Rare events of small-noise Doob conditioned processes
summary
The gist
Doob fixed-time conditioning enables the sampling of rare trajectories of Markov processes by modifying the drift so that reaching a prescribed target at a given time is guaranteed The paper studies…
In short
The paper introduces Doob fixed-time conditioning to sample rare trajectories of Markov processes by modifying the process drift. It studies these conditioned paths using a moment generating function in the weak-noise large deviation regime, showing how to derive an action principle for fluctuations.
Key concepts
- Doob's h-function
- This function, ht(x), represents the probability that a Markov process starting at state x will reach a specific target state within a given time T. It is used to dynamically modify the process drift to ensure it hits the target at time T.
- Modified diffusion (X~ϵ t)
- This is the Doob conditioned process. Its dynamics are altered by adding a drift term proportional to the gradient of Doob's h-function and scaled by $\sqrt{\epsilon}$. This modification makes the resulting path measure optimal in terms of minimizing relative entropy compared to the original process.
- Action (M~τ,x(s))
- The action is a leading exponential term extracted from the moment generating function. It can be represented as a variational principle involving a Lagrangian. This allows researchers to characterize fluctuations of conditioned processes by treating them as optimal control problems.
- Weak-noise limit ($\epsilon \to 0$)
- When the noise parameter $\epsilon$ approaches zero, the generalized Feynman–Kac equation simplifies into a Hamilton–Jacobi problem. The solution to this problem yields the leading exponential contribution to the moment generating function, which is crucial for finding rare event statistics.
Terminology used across episodes
This episode discusses
The paper
Rare events of small-noise Doob conditioned processes · Read on arXiv
Iago N. Mamede, Francesco Coghi
Instituto de Física da Universidade de Sãao Paulo · UHasselt Faculty of Sciences, Theory Lab · School of Physics and Astronomy University of Nottingham · Centre for the Mathematics and Theoretical Physics of Quantum Non-Equilibrium Systems University of Nottingham
Doob fixed-time conditioning enables the sampling of rare trajectories of Markov processes by modifying the drift so that reaching a prescribed target at a given time is guaranteed. We study the statistics of this conditioned path ensemble through the moment generating function in the weak-noise large deviation regime. Since the Doob drift is rarely available in closed form, we reinterpret the conditioned ensemble as the original process post-selected on the terminal constraint, thereby avoiding explicit construction of the Doob transform. This viewpoint then yields an optimal-control representation for the leading exponential contribution to the generating function, expressed as a variational principle with terminal boundary conditions set by the Doob end-point constraint. We illustrate the framework with two analytical examples and with an application to heat dissipation under Doob-controlled dynamics in a minimal model of biomolecular folding.
Transcript
Introduction to the show: ident: Condensed Matter. The day in condensed matter: materials, phases and the measurements under them.
Kai: We're looking at the paper: "Rare events of small-noise Doob conditioned processes".
Paper summary: Kai: So, looking at the paper "Rare events of small-noise Doob conditioned processes," it really boils down to showing how you can encode conditioning through a terminal boundary term <ref:2605.24537#pg2>. They derive a generalized Feynman-Kac equation for the unnormalised generating function from this setup <ref:2605.24537#pg1>.
Mira: And the crucial part is that they found a Lagrangian variational representation for the action, which means we can treat these fluctuations as an optimal control problem <ref:2605.24537#pg1>. This method is useful because it lets us tackle systems where the explicit Doob drift isn't available analytically <ref:2605.24537#pg1>.
Lev: For hardware, this implies we might not need to solve the difficult backward Kolmogorov equation for every system; instead, we can use these optimization schemes directly when characterizing fluctuations of conditioned processes <ref:2605.24537#pg1>. It moves the complexity from solving an intractable differential equation to setting up an optimization problem <ref:2605.24537#pg1>.
Kai: The authors used this framework on path observables, like heat dissipated by a Doob-conditioned process <ref:2605.24537#pg1>. They found that for the Ornstein-Uhlenbeck bridge, the rare trajectories that connect the bottom of the potential to a value far away in fixed time T will dissipate heat <ref:2605.24537#pg1>.
Mira: That connects to how we think about thermodynamics and rare events in these systems <ref:2605.24537#pg1>. It shows that the larger the parameter s, which relates to the conditioning, the greater the heat dissipated under this Doob protocol <ref:2605.24537#pg1>.
Lev: But I have to mention what they didn't do here; they focus on small-noise regimes and weak-noise large deviation regime for their statistics <ref:2605.24537#pg1>. So, applying this directly to very high noise or highly non-linear systems might need more work <ref:2605.24537#pg1>.
Kai: Right, so the paper proves we can use post-selection to find a way around the difficulty of calculating Doob's h-function and instead get an action that comes from an optimal control formulation <ref:2605.24537#pg1>. That’s what they set up in this paper.
Conclusion: Kai: So, this paper is about finding rare paths in Markov processes by using a technique called Doob conditioning to make sure you hit a specific target at exactly one time.
Mira: It’s looking at how to find these rare paths when the noise is small, which is kind of what we actually expect in many physical systems.
Kai: The authors are focusing on modifying the drift term in a way that guarantees the process reaches a certain spot by a specific time, and they're using this to study the statistics of those conditioned paths.
Mira: It’s really about taking these complicated paths and turning them into something we can actually calculate using tools like moment generating functions.
Kai: They show that you can derive a kind of action principle from this whole conditioning setup, which turns the problem into an optimal control problem for characterising fluctuations.
Mira: That means they aren't just looking at one path; they’re looking at the whole ensemble of paths and finding a way to quantify how much those rare events contribute to the overall statistics.
Kai: It seems like this method could be useful for understanding things in quantum hardware where you want to know about very specific, rare error events that might otherwise be too hard to measure directly.
Mira: Exactly, because it gives us a way to encode that desired outcome into the math itself rather than just hoping the path happens.
Kai: We'll see how this variational approach plays out when we apply it to things like heat dissipation in these conditioned processes next.
More episodes
- 2610.32678-Probing the limits of materials design via a comprehensive crystal database of 14.6 million structures with 140 properties
- 2610.33595-Native-oxide-driven layer asymmetry, two-step switching and exchange bias in Co/Ir/Co trilayers
- 2503.19572-Spin models from nonlinear cellular automata
- 2606.23350-Irreversibility Enhances Quantum-Enhanced Markov-Chain Monte Carlo
- 2608.02744-Two-Parameter Ansatz for the Violation of Eigenstate Thermalization
- 2210.00806-Low- and high-energy localization landscapes for tight-binding Hamiltonians in two-dimensional lattices
- 2409.09625-Selective switch for band-edge alignments in pentagonal CdSeTe monolayer: Atom-valley locking
- 2606.20168-Norms, overlaps and Yangian descendants for the Haldane-Shastry spin chain
- 2512.19503-Strain-induced splitting of the CCDW-NCCDW phase transition in 1T-TaS 2
- 2307.12638-Strong localization of microwaves beyond two dimensions in aperiodic Vogel spirals