Proxitaxis: an adaptive search strategy based on proximity and stochastic resetting

This paper introduces "proxitaxis," an adaptive search strategy that combines distance-dependent local moves, stochastic resetting, and dynamic resetting-position updates to maximize target capture probability, revealing that the optimal strategy undergoes generic phase transitions across all dimensions.

The Gist

Imagine you are lost in a massive, foggy forest, and you are looking for a hidden treasure chest. You have a special device: a Geiger counter-like gadget that tells you how far you are from the treasure (e.g., "You are 500 meters away"), but it gives you zero information about which direction to go. It doesn't point North, South, East, or West. It just beeps faster as you get closer.

This is the problem scientists call "Proxitaxis" (a fancy word combining "proximity" and "taxis," which means movement).

In this paper, a team of physicists proposes a clever, three-step strategy to find the treasure as quickly as possible when you only know the distance, not the direction. They call this strategy Proxitaxis.

Here is how it works, broken down into simple analogies:

The Three Ingredients of the Strategy

1. The "Smart Shuffle" (Distance-Dependent Speed)
Imagine you are walking through the forest.

• When you are far away: You walk slowly and take small, careful steps, looking around in every direction. You are "sluggish." This is because you don't know where to go, so you need to cover a lot of ground locally to find a clue.
• When you get closer: Your gadget beeps faster. You realize, "Hey, I'm getting warm!" So, you start running faster and taking bigger, more energetic steps. You become "hyper-active."
• The Analogy: Think of a dog sniffing for a bone. When the scent is faint, it sniffs slowly and methodically. When the scent gets strong, it starts sprinting toward the source. In the paper, they mathematically prove that speeding up as you get closer is a smart move.

2. The "Do-Over" Button (Stochastic Resetting)
Sometimes, even if you are moving fast, you might accidentally wander into a dead-end or a swamp, moving away from the treasure without realizing it.

• The Strategy: If you've been walking for a while and haven't found the treasure, you press a "Do-Over" button. You instantly teleport back to where you started (or a specific checkpoint) and try again.
• The Analogy: Imagine you are looking for a specific book in a giant library. You wander an aisle, realize you're in the wrong section, and instead of walking all the way back, you just teleport back to the entrance to pick a new aisle. This prevents you from wasting time wandering in circles.

3. The "Checkpoint Update" (The Inspection Move)
This is the most clever part.

• The Problem: If you teleport back to your original starting point every time, you might be teleporting back to a spot that is actually farther away from the treasure than where you are right now! That would be silly.
• The Solution: Every so often, you stop and check your gadget.
  • If you are closer to the treasure than when you started this round, you say, "Great! I'm making progress." You update your "Do-Over" checkpoint to be your current location. Now, if you get lost later, you teleport back to this new, better spot.
  • If you are farther away, you keep your old checkpoint.
• The Analogy: Think of a hiker climbing a mountain. Every hour, they check their GPS. If they are higher up than when they started the hour, they mark that spot as their new "safe base." If they slipped and went lower, they ignore it and keep their old base. This ensures they are always drifting upward toward the peak, even if they take wrong turns in between.

The Big Discovery: Phase Transitions

The authors did some heavy math to figure out the perfect way to use this strategy. They asked: "How fast should I run? How often should I press the 'Do-Over' button?"

They found something surprising: The answer changes drastically depending on how far away you start. It's like a traffic light system that changes based on how far you are from the city center.

• Scenario A (Very Far Away): The best strategy is to just walk normally. Don't run, don't teleport. Just keep exploring.
• Scenario B (Medium Distance): Now, you should start using the "Do-Over" button frequently, but still walk at a normal speed.
• Scenario C (Very Close): Now, you should run super fast (increase your speed) AND use the "Do-Over" button frequently.

The scientists call these sudden changes in strategy "Phase Transitions." It's like water suddenly turning into ice when the temperature drops. In this case, the "temperature" is your distance from the target, and the "ice" is a completely different way of moving.

Why Does This Matter?

1. For Nature: Animals often search for food or mates without knowing the exact direction (e.g., a sperm cell looking for an egg, or a shark smelling blood). This paper suggests that animals might naturally be using a version of this "Proxitaxis" strategy without even knowing the math behind it.
2. For Robots: If we build robots to search for oil leaks, radioactive sources, or shipwrecks in the deep ocean, they often can't "see" the target. They only have sensors that tell them "it's close." This paper gives engineers a perfect recipe for programming those robots to be super efficient.

The Bottom Line

Proxitaxis is a recipe for finding things when you only know "how close" you are, not "which way."

• Move slowly when far away.
• Move fast when close.
• Reset (teleport back) if you get lost.
• Update your base camp every time you get closer.

By mixing these three ingredients in the right proportions, you can find your target much faster than if you just wandered aimlessly. And the best part? The "right proportions" change automatically depending on how far you are from the goal!

Technical Summary

1. Problem Statement

The paper addresses the target search problem in scenarios where the searcher possesses partial information: they know their instantaneous distance ($R$) to the target but lack information regarding the direction.

• Context: This is distinct from chemotaxis (where a concentration gradient provides direction) and infotaxis (which relies on information gradients but is often computationally complex).
• Real-world Examples: Searching for a radioactive source with a Geiger counter, detecting gas leaks, or animals foraging in turbulent environments where signals are intermittent.
• Challenge: How to design an efficient search protocol using only distance data to maximize the probability of capturing a target within a finite time, without needing to store the entire history of the search.

2. Methodology: The Proxitaxis Strategy

The authors propose a new search strategy called Proxitaxis, which combines three specific components into a recursive, memory-based protocol:

1. Distance-Dependent Diffusion: The searcher performs local stochastic moves (diffusion) where the diffusion coefficient $D(R)$ depends on the distance $R$ from the target.

