Java: Random Class & Simulating Particles in a Box

[Feldoh]

I'm trying to program a java simulation of some particles in a box. Anyways I want to initialize the particles with random velocities such that on average they're distributed around some velocity.

To do this I was thinking about using the nextGaussian() method.

Anyways I was initializing my particles with something like (in a simplified version):

Random r = new Random();
double avg = 10;
Particle a = new Particle(xpos,ypos, r.nextGaussian()+avg,r.nextGaussian()+avg);

Where particle takes in 4 doubles: xpos,ypos,xvel,yvel.

Anyways I'm worried about the way java stores the variables. I'm worried that my particle class will store the addr to r.nextGaussian rather than just a number. This would be bad since then everytime I would call the xvel or yvel the number would change.

Is that assessment correct and does anyone know of a way to fix it if this is correct?

 

[Mark44]

You didn't show the definition of the Particle class, but presumably it has four member variables of type double that the constructor sets. After a given Particle instance is constructed, the xvel and yvel values won't change unless some member function changes them.

 

[chiro]

I'm trying to program a java simulation of some particles in a box. Anyways I want to initialize the particles with random velocities such that on average they're distributed around some velocity.

To do this I was thinking about using the nextGaussian() method.

Anyways I was initializing my particles with something like (in a simplified version):

Random r = new Random();
double avg = 10;
Particle a = new Particle(xpos,ypos, r.nextGaussian()+avg,r.nextGaussian()+avg);

Where particle takes in 4 doubles: xpos,ypos,xvel,yvel.

Anyways I'm worried about the way java stores the variables. I'm worried that my particle class will store the addr to r.nextGaussian rather than just a number. This would be bad since then everytime I would call the xvel or yvel the number would change.

Is that assessment correct and does anyone know of a way to fix it if this is correct?

You could specify a variable on the stack and just assign r.nextGaussian() to that variable.

Typically when you want to force a register or some word to take the value instead of the address the function has to pass the arguments BYVAL (by value) instead of BYREF (by reference).

In BASIC languages you use BYVAL or BYREF. In C/C++ you simply use the variable (eg int value) or you use the reference symbol (int &value) for passing the address. I'm not sure what you do in java but there's probably something like the syntax in C++ and if i recall correctly Java doesn't allow you to use pointers which lends me to think that there is definitely a reference syntax thing in there somewhere (I haven't used Java in over 5 years)

 

[chiro]

Also I should mention that if the Particle class requires a reference then your code would not compile because you have not supplied a correct variable.

If the code compiles it should most likely be constructing the variable on the stack and then passing by value.

 

[LudusRex]

I am familiar with using Java to simulate particles in a box. I can understand your concern about how Java stores variables and how it may affect the simulation.

Firstly, let me assure you that Java stores variables as values, not addresses. So in your code, the xvel and yvel values will be stored as numbers, not addresses.

However, there is a potential issue with your approach. Using the nextGaussian() method, you are generating random values that follow a normal distribution with a mean of 0 and a standard deviation of 1. By adding your desired average velocity (avg) to this value, you are essentially shifting the distribution of velocities. This may result in some particles having velocities that are significantly higher or lower than your desired average.

To ensure that your particles have velocities distributed around your desired average, you could use the nextDouble() method instead. This will generate random values between 0.0 and 1.0, which you can then scale to fit your desired velocity range. For example, if you want velocities between -10 and 10, you could do something like this:

Random r = new Random();
double avg = 10;
Particle a = new Particle(xpos, ypos, (r.nextDouble() * 20) - 10, (r.nextDouble() * 20) - 10);

This approach will generate velocities that are more evenly distributed around your desired average.

Additionally, you can also use the setSeed() method to set a specific seed for your random number generator. This will ensure that your simulation is reproducible and consistent each time it is run.

In conclusion, while your assessment of how Java stores variables is not entirely correct, there are some improvements that can be made to your approach to ensure a more accurate simulation. I hope this helps and good luck with your simulation!