Evaluate for Free

2025 R2 Release Notes

Caution_symbol_selfmade.pngWarning: In Ansys Lumerical FDTD™ 2025 R2, explicitly saving a simulation file in analysis mode after running a GPU simulation causes monitors to lose results when you reload the file. See this Knowledge Base article on the issue for further information and workaround. This issue does not affect simulations run on CPU, or does it affect any previously released versions of the software.

Shared Lumerical Enhancements

  • Ansys Engineering Copilot: Ansys Engineering Copilot is now integrated into Ansys Lumerical FDTD™, Ansys Lumerical MODE™, Ansys Lumerical Multiphysics™ and Ansys Lumerical INTERCONNECT™. This new feature provides users with access to the AI-powered AnsysGPT™ assistant, which is trained with all learning and support contents (including KBs, examples, forums, webinars, Ansys blogs, and more) directly within the Lumerical products. The feature requires the GPT AI+ license increment, which is included for free with TECS maintenance. A new Ansys Engineering Copilot icon in the Learning and Support tab opens the application offering access to AnsysGPT, Support (ACSS), Courses, Knowledge, and Forums.
  • CAD objects import capacity: Ansys Lumerical FDTD™, MODE™ and Multiphysics™ now can import CAD objects from Parasolid (*.x_t, *.b_t), SolidWorks (*.sldprt), CATIA (*.CATPart), Autodesk Inventor (*.ipt, *.iam) and ACIS files (*.sab, *.sat).
  • Lumerical for Ansys Access / Ansys Gateway: The full Lumerical software suite, release 2025 R1.2, was released to Ansys Access and Ansys Gateway on May 14th, 2025.
  • Deprecated mesh refinement in Ansys Lumerical FDTDTM and Ansys Lumerical MODETM: Low-usage options (dielectric volume average, volume average, Yu-Mittra method 1 and Yu-Mittra method 2) are deprecated for new projects from 2025 R2 version because other current and future mesh options are more efficient. We now recommend conformal mesh refinement instead. Compatibility with older versions is maintained, the deprecated option previously chosen will be hidden in the UI unless in use (FDTD, varFDTD, RCWA, FDE, EME). The script command set(“mesh refinement”, “name_of_deprecated_mesh”); still support deprecated mesh options and now display a warning message. The warning also appears in the Simulation and Memory Report (FDTD, varFDTD, RCWA). Deprecated mesh refinement options are no longer supported with FDTD GPU.
  • Python Version Upgrade: Lumerical products have upgraded their integrated Python version to 3.13.1. 

Ansys Lumerical FDTD™

  • Multi-Node Multi-GPU acceleration: We support FDTD GPU simulations distributed amongst multiple GPUs across multiple host machines. The simulation is accelerated using all supported and available GPU devices on all host machines, even if each machine may have different numbers of GPU devices or different supported models. Significant speed up, with nearly linear scaling of solver speed versus number of compute nodes can be achieved for large simulations (benchmarked with up to 8 nodes of 2 H100 GPUs and InfiniBand network connection between nodes). This result makes multi-node multi-GPU acceleration attractive for simulating large structures such as Metalens, allowing simulations to complete in a fraction of time compared to running on a single node.
  • Inverse Design Performance Improvements: The inverse design optimization module, lumopt, now allow you to specify the expected spatial localization of a given parameter for shape-based optimization problems. This improvement allows parallelization of meshing when calculating how each parameter changes the permittivity and can significantly improve performance in problems with large number of parameters.
  • FDTD GPU Performance Improvement:
    • Time monitors: Time monitors have been optimized for FDTD GPU simulations. The new implementation buffers data on the GPU, eliminating the need for synchronization and data transfer to the CPU at every time step. This enhancement significantly reduces computational overhead and improves overall simulation efficiency.
  • RCWA Object & Solver enhancements
    • Layer Repetition feature improvement: The RCWA object has been enhanced to allow users to repeat grouped sets of layers at specific locations. Users can now define layer repetition by using start and end reference positions, similar to how periodicity is set with EME object. This new approach provides greater control over interface definition within the unit cell, allowing for more precise and flexible structure modeling in optical metrology and AR/VR applications.
    • Materials default to interpolation: Materials now default to using the interpolated model instead of the multi-coefficient model in RCWA.
  • Assembly groups support Import geometry: Assembly group can now handle imported surface (*.txt) object, binary material, (n,k) material and imported image.
  • Passive ports in Ansys Lumerical FDTDTM: It is possible to choose from script or the UI to have no source port, to do an FDTD simulation with all ports being passive ports. To select this through the UI, a new blank option is added.

Ansys Lumerical MODE™

  • Snap-to-PEC option activated by default for varFDTD: The Snap-to-PEC option in the varFDTD solver ensures that any structures defined as Perfect Electric Conductors (PEC) have their interfaces aligned with the Yee cell boundaries. This alignment guarantees that all electric field components at the PEC interface are tangential to the interface. This option is now selected by default in 2025 R2 release, like the other solvers FDTD since 2017, FDE or EME since 2016.

Associated files

Related articles