation.
In one line I get a NullpointerException and I do not exactly understand what I do wrong.

The error is in the line (in ParticleSystem.java)
this.P[ i * Ny * Nz + j * Nz + k ] = new Particle( pos , v , R.getRN() );

In case the problem lies somewhere else I copy the rest of the code. If you have a hint about a general missunderstanding how I am supposed to use java I would also appreciate it.

MolecularD.java

public class MolecularD {
	public static void main(String[] Args) {
		//Parameters
		int Nx = 10; //particles in x direction
		int Ny = 10; //particles in y direction
		int Nz = 10; //particles in z direction
		int N = Nx * Ny * Nz; //Total number of particles
		float dim[] = new float[3]; //System dimension
		dim[0] = 100; //length of the box in x direction
		dim[1] = 100; //length of the box in y direction
		dim[2] = 100; //length of the box in z direction
		float t = 1; //Simulation length
		float dt = 0.0001f; //Time step
		ParticleSystem PS = new ParticleSystem( Nx , Ny , Nz , dim );
		for( float i = 0 ; i < t ; i += dt) {
			PS.evolve( dt );
		}
	}
}

ParticleSystem.java

public class ParticleSystem {
	public Particle P[];
	private int N; //Number of particles
	ParticleSystem( int Nx, int Ny, int Nz , float dim[]) {
		this.N = Nx * Ny * Nz;
		Particle P[] = new Particle[ Nx * Ny * Nz ];
		float pos[] = new float[3];
		float v[] = new float[3];
		RNG R = new RNG(1238); //generate seeded RNG
		for( int i = 0 ; i < Nx ; ++i )  {
			for( int j = 0 ; j < Ny ; ++j )  {
				for( int k = 0 ; k < Nz ; ++k )  {
					pos[0] = ( i + 0.5f ) * dim[0] / (Nx + 1f);
					pos[1] = ( j + 0.5f ) * dim[1] / (Ny + 1f);
					pos[2] = ( k + 0.5f ) * dim[2] / (Nz + 1f);
					v[0] = 0;
					v[1] = 0;
					v[2] = 0;
					this.P[ i * Ny * Nz + j * Nz + k ] = new Particle( pos , v , R.getRN() );
				}
			}
		}
	}
	void evolve( float dt ) { //Let the simulation run, according to the velocity-verlet algorithm
		//update positions
		for ( int i = 0 ; i < N ; ++i) {
			this.P[i].p[0] += dt * this.P[i].p[0] + dt * dt * this.P[i].anew[0] / 2;
			this.P[i].p[1] += dt * this.P[i].p[1] + dt * dt * this.P[i].anew[1] / 2;
			this.P[i].p[2] += dt * this.P[i].p[2] + dt * dt * this.P[i].anew[2] / 2;
		}
		//calculate the forces
		for ( int i = 0 ; i < N ; ++i) { //set new forces to zero and old forces to new forces
			this.P[i].aold[0] = this.P[i].anew[0];
			this.P[i].aold[1] = this.P[i].anew[1];
			this.P[i].aold[2] = this.P[i].anew[2];
			this.P[i].anew[0] = 0;
			this.P[i].anew[1] = 0;
			this.P[i].anew[2] = 0;
		}
		for ( int i = 0 ; i < N ; ++i) {
			for ( int j = i+1 ; j < N ; ++j) {
				this.P[i].interact(this.P[j]);
			}
		}
		//update velocities
		for ( int i = 0 ; i < N ; ++i) {
			this.P[i].v[0] += ( this.P[i].anew[0] + this.P[i].aold[0]) / ( 2 * dt);
			this.P[i].v[1] += ( this.P[i].anew[1] + this.P[i].aold[1]) / ( 2 * dt);
			this.P[i].v[2] += ( this.P[i].anew[2] + this.P[i].aold[2]) / ( 2 * dt);
		}
	}
}

Particle.java

public class Particle {
	public float[] p = new float[3]; //position
	public float[] v = new float[3]; //velocity
	public float[] aold = new float[3]; //force
	public float[] anew = new float[3]; //force
	public float m; //mass
	Particle( float pos[], float vel[], float mass ) {
		this.p = pos;
		this.v = vel;
		this.m = mass;
		this.aold[0] = 0;
		this.aold[1] = 0;
		this.aold[2] = 0;
		this.anew[0] = 0;
		this.anew[1] = 0;
		this.anew[2] = 0;
	}
	void interact( Particle P2 ) {
		float dx,dy,dz,d2;
		float F; //Force
		dx = this.p[0] - P2.p[0];
		dy = this.p[1] - P2.p[1];
		dz = this.p[2] - P2.p[2];
		d2 = dx * dx + dy * dy + dz * dz ; //squared distance
		F = 1/(d2*d2*d2) - 1/(d2*d2*d2*d2*d2*d2);
		this.anew[0] -= F * dx / this.m;
		this.anew[1] -= F * dy / this.m;
		this.anew[2] -= F * dz / this.m;
		P2.anew[0] += F * dx / this.m;
		P2.anew[1] += F * dy / this.m;
		P2.anew[2] += F * dz / this.m;
	}
}

