On Non-Lipschitz Path-Dependent SDEs with Weighted Running Extrema: Existence, Uniqueness, and Strong Convergence Rates

Many of the most critical systems in our world do not react only to what is happening right now; they react to their entire history. A bridge does not simply feel the weight of a single truck crossing it today; its structural integrity depends on the cumulative stress of every heavy load it has ever carried. Similarly, a financial market does not just respond to the current price of a stock; it often reacts to the highest price that stock has ever reached, or the lowest point it has fallen to. These historical highs and lows, known as running extrema, act as memory triggers that can change the behavior of a system in profound ways. To understand and predict such systems, scientists use a type of mathematical equation called a stochastic differential equation. These are tools for describing how things change over time when they are influenced by both predictable forces and random, unpredictable jolts, like the wind or market panic. For decades, mathematicians have been able to solve these equations easily, but only when the rules governing the system were smooth and predictable. However, the real world is often rougher, with rules that change abruptly or behave in ways that defy simple smoothing.

This paper tackles a difficult problem at the intersection of history and randomness. The researchers focused on a specific class of equations where the future path of a system depends on its highest and lowest points up to that moment, but where the rules governing the system are "non-Lipschitz." In plain language, this means the rules are not perfectly smooth; they can have sharp corners or behave in ways that make standard mathematical tools break down. The team, led by Mohamed Bourza and Kamal Hiderah, set out to prove two things: first, that these complicated equations actually have a single, well-defined solution, and second, that we can calculate that solution using a computer with a known level of accuracy. They did not just guess that these solutions exist; they provided a rigorous mathematical proof that they do, even under these difficult, jagged conditions. Furthermore, they designed a specific computer algorithm to simulate these systems and proved exactly how close the computer's answer gets to the true answer as the calculation becomes more detailed.

The researchers began by looking at equations that include weighted running maxima and minima. Imagine a river flowing into a dam. The water level at any moment depends on the current flow, but also on the highest water level the dam has ever seen, which might trigger a spillway, and the lowest level, which might trigger a release of stored water. The "weights" in the equation are like adjustable knobs that determine how much influence these historical highs and lows have on the current state. The team showed that if these knobs are set within a certain safe range—specifically, if the influence of the history is not so strong that it causes the system to spiral out of control—then the system behaves in a stable, predictable way. They proved that for any starting point, there is one and only one path the system will follow. This is a crucial step because, without such a guarantee, any computer simulation would be meaningless, as it might be calculating a path that doesn't actually exist or finding multiple conflicting paths.

To reach this conclusion, the authors had to develop new mathematical techniques. Standard methods for solving these equations rely on the assumption that the rules are smooth, like a gentle slope. But in the scenarios they studied, the rules are more like a jagged cliff face. The team used a specialized inequality, a type of mathematical bound that helps control how fast a system can grow, to handle these rough edges. They combined this with a step-by-step construction method, building the solution layer by layer to show that it converges to a single, stable result. This proof is significant because it extends the reach of mathematical certainty into areas where previous tools failed, covering applications from the pricing of complex financial options to the fatigue analysis of materials that degrade under peak stress.

Once the existence of the solution was secured, the researchers turned to the practical problem of how to find it. Since these equations cannot be solved by hand, they must be approximated using computers. The team designed a discrete version of a standard algorithm known as the Euler-Maruyama scheme, but they modified it to explicitly track the highest and lowest points as the simulation moves forward in time. In a computer, time is chopped into tiny steps, and the algorithm calculates the state of the system at each step. The challenge was that the "running maximum" in a computer is just a list of past values, and switching from a continuous history to a list of steps introduces a small error. The researchers proved that this error does not accumulate uncontrollably. Instead, they showed that the difference between the computer's answer and the true mathematical answer shrinks at a predictable rate as the time steps get smaller.

The team tested their theory with a specific example involving a system where the rules change in a logarithmic way, a common feature in non-smooth physical and economic models. They ran thousands of simulations on a computer, comparing the results of their new method against a highly precise benchmark. The results confirmed their theoretical predictions: the error decreased steadily as the simulation became more detailed, following the exact rate they had calculated. This means that engineers and financial analysts can now use this method with confidence, knowing exactly how accurate their simulations will be. The work bridges the gap between abstract mathematical theory and real-world application, providing a reliable tool for modeling systems where history matters and the rules are not perfectly smooth. By establishing both the existence of solutions and the reliability of the methods to find them, the paper opens the door to more accurate simulations of complex, memory-dependent systems in fields ranging from structural engineering to risk management.