https://surrogatemodels.com/ News, papers, datasets, and funding and deals in Physics AI for engineers: Electromagnetics. en https://surrogatemodels.com/papers/clifford-layershttps://surrogatemodels.com/papers/clifford-layersThu, 12 Nov 2026 09:00:00 GMTScalar and vector fields at each point are packed into one multivector, and convolutions and Fourier layers act on these multivectors with Clifford algebra, so the coupling between, for example, pressure and velocity components is built into each layer.Aero & CFDElectromagneticsWeather & climate https://surrogatemodels.com/news/xplore-rwe-benchhttps://surrogatemodels.com/news/xplore-rwe-benchTue, 27 Oct 2026 09:00:00 GMTXplore builds what it calls high-fidelity reinforcement learning environments and evals around real engineering workflows. Its first public benchmark, Real World Engineering Bench (RWE-Bench), evaluates frontier AI models on mechanical tasks (parts and mechanisms modelled in CAD and checked for strength, fit and motion), electrical tasks (high-speed PCB routing, including DDR memory and PCIe links) and aerospace tasks (CFD and conjugate heat transfer, from building the mesh to a converged solution). Agents work in a sandbox with CadQuery, Gmsh, CalculiX and Python, and every run is graded by a deterministic verifier that runs physics solvers, with no LLM or human grading. Podcaster Dwarkesh Patel wrote that he is backing the company, founded by engineers Miguel and Carlos; the amount was not disclosed.Aero & CFDThermalElectromagnetics https://surrogatemodels.com/news/arena-physica-heaviside-microsoft-discoveryhttps://surrogatemodels.com/news/arena-physica-heaviside-microsoft-discoveryTue, 13 Oct 2026 09:00:00 GMTArena Physica announces that Heaviside, its foundation model for electromagnetism, is available on Microsoft Discovery. Heaviside-1, released in early September, has 350 million parameters and was trained on 250,000 3D field simulations; the company says it predicts the full 3D field response in milliseconds, roughly 100,000 times faster than a traditional solver and within a decibel of commercial accuracy. An agent on Discovery can switch between Heaviside's predictions and a full-fidelity solver for final validation. The terms of the collaboration are not given.Electromagnetics https://surrogatemodels.com/news/arena-physica-heaviside-1https://surrogatemodels.com/news/arena-physica-heaviside-1Tue, 01 Sep 2026 09:00:00 GMTArena Physica releases Heaviside-1, which takes 3D structures, material properties and excitations as input and predicts the full electric and magnetic vector fields at arbitrary points, not only S-parameters. It has 350M parameters, more than ten times Heaviside-0 from March, and was trained on 250,000 designs with 500 billion field samples. Arena reports that it runs 10,000 times faster than commercial solvers while staying within 1 dB, and that supervising on fields reduced out-of-distribution S-parameter error from 0.99 dB to 0.53 dB. The company also introduces EMVal v1.0, a benchmark with S-parameter and near-field tracks, and a public version of its standard split.Electromagnetics https://surrogatemodels.com/news/nvidia-physicsnemo-agent-toolkithttps://surrogatemodels.com/news/nvidia-physicsnemo-agent-toolkitSun, 26 Jul 2026 09:00:00 GMTNVIDIA re-architected PhysicsNeMo into agent-friendly libraries within the NVIDIA Agent Toolkit, so AI agents can train and deploy physics models alongside simulation and design tools. Cadence, Siemens, Synopsys, Samsung, Keysight and TSMC were named among the companies using the toolkit, with Cadence reporting up to 20× faster multiphysics in packaging and PCB design.Electromagnetics https://surrogatemodels.com/market/quanscient-series-ahttps://surrogatemodels.com/news/quanscient-series-aTue, 26 May 2026 09:00:00 GMTFinland's Quanscient closed a €10 million Series A led by 55 North and B&C Group, with Maki.vc, Crowberry Capital, QAI Ventures and First Fellow Partners. Its code-driven cloud multiphysics platform targets hardware such as motors, antennas and fusion magnets, and the company says it is built to produce the volume of simulation data that machine-learning models need.ElectromagneticsEnergy & fusion https://surrogatemodels.com/news/arena-physica-heaviside-0-rf-studiohttps://surrogatemodels.com/news/arena-physica-heaviside-0-rf-studioTue, 31 Mar 2026 09:00:00 GMTArena Physica releases Atlas RF Studio Beta, a web sandbox for inverse design of RF components, with two models. Heaviside-0, a transformer, predicts S-parameters from a geometry; Marconi-0, a conditional diffusion model, generates a geometry from a target S-parameter response, and its candidates are ranked with Heaviside-0. Both were trained on 3 million simulated designs across 25 expert templates mixed with random structures, over 20 years of combined simulation time according to the company. Arena reports a magnitude weighted MAE well under 1 dB for Heaviside-0, and 13 milliseconds per board against roughly 4 minutes for a commercial field solver.Electromagnetics