   • Rationale: Inspired by biological observations (e.g., sperm hyper-activation), the strategy assumes the searcher moves more sluggishly (low $D$) when far away to explore broadly, and becomes more active (high $D$) when close to the target to scan intensively.
   • Model: $D(R) = R^{-\alpha}$, where $\alpha$ is a control parameter.

2. Stochastic Resetting: The searcher intermittently resets its position to a specific location $\vec{R}_0$ with a rate $r$.

   • Rationale: This eliminates "rogue" trajectories where diffusion takes the searcher too far away from the target, effectively restarting the search from a known point.

3. Dynamic Inspection and Updating: The searcher does not reset to a fixed initial point forever. Instead, it performs an "inspection" at random intervals (Poisson distributed with rate $b$).

   • Logic: At the end of an interval $\tau$, the searcher compares its current distance $R(\tau)$ with the distance at the start of the interval $R_0$.
   • Update Rule: If $R(\tau) < R_0$ (closer to the target), the new reset point $\vec{R}_0^{new}$ is updated to the current position. If $R(\tau) > R_0$, the reset point remains unchanged.
   • Effect: This creates a net drift toward the target without requiring full directional memory, only a comparison of distances.

Mathematical Framework:

• The system is modeled in $d$-dimensions with a spherical target of radius $\epsilon$.
• The dynamics are governed by a Langevin equation with multiplicative noise (due to $D(R)$) and stochastic resetting.
• The authors derive the capture probability $C(R_0)$ (the probability of finding the target within the lifetime of the strategy) using the backward Fokker-Planck equation and renewal theory.
• The survival probability $Q_0$ (without resetting) is solved analytically using modified Bessel functions, leading to an exact expression for $C(R_0)$ in terms of the control parameters $(r, \alpha)$.

3. Key Contributions

1. Definition of Proxitaxis: Introduction of a simple, analytically tractable search strategy specifically designed for distance-only information.
2. Analytical Optimization: Derivation of an exact formula for the capture probability $C(R_0)$ in arbitrary dimensions $d$, allowing for the explicit optimization of control parameters $(r, \alpha)$.
3. Discovery of Phase Transitions: Identification of multiple phase transitions in the optimal parameters $(r, \alpha)$ as a function of the initial distance $R_0$ and the inspection rate $b$.
4. Universality: Demonstration that the leading-order behavior of optimal parameters near the target ($R_0 \to 1^+$) is universal across all dimensions.

4. Key Results

A. Optimization of Capture Probability

The authors show that for any fixed initial distance $R_0$ and inspection rate $b$, there exists a unique global maximum for the capture probability $C(R_0)$ by tuning the resetting rate $r$ and the diffusion exponent $\alpha$.

B. Phase Transitions in Optimal Parameters

The optimal strategy undergoes distinct phase transitions depending on the initial distance $R_0$ and the inspection rate $b$. A critical threshold $b^*$ exists (approx. 1.195 in 1D).

• Case 1: Low Inspection Rate ($b < b^*$)

  • Far from target ($R_0 > R^*_{02}$): Optimal strategy is standard diffusion ($r=0, \alpha=0$).
  • Intermediate distance ($R^*_{01} < R_0 < R^*_{02}$): Optimal strategy is standard diffusion with stochastic resetting ($r>0, \alpha=0$).
  • Close to target ($R_0 < R^*_{01}$): Optimal strategy involves both resetting and distance-dependent diffusion ($r>0, \alpha>0$).

• Case 2: High Inspection Rate ($b > b^*$)

  • Far from target: Standard diffusion ($r=0, \alpha=0$).
  • Intermediate distance ($R^*_{03} < R_0 < R^*_{04}$): Only distance-dependent diffusion is optimal, without resetting ($r=0, \alpha>0$). This is a unique regime where the inspection mechanism alone drives efficiency without the need for resetting.
  • Close to target: Both resetting and distance-dependent diffusion are active ($r>0, \alpha>0$).

C. Universal Divergence

As the searcher approaches the target ($R_0 \to 1^+$), the optimal parameters diverge universally across all dimensions:
$$\alpha \approx \frac{1}{R_0 - 1}, \quad r \approx \frac{9}{16e(R_0 - 1)^2}$$
This indicates that as the searcher gets very close, it should increase its activity ($\alpha \to \infty$) and reset infinitely often ($r \to \infty$) to ensure capture, a behavior regularized in physical reality by a short-distance cutoff.

D. Dimensional Robustness

The phase transitions and the universal divergence of parameters are generic and observed in 1D, 2D, and 3D, suggesting the strategy is robust regardless of spatial dimensionality.

5. Significance

• Theoretical Benchmark: Proxitaxis provides a rare example of a realistic search strategy that is fully analytically solvable and optimizable. It serves as a benchmark for comparing more complex strategies (like infotaxis) or experimental data.
• Biological Insight: The results offer a theoretical framework for understanding animal foraging behaviors where directional cues are absent but distance cues (e.g., intensity of a scent or sound) are available. The "inspection" mechanism mimics biological decision-making processes.
• Engineering Applications: The strategy is highly relevant for designing autonomous search robots (e.g., underwater drones, mining robots) operating in environments where GPS or directional sensors fail, and only proximity data is available.
• Fundamental Physics: The work deepens the understanding of stochastic resetting in heterogeneous media, revealing that the interplay between space-dependent diffusion and resetting leads to rich non-equilibrium phase transitions.

In conclusion, the paper establishes that a simple, adaptive strategy combining distance-dependent diffusion, stochastic resetting, and periodic distance comparison can achieve near-optimal search efficiency in the absence of directional information, exhibiting complex phase behavior that is universal across dimensions.