Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Martingale Neural Operators: Learning Stochastic Marginals via Doob-Meyer Factorization

About

Neural operators excel as deterministic surrogates, but inevitably collapse to the conditional mean when applied to stochastic PDEs, discarding the variance and tail structure upon which uncertainty quantification depends. Recovering this structure typically requires Monte Carlo rollouts or grafted generative models, both of which surrender the one-shot efficiency and resolution invariance that define the operator paradigm. To resolve this, we draw on the Doob-Meyer theorem, which establishes that any semimartingale fundamentally decomposes into a predictable drift and an unpredictable, zero-mean martingale. Translating this theorem into an architectural prior, we introduce the Martingale Neural Operator (MNO). MNO maps an initial condition directly to the conditional mean and covariance of the terminal law, parameterized by a drift-like mean and a low-rank factor $B_\phi$ with $B_\phi^\top B_\phi$ positive semi-definite by construction. For our experiments, we use a Gaussian residual instantiation. Across 1D SPDEs, rough volatility, and 2D operator tasks, MNO reduces Wasserstein distance by up to $120\times$ on $\phi^4$ field theory and $68\times$ on stochastic Burgers, evaluating $\sim 3\times$ faster than a conditional diffusion baseline at matched wall-clock training budgets. On 2D tasks, MNO is comparable to FNO on zero-shot resolution transfer and turbulent flow, while quasi-deterministic systems such as Gray-Scott remain a failure mode.

Kai Hidajat• 2026

Related benchmarks

TaskDatasetResultRank
Rough Volatility modelingRough Volatility terminal-marginal results
Mean RMSE0.0223
15
1D Super-resolution1D Super-resolution data
Mean RMSE0.0205
12
1D stochastic terminal-law estimationStochastic Burgers (test)
Mean RMSE0.023
6
1D stochastic terminal-law estimationSPDEBench phi^4 (test)
Mean RMSE0.0086
4
Generative ModelingStochastic Burgers
Time (s)0.007
4
Event Risk PredictionReaction-diffusion
Mean RMSE0.9948
4
2D resolution transfer2D Gray-Scott 32 x 32 resolution
Mean RMSE0.7602
4
2D resolution transfer2D Gray-Scott 64 x 64 resolution
Mean RMSE0.7602
4
2D resolution transfer2D Gray-Scott 128 x 128 resolution
Mean RMSE8.4394
4
2D Field PredictionGray-Scott 2D (test)
Mean RMSE0.0345
4
Showing 10 of 20 rows

Other info

Follow for update