Search papers, labs, and topics across Lattice.
To address the unmeasured fragility of neural PDE surrogates under changing sensor distributions, this work establishes PDE-OBS, a benchmark spanning 560,000 simulation fields across seven PDE families with decoupled, parameterized observation operators. Across 3,969 evaluations comparing seven baselines across nine observation patterns, cross-pattern transfer degraded accuracy across all 49 PDE-method pairs relative to pattern-matched references. Most critically, the findings reveal that test-time performance is not monotonic with sensor density: denser observations frequently fail to reduce error for fixed models without pattern-mixed training.
Giving a trained neural PDE surrogate denser sensor measurements at test time can actually degrade its predictions, exposing widespread distribution collapse under spatial observation shifts.
Physical-field reconstruction and forecasting depend on both measurement density and spatial layout, yet evaluation under a single observation pattern does not characterize performance when that pattern changes. We introduce PDE-OBS, an integrated benchmarking platform spanning numerical data generation, model training, and inference and evaluation under varying observation conditions. It combines 560,000 fields and trajectories from seven partial differential equation families with configurable observation operators and seven adapted baseline methods for stationary reconstruction and short-horizon forecasting. Separating observation construction from physical records allows users to specify parameterized patterns and deterministic mixtures for training and testing while preserving prediction targets and data splits. The evaluation protocol uses references trained for each test pattern to compare models on identical test observations and targets, alongside equal-count groups for spatial-layout comparisons. On a 14,000-record subset, we evaluate 441 trained models under nine test patterns, yielding 3,969 evaluations. Mean cross-pattern error exceeds mean matched-pattern error in all 49 PDE-method pairs, and this finding persists in a configuration-matched subset of 117 models. Denser test observations do not consistently reduce error for a fixed model. Mixed-pattern training on five completed pairs reduces large single-pattern transfer errors, although destination-trained references usually remain more accurate. Together, the benchmark and findings support systematic evaluation of observation-pattern sensitivity and provide a reusable workflow for developing methods under changing measurement conditions. Code: https://github.com/ru1ch3n/PDE-OBS.
