Mean first passage times of velocity jump processes in higher dimensions

This paper presents a general framework for estimating mean first passage times of velocity jump processes in higher dimensions, deriving universal formulas for biased motion and revealing how directional persistence leads to anomalous scaling and finite passage times in the narrow capture limit where standard diffusion fails.

The Gist

Imagine you are trying to find a specific coffee shop in a giant, foggy city. You have a map, but you can't see very far. How you move matters just as much as where you are.

This paper is about understanding how long it takes, on average, for a moving object to find a target when that object doesn't just wander randomly like a drunk person (standard diffusion). Instead, this object has a "momentum." It runs in a straight line for a while, then suddenly stops, spins around, and picks a new direction to run again.

The authors call this a "Velocity Jump Process." Think of it like a bacterium swimming, a dog on a leash that pulls in one direction before the owner yanks it another way, or a stock price that trends up for a while before a sudden crash.

Here is the breakdown of their discovery, using simple metaphors:

1. The Problem: The "Drunk" vs. The "Runner"

In standard physics, we often assume things move like a drunk person: they take a step left, then right, then left again, with no memory of where they were going. This is called diffusion. If the target is very small (like a needle in a haystack), it can take a very long time for a drunk person to find it.

But many things in nature don't act like drunks.

• Bacteria swim in straight lines ("runs") and then tumble to change direction.
• Animals follow scent trails or magnetic fields.
• Stocks trend in one direction before a sudden shift.

These are Velocity Jump Processes. They have "persistence." They keep going in the same direction until something forces them to change. The big question the authors asked is: Does this persistence make them find the target faster or slower, and how do we calculate that time?

2. The Solution: A New "GPS" for Moving Things

The authors created a new mathematical "GPS" (a set of equations) to predict the Mean First Passage Time (MFPT).

• MFPT is just a fancy way of saying: "On average, how long does it take to hit the target?"

They found that when the "straight runs" are short compared to the size of the city (a condition they call a "low Knudsen number"), the complex, jerky motion of the runner can be smoothed out into a simpler equation.

The Analogy:
Imagine you are driving a car that has a mind of its own. Sometimes it wants to go straight, sometimes it wants to turn left.

• If the car is unbiased (it turns randomly), it acts like a standard car drifting in traffic.
• If the car is biased (it really wants to go North), it acts like a car with a strong GPS signal pulling it toward the destination.

The authors' equation calculates the travel time by looking at two "bias functions":

1. The Drift: How much is the wind pushing you in a specific direction?
2. The Spread: How much does the car wobble around that direction?

3. The Surprising Discovery: The "Narrow Capture" Paradox

The most exciting part of the paper is what happens when the target is tiny (like a tiny door in a giant room).

• Standard Diffusion (The Drunk): If you are wandering randomly, the time it takes to find a tiny door grows infinitely large as the door gets smaller. It's like looking for a needle in a haystack; if the needle shrinks, you might never find it.
• Velocity Jump (The Runner): The authors found that if the runner has a strong "bias" (a strong desire to go in a specific direction), the time to find the tiny target stays finite, even if the target is microscopic.

The Metaphor:
Imagine you are in a dark room looking for a tiny keyhole.

• If you are spinning in circles randomly, you might miss the keyhole forever.
• But if you are running in a straight line toward a specific wall, you will eventually hit that wall. Even if the keyhole is the size of a pinprick, your momentum ensures you don't just wander away forever. You will eventually slam into the wall and find it.

This is called anomalous scaling. It means that in the real world (biology, finance, ecology), having a "direction" or a "goal" can make finding a tiny target infinitely easier than pure randomness would suggest.

4. The "Langevin" Shortcut

Finally, the authors realized that all this complex math about "jumping velocities" can be replaced by a much simpler simulation called a Langevin process.

The Analogy:
Instead of simulating every single "run and tumble" of a bacterium (which is computationally heavy), you can just simulate a particle moving in a river with a current (the bias) and some turbulence (the randomness).

