Papers for

computational physics engineers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

TangentPDE improves neural PDE operators with local sensitivity training

TangentPDE: Learning Local Operator Responses with Sparse Finite-Difference Supervision

Abstract: Neural PDE operators are often optimized through state reconstruction, which largely constrains operator values at observed inputs but does not fully characterize their sensitivity to local perturbations. A more faithful approximation of operator behavior requires not only accurate state predictions, but also correct variations along local perturbation directions. We present TangentPDE, which uses finite differences to approximate directional derivatives and constructs task-relevant finite-step responses from the state outputs of a reference solver at sparse paired inputs. Its objective jointly constrains state fidelity, response-direction alignment, and response-magnitude calibration, thereby learning both operator values and their local variations. Because the response labels are constructed solely from state outputs at paired inputs, the method requires neither explicit derivative labels, linearized or adjoint PDE solves, nor full Jacobians, and remains applicable to nonlinear solvers without derivative interfaces. Building on this, we additionally propose a selectively executed adaptation mechanism to strengthen local operator representations under finite-difference supervision. Budget-matched and correspondence-perturbed controls isolate the effect of relational supervision from improvements attributable to additional state observations or increased data volume. Designed as a plug-and-play training framework, TangentPDE integrates with diverse neural-operator backbones. Across four PDE systems and 20 task-backbone settings, TangentPDE lowers the three-seed mean state error in 18 settings and the corresponding Rollout@20 error in 18, while intrinsic finite-step response gains are strongest on DiffReact and remain task- and backbone-dependent elsewhere.

Wed 18 NovMachine Learning
The gist
Solving partial differential equations (PDEs) with neural networks usually focuses on matching the final solution but overlooks how the solution changes when inputs are slightly altered. The authors present TangentPDE, a method that teaches neural models to better understand these local changes by comparing how small input tweaks affect outputs using finite differences. This approach doesn’t need complex derivative calculations and works with nonlinear PDE solvers. Their tests show TangentPDE reduces errors in many cases by learning both the solution and its local variations more accurately.
Open → 2611.24873v1

Ermakov-based embedding improves physics-informed neural network accuracy

Ermakov--Based embeddings as a parameter-free layer for physics-informed neural networks

Abstract: Evolution equations with time-dependent coefficients force a neural solver to spend capacity on a scale that the equation already determines. We show that a single classical object, the Ermakov-Pinney scale function $ρ$, can be written into the input layer of a physics-informed neural network as a parameter-free embedding. One construction covers three problem families through the two signs of the Pinney constant $c$: linear diffusion ($c=-1$, $ω\equiv0$), diffusion in a time-dependent confining trap ($c=-1$, $ω\neq0$), and the quantum oscillator with time-dependent frequency ($c=+1$). An exact moment identity gives $ρ$ in closed form from the initial data alone, so no auxiliary network or knowledge of the solution is needed. Our central finding is comparative: each family requires a different component of the construction and has a different honest control. In family I, the diffusive-time reparametrisation alone reproduces most of the gain, so that family is a poor test of the idea and shows why an ablation is mandatory. In family II no time reparametrisation exists, the control degrades accuracy, and the amplitude prefactor becomes essential (40.8x). In family III the adiabatic approximation also fails, and the quadratic phase factor is what matters (5.7x). The method does not help when the reduced solution does not converge to a fixed point: its extrapolation advantage then disappears. The reduction itself is the classical scaling-variable change of the self-similar-asymptotics literature; what is new is the moment identity that makes it architectural for time-dependent coefficients, and the comparative anatomy above.

Mon 2 NovMachine Learning
The gist
Some complex physics problems involve equations where certain scales change over time, and neural networks have to learn these scales from scratch, which wastes effort. The authors show that by adding a specific formula called the Ermakov-Pinney scale function directly into the neural network's input, the network can better handle these time-dependent problems without extra parameters or guesswork. They tested this approach on three different types of physics problems and found that different parts of the formula mattered depending on the problem type. However, this method doesn't help when the underlying solution is too complex or doesn't settle down over time.
Open → 2611.03964v1

Jet tagger performance differences traced to missed physics information

What the Primary Lund Plane Omits: A Model-Stitching Diagnosis of a Jet-Tagger Representation Gap

Abstract: Lund-plane and Particle Transformer (ParT) jet taggers can differ in performance because one representation is weaker, or because a weaker downstream computation reads an adequate one. Standalone performance cannot separate these. Model stitching can: it links the frozen early layers of one network to the frozen late layers of another through a single trained linear map. As a positive control we use an incomplete Lund representation, the primary declustering branch, whose deficit is known independently. Stitched into ParT's downstream, it reaches a background rejection at least as much as its own downstream recovers, yet far below ParT's background rejection: the deficit is representational, not computational. A hypergraph reader over splitting triplets does no better, ruling out higher-order correlations as the cause. The gap is invisible in AUC, which stays within 1\% of its target value, and appears only in background rejection at low signal efficiency. It localizes to 0.6\% of background jets, at and above the top mass, with more constituents but fewer primary splittings than typical background. Rebuilding the representation on the full Cambridge/Aachen declustering tree repairs the deficit as predicted: standalone rejection rises, closing 72\% of the gap. Masking nodes at inference shows the repaired network still relies almost entirely on the primary branch; the secondary splittings make the representation learnable rather than supplying information it reads. Along the primary branch, the widest-angle splitting, near the top and $W$ opening angles, is worth nearly twice a random one. For quark/gluon tagging the primary branch is far less deficient, retaining 77.6\% of ParT's rejection (46.9\% for top), and the full tree closes only 33\% of the gap (72\% for top).

Tue 27 OctArtificial Intelligence
The gist
Jet taggers are tools used to identify particle jets in physics experiments, but some do better because of the information they use. The authors show that a commonly used jet representation called the primary Lund plane misses important details, causing weaker performance compared to a Particle Transformer method. By combining parts of both approaches, they pinpoint that the problem is missing representation, not poor computation. Adding back the missed parts improves performance, especially for identifying top quark jets.
Open → 2610.37137v1