Simulation of Bernoulli TASEP with step initial condition. Particles at sites {-r+1,...,0} each flip a biased coin and (if heads) attempt to jump right by one site; exclusion prevents two particles from occupying the same site. Two update rules: parallel (snapshot-based, simultaneous) and sequential (right-to-left cascading, allowing chains of jumps). Outputs are the averaged empirical density profile as a function of xi = x/T, and the fraction of active particles A_n/n vs time.
Definitions, stationary currents, and limit shapes
Setup. State space $\{0,1\}^{\mathbb{Z}}$; write $\eta_t(x) \in \{0,1\}$ for the occupancy of site $x \in \mathbb{Z}$ at time $t \in \mathbb{Z}_{\ge 0}$. Fix $r \in \mathbb{Z}_{\ge 1}$ and $p \in (0,1]$. Step initial condition: $\eta_0(x) = \mathbf{1}_{-r < x \le 0}$. Let $(\xi_{t,x})_{t \ge 0,\, x \in \mathbb{Z}}$ be i.i.d. $\mathrm{Bernoulli}(p)$.
Definition 1 (parallel update). $\eta_{t+1}$ is obtained from $\eta_t$ by moving, simultaneously, every particle at $x$ with $\xi_{t,x} = 1$ and $\eta_t(x+1) = 0$.
Definition 2 (right-to-left sequential update). $\eta_{t+1}$ is obtained from $\eta_t$ by processing sites in strict decreasing order of $x$: apply $(\eta(x), \eta(x+1)) \mapsto (0,1)$ whenever the current configuration satisfies $\eta(x) = 1$, $\eta(x+1) = 0$, and $\xi_{t,x} = 1$.
Proposition 1 (Evans–Rajewsky–Speer [2]). The stationary current of Definition 1 on $\mathbb{Z}/N\mathbb{Z}$ at density $\rho \in [0,1]$ is $$j_{\mathrm{par}}(\rho) = \tfrac{1}{2}\!\left(1 - \sqrt{1 - 4p\rho(1-\rho)}\,\right).$$
Proposition 2 (Rajewsky–Santen–Schadschneider–Schreckenberg [1]). The stationary current of Definition 2 on $\mathbb{Z}/N\mathbb{Z}$ at density $\rho \in [0,1]$ is $$j_{\mathrm{seq}}(\rho) = \frac{p\,\rho(1-\rho)}{1 - p\rho}.$$
Both currents are strictly concave on $[0,1]$ for $p \in (0,1)$. At the deterministic boundary $p = 1$ one has $j_{\mathrm{par}}(\rho) = \min(\rho, 1-\rho)$ and $j_{\mathrm{seq}}(\rho) = \rho$.
Scaling limit under step initial condition. Under Eulerian scaling $\xi = x/T$, the empirical measure $T^{-1}\sum_k \delta_{x_k(T)/T}$ is expected to converge in probability to $\rho_\infty(\xi)\, d\xi$, where $\rho_\infty$ is the entropy solution of $\partial_t \rho + \partial_x j(\rho) = 0$ with initial data $\rho_0(\xi) = \mathbf{1}_{\xi \le 0}$; see Kipnis–Landim [4]. For a strictly concave flux this yields the rarefaction $$\rho_\infty(\xi) = \begin{cases} 1, & \xi \le j'(1), \\ (j')^{-1}(\xi), & j'(1) < \xi < j'(0), \\ 0, & \xi \ge j'(0). \end{cases}$$
Corollary (explicit fans).
Parallel. $j_{\mathrm{par}}'(0) = p$, $j_{\mathrm{par}}'(1) = -p$; on $\xi \in (-p, p)$, $$\rho_\infty^{\mathrm{par}}(\xi) = \tfrac{1}{2}\!\left(1 - \mathrm{sgn}(\xi)\sqrt{\tfrac{\xi^2(1-p)}{p\,(p - \xi^2)}}\right).$$
Sequential. $j_{\mathrm{seq}}'(0) = p$, $j_{\mathrm{seq}}'(1) = -p/(1-p)$; on $\xi \in \bigl(-\tfrac{p}{1-p},\, p\bigr)$, $$\rho_\infty^{\mathrm{seq}}(\xi) = \tfrac{1}{p}\!\left(1 - \sqrt{\tfrac{1-p}{1 - \xi}}\,\right).$$
References
- N. Rajewsky, L. Santen, A. Schadschneider, M. Schreckenberg. The asymmetric exclusion process: Comparison of update procedures. J. Stat. Phys. 92 (1998), 151–194.
- M. R. Evans, N. Rajewsky, E. R. Speer. Exact solution of a cellular automaton for traffic. J. Stat. Phys. 95 (1999), 45–96. arXiv:cond-mat/9903287.
- A. M. Povolotsky, V. B. Priezzhev. Determinant solution for the totally asymmetric exclusion process with parallel update. J. Stat. Mech. (2006), P07002.
- C. Kipnis, C. Landim. Scaling Limits of Interacting Particle Systems. Grundlehren der mathematischen Wissenschaften 320, Springer, 1999.
- A. Borodin, P. L. Ferrari, M. Prähofer, T. Sasamoto. Fluctuation properties of the TASEP with periodic initial configuration. J. Stat. Phys. 129 (2007), 1055–1080. arXiv:math-ph/0608056.
code
(note: parameters in the code might differ from the ones in simulation results below)-
Link to code(Interactive simulation — see source) -
Link to code(C++ source for WASM (128-bit SIMD Bernoulli sampler))