Track new results across pure and applied mathematics.
Search papers, labs, and topics across Lattice.
The authors construct the Latent Twin Operator (LTO), a neural operator that combines a ConvCNP-style continuous encoder-decoder with an arbitrary-time latent evolution map to fully decouple temporal PDE integration from spatial sensor discretizations. Crucially, the work proves an explicit $\mathcal{O}(N^{-2/(3D)})$ bound on context discretization error for spatial dimension $D$, establishing formal convergence guarantees under quasi-uniform sensor refinement. On benchmarks including 2D Navier–Stokes, LTO transfers seamlessly across spatial sensor densities with fixed weights while direct latent stepping avoids the compounding error typical of autoregressive rollouts.
Arbitrary sensor layouts no longer break PDE surrogates: direct latent evolution achieves zero-shot spatial transfer while proving an explicit $\mathcal{O}(N^{-2/(3D)})$ bound on sensor discretization error.
Surrogate models deployed on real physical systems rarely see data at fixed resolutions: sensor configurations vary across deployments and may evolve over time as sensing infrastructure changes. We introduce the Latent Twin Operator (LTO), a latent-space surrogate for time-evolving PDEs with a Convolutional Conditional Neural Process (ConvCNP)-style encoder and decoder that accepts a context set of $N$ sensor observations with arbitrary placement and can be queried at any resolution. A learned latent evolution map advances the encoded state directly between arbitrary time points, decoupling temporal evolution from the observation and query discretizations. We derive an explicit $\mathcal{O}(N^{-2/(3D)})$ rate for the context discretization error in spatial dimension $D$ under quasi-uniform refinement. We verify the predicted decay empirically on a 2D heat-equation benchmark. Across time-dependent PDE benchmarks, LTO achieves strong accuracy under one-step comparisons and transfers from native to coarser spatial resolutions with fixed parameters. On Navier--Stokes, its direct latent evolution further reduces error over longer prediction horizons relative to recursive evaluation.