- 6

- 0

## Main Question or Discussion Point

Hi,

I am doing simulations on a particle in some potential together with a fluctuating force and friction. To do that, I use the Langevin equation with the fluctuating force being a random number from a normal distribution with a temperature-dependent variance. I use a Verlet-algorithm for the calculation of position, velocity and acceleration (x,v and a) at each timestep.

My problem is how I can implement a high potential barrier - basically a wall. What I tried so far was to change the sign of x and v once it happens that x<0 (I have the barrier at x=0, and a linear potential at x>0).

But this gives unrealistic results, the particle goes way to far away from the barrier at very low temperatures - it looks like the particle in some way gets quite much energy when it hits the wall.

Does anyone have some suggestion on how to implement such a barrier? Or help me explain why my current method fails?

I can mention it probably isn't problems with the Verlet-algorithm itself or something else, since it yields very nice results if I use a harmonic potential.

Let me know if you need more information :)

I am doing simulations on a particle in some potential together with a fluctuating force and friction. To do that, I use the Langevin equation with the fluctuating force being a random number from a normal distribution with a temperature-dependent variance. I use a Verlet-algorithm for the calculation of position, velocity and acceleration (x,v and a) at each timestep.

My problem is how I can implement a high potential barrier - basically a wall. What I tried so far was to change the sign of x and v once it happens that x<0 (I have the barrier at x=0, and a linear potential at x>0).

But this gives unrealistic results, the particle goes way to far away from the barrier at very low temperatures - it looks like the particle in some way gets quite much energy when it hits the wall.

Does anyone have some suggestion on how to implement such a barrier? Or help me explain why my current method fails?

I can mention it probably isn't problems with the Verlet-algorithm itself or something else, since it yields very nice results if I use a harmonic potential.

Let me know if you need more information :)