# Robin-Boundary Diffusion Sampler

- ID: 3075
- Canonical URL: https://synthcore.org/idea/3075/robin-boundary-diffusion-sampler
- API JSON: https://synthcore.org/api/idea/3075.json
- API Markdown: https://synthcore.org/api/idea/3075.md
- Verification status: unverified
- Source: [arXiv:2609.02665](https://arxiv.org/abs/2609.02665)
- Category: sampling
- Solves: accuracy, stability, sample-efficiency
- ML areas: diffusion, diffusion-sampling, architecture
- Math tags: stochastic-processes, pde, dynamical-systems, geometry
- Ratings: usefulness 5/10; difficulty 6/10; novelty 8/10

## Idea description

Train a score-based diffusion model on a domain with partially reactive constraints, replacing hard rejection or large boundary penalties by a Robin boundary condition. Samples approaching the constraint boundary acquire a hazard proportional to accumulated boundary local time, giving a continuous interpolation between reflection and absorption.

## Mathematical statement

Let X_t be Brownian motion with diffusion coefficient D in the wedge Omega = {(r, theta): r greater than 0, 0 less than theta less than alpha}. Let L_t denote boundary local time and q greater than or equal to 0 the boundary reactivity. The Robin survival weight is W_t = exp(-q L_t), so S_q(t given x_0) = E_x0[W_t]. The backward survival function obeys partial_t S_q = D Delta S_q in Omega, with normal_derivative S_q + q S_q = 0 on the wedge sides. For perfectly absorbing sides, the long-time survival law is S_infinity(t given x_0) proportional to t raised to the power -pi divided by (2 alpha). Dimensional analysis gives the Robin crossover time t_c proportional to 1 divided by (D q squared).

## Key formulas

- partial_t S_q(x,t) = D Delta S_q(x,t), and normal_derivative S_q(x,t) + q S_q(x,t) = 0 for x on the boundary of Omega.
- S_q(t given x_0) = E_x0[ exp(-q L_t) ].
- S_infinity(t given x_0) is proportional to t^(-pi/(2 alpha)) as t tends to infinity.
- t_c is proportional to 1/(D q^2), and d log S_q(t) divided by d log t tends to -pi/(2 alpha) when t is much larger than t_c.

## Implementation notes

Use a two-dimensional constrained latent space whose admissible region is a wedge, or locally approximate a more general constraint manifold by a wedge. Train a score network s_theta(x,t) with the standard denoising score-matching objective, then add boundary points and a Robin residual. If the network models a density p_theta, use R_b = normal_derivative p_theta + q p_theta. If it models a score, estimate the normal derivative by finite differences and penalize the equivalent logarithmic condition n dot gradient_x log p_theta + q = 0. In reverse Euler-Maruyama sampling, detect a proposed boundary crossing. Reflect the normal displacement with probability 1 minus min(1, q times sqrt(delta_t divided by D)), and absorb or terminate the trajectory otherwise; calibrate this discretization against direct local-time estimates. The paper contributes the Robin PDE, the local-time Feynman-Kac weight, and the absorbing persistence exponent; the network estimates the score and boundary density. First experiment: train a two-dimensional variance-preserving diffusion on Gaussian data restricted to wedges with alpha equal to pi/2, pi/3, and pi/4, comparing hard rejection, pure reflection, and Robin sampling for q equal to 0.1, 0.3, 1, and 3. Measure survival curves and their local log-slopes. The predicted signature is a crossover near t_c = 1/(D q^2), within a factor of two, followed by a late-time slope near -pi/(2 alpha), within 20 percent. If the crossover does not shift as q raised to the power -2, or the late slope is angle-independent, reject the implementation.

## Disclaimer

AI-generated research hypothesis, automatically tested. Not peer-reviewed.