RNG.java

public class RNG { //Not the best RNG, but does not have to be for this use
	private long Seed;
	private long c = 22695477;
	private long m = 2147483647;
	RNG ( int S ) {
		this.Seed = S;
	}
	float getRN() {
		this.Seed = (this.Seed * this.c) % this.m;
		return (float)this.Seed/this.m;
	}
}

Edit: Ok, Sorry the format gets somehow messed up.

[Added code tags - see UseCodeTags for details]

[Matthew Brown]

Hi Samuel. Welcome to the Ranch!

In that line:

this.P[ i * Ny * Nz + j * Nz + k ] = new Particle( pos , v , R.getRN() ); 

There are two things that could cause a NullPointerException. If either this.P or R haven't been initialised you'll get the error. So that's where to start looking.

Starting with R...that looks to have been initialised correctly on line 9 of ParticleSystem.java. So that's OK.

What about this.P? What's happened there is actually a common error. On line 6 you're declaring and initialising a local variable P. But this.P is an instance variable. They've got the same name, but they are not the same thing - the local variable "shadows" the instance variable. And your instance variable is never initialised, as far as I can see.

[Jesper de Jong]

Welcome to the Ranch.

The problem is that in that line, P is null.

The real problem is in line 6 of ParticleSystem.java, where you declare a new local variable P that hides the member variable P that was declared on line 2. So in line 6 you are not initializing the member variable P; you're declaring a new, local variable P that's initialized. The member variable P remains on its default value, null.

Change line 6 to this:

P = new Particle[Nx * Ny * Nz];

[Samuel Andermatt]

Thank you both, It runs now.

[Winston Gutkowski]

Samuel Andermatt wrote:The error is in the line (in ParticleSystem.java)
this.P[ i * Ny * Nz + j * Nz + k ] = new Particle( pos , v , R.getRN() );

Right. Well, first, do you understand Java operator precedence? The above line is the equivalent of:
this.P[ (i * Ny * Nz) + (j * Nz) + k ] = new Particle( pos , v , R.getRN() );
Is that what you want?

Other than that, I can't see any other cause off the top of my head; and if it is the problem, I'm surprised it didn't throw ArrayIndexOutOfBoundsException.

Second: Your RNG class is redundant. Java has a class called java.util.Random, and it is a good RNG - and seedable.

Winston

[Edit] Doh-h! The others have seen what I stupidly missed. However, I still reckon that you should read my post, because the other things may be relevant.

[Winston Gutkowski]

Samuel Andermatt wrote:Thank you both, It runs now.

OK, but you're still giving yourself a lot of headaches by ploughing so much into a single statement. Break things up a bit.

Winston

[Samuel Andermatt]

Winston Gutkowski wrote:

Samuel Andermatt wrote:The error is in the line (in ParticleSystem.java)
this.P[ i * Ny * Nz + j * Nz + k ] = new Particle( pos , v , R.getRN() );

Right. Well, first, do you understand Java operator precedence? The above line is the equivalent of:
this.P[ (i * Ny * Nz) + (j * Nz) + k ] = new Particle( pos , v , R.getRN() );
Is that what you want?

Other than that, I can't see any other cause off the top of my head; and if it is the problem, I'm surprised it didn't throw ArrayIndexOutOfBoundsException.

Second: Your RNG class is redundant. Java has a class called java.util.Random, and it is a good RNG - and seedable.

Winston

[Edit] Doh-h! The others have seen what I stupidly missed. However, I still reckon that you should read my post, because the other things may be relevant.

About precedence. Yes the statement you wrote is what I wanted. Mathematically I chose a long vector instead of making a 3 dimensional matrix.
Thanks for the help with the RNG.

[Winston Gutkowski]

Samuel Andermatt wrote:Mathematically I chose a long vector instead of making a 3 dimensional matrix.

Hmmm. Not necessarily the best, since Java "matrices" are arrays of arrays. By doing that, you've forced the JVM into allocating contiguous memory, which may cause spikes and/or delays that you couldn't foresee. You also may suffer from 'locality of reference' issues if this matrix can get very large.

Java is NOT C++.

Winston