• Old way: Simulate 10,000 steps of a zig-zagging robot.
• New way: Simulate a boat floating down a river with a slight current.

They proved that this simpler "boat in a river" model gives the exact same answer for how long it takes to reach the destination. This is huge for scientists because it makes computer simulations much faster and easier.

Why Does This Matter?

This isn't just about math; it explains real life:

• Biology: How long does it take for an immune cell to find a virus? If the cell has a "sense" of direction (chemotaxis), it finds the virus much faster than if it just wandered randomly.
• Ecology: How long does it take a lost animal to find its way home?
• Finance: How long until a stock hits a "sell" threshold? If the market has a strong trend, the time to hit that threshold is different than if the market is just noisy.

In a nutshell:
This paper gives us a new rulebook for predicting how long it takes to find something when the searcher has momentum and a sense of direction. It shows that persistence is powerful: even a little bit of direction can prevent a search from taking forever, turning a hopeless wander into a successful hunt.

Technical Summary

1. Problem Statement

First-passage phenomena are fundamental across physics, biology, and finance, describing the time it takes for a stochastic process to reach a specific threshold or target. While standard diffusion models are widely used, many real-world systems exhibit velocity jump processes (also known as run-and-tumble or persistent random walks). In these systems, particles move with a fixed speed in a specific direction for a random duration before stochastically reorienting.

The core challenges addressed in this paper are:

• Higher Dimensions: Existing analytical frameworks for first-passage times (FPT) are well-developed for 1D (telegrapher's equations) but remain underdeveloped in higher dimensions ($d > 1$), particularly when directional bias (persistence) is present.
• Complex Reorientation: Unlike 1D where particles only turn left or right, higher-dimensional particles can reorient along a continuum of directions.
• Anisotropy and Bias: Many biological and physical systems (e.g., bacteria chemotaxis, animal movement, financial markets) exhibit non-isotropic reorientation distributions (e.g., von Mises, Fisher) where motion is biased toward a preferred direction.
• Narrow Capture: Understanding how directional persistence affects the time to reach a vanishingly small target (narrow capture limit), where standard diffusion often predicts divergence.

2. Methodology

The authors develop a general mathematical framework to estimate the Mean First Passage Time (MFPT), denoted as $\Theta(x, v)$, for a particle starting at position $x$ with velocity $v$.

• Governing Equations:

  • The system is modeled using the forward Kolmogorov equation for the probability density $P(x, v, t)$ and the backward Kolmogorov equation for the survival probability $S(x, v, t)$.
  • The jump process is characterized by a reorientation rate $\mu$ and a memoryless probability density function $q(x, v)$ governing the new velocity direction.
  • The MFPT is derived as the first moment of the survival probability: $\Theta(x, v) = \int_0^\infty S(x, v, t) dt$.

• Asymptotic Analysis (Small Knudsen Number):

  • The authors assume a small Knudsen number $\epsilon = \ell/L \ll 1$, where $\ell = \sigma/\mu$ is the mean free path and $L$ is the characteristic distance to the target.
  • They employ a multiple-scale asymptotic expansion to separate fast (velocity reorientation) and slow (spatial diffusion) time scales.
  • This leads to a closed-form, self-consistent differential equation for the leading-order MFPT $T(x)$, which is independent of the initial velocity orientation when the distance to the target is large compared to the mean free path.

• Langevin Representation:

  • By interpreting the derived differential operator via Itô calculus, the authors construct an equivalent Langevin stochastic differential equation (SDE). This allows for efficient numerical simulation and reconstruction of bias terms from experimental data.

3. Key Contributions

A. Generalized MFPT Equation

The paper derives a universal partial differential equation for the MFPT $T(x)$ in $d$-dimensions:
$$D(x) : \nabla \otimes \nabla T(x) + b(x) \cdot \nabla T(x) = -1$$
where:

• $D(x)$ is a diffusion tensor dependent on the angular distribution of reorientations.
• $b(x)$ is a drift term arising from directional bias.
• This equation generalizes the standard diffusion equation ($D \nabla^2 T = -1$) to include anisotropy and drift, valid for fixed-speed velocity jump processes.

B. Universal Form for Angular Distributions

The authors demonstrate that for a broad class of angular distributions (including von Mises-Fisher, wrapped Cauchy, and elliptical families), the MFPT equation can be parameterized by two bias functions, $\alpha(x)$ and $\beta(x)$, which depend on the concentration parameter $\kappa(x)$:

• $\alpha(x)$ modulates the diffusion tensor (anisotropy).
• $\beta(x)$ modulates the drift velocity.
• These parameters interpolate between isotropic diffusion ($\kappa \to 0$, $\alpha, \beta \to 0$) and ballistic motion ($\kappa \to \infty$, $\alpha, \beta \to 1$).

C. Narrow Capture and Anomalous Scaling

In the limit of a vanishingly small target (narrow capture), the authors derive scaling laws for the global MFPT $\tau_g$:

• Radial Bias: The MFPT scales as $\tau_g \sim \zeta^{\frac{d(\alpha-1)+2}{\alpha+1}}$ (where $\zeta$ is the target size ratio).
  • Crucially, for $d=2$ and $\alpha > 0$, and $d=3$ and $\alpha > 1/3$, the MFPT remains finite as the target size approaches zero. This contrasts sharply with standard diffusion, where the MFPT diverges in 2D ($\sim \log \zeta$) and 3D ($\sim \zeta^{-1}$).
• Tangential Bias: In 2D, tangential bias leads to a different scaling $\tau_g \sim \zeta^{-2\alpha/(1-\alpha)}$, which can diverge even for small biases, indicating that tangential preferences can significantly delay capture.

D. Langevin Approximation

The authors provide a Langevin equation $dx = b(x)dt + [2D(x)]^{1/2} dW_t$ that accurately reproduces the first-passage statistics of the underlying velocity jump process. This offers a computationally efficient alternative to tracking full velocity dynamics.

4. Results

• Analytical vs. Numerical Agreement: The analytical solutions derived from the generalized MFPT equation (Eq. 9) show excellent agreement with direct particle simulations for $d=2$ and $d=3$ across various bias scenarios (radial and tangential).
• Sensitivity to Bias: The MFPT is highly sensitive to directional bias. Even modest biases can alter the MFPT by several orders of magnitude compared to isotropic diffusion.
  • Positive radial bias (towards the target) significantly reduces the MFPT.
  • Negative radial bias (away from the target) induces detours, delaying exit.
• Boundary Layer Effects: Deviations between the asymptotic theory and simulations occur only in a thin boundary layer where the distance to the target is comparable to the mean free path ($r \approx R_0 - \ell$).
• Application to Plume Tracking: The framework was applied to a particle searching for the maximum concentration of a Gaussian plume. The Langevin approximation successfully captured both the MFPT and the full survival probability distribution.

5. Significance

• Unified Framework: The paper provides a unified theoretical framework to study first-passage times for diverse systems, including run-and-tumble bacteria under chemotaxis, active swimmers, animal movement, and field-driven colloidal matter.
• Biological and Ecological Relevance: The results offer new insights into how persistence and directional cues affect search efficiency, resilience in ecosystems, and the onset of biological events (e.g., immune responses, biochemical reactions).
• Experimental Utility: The derived Langevin representation allows researchers to reconstruct bias terms ($\alpha, \beta$) directly from experimentally observed trajectories, bridging the gap between microscopic dynamics and macroscopic transport properties.
• Theoretical Advancement: By resolving the narrow capture problem for persistent random walks, the paper corrects the misconception that diffusion is the only relevant limit, showing that persistence can fundamentally alter scaling laws and prevent divergence in search times.