changelog.md

March 11, 2026 · View on GitHub

New features:

Support for complex input beams, on either a Hermite-Gauss or Laguerre-Gauss basis with frequency-dependent beamwaist, rotation, 3D offset and angle.
offsetAperture() sequence function for including a non-centered iris
Free code signing on Windows provided by SignPath.io, certificate by SignPath Foundation

Bug fixes:

Will load the last crystal in the database even if there isn't a newline at the end.
Circular polarization settings are explained in the tooltip, and values of outside of -1..1 are no longer passed to the simulation with odd results.
2026.1.1 update: fix glitches when rendering beam view, and TEM00 always replaces previous beam
2026.1.2 update: fix x-angle in beam specification
2026.1.3: further fixes for beam view glitches for long renders
2026.1.4: fix addPulse sequence function
2026.1.5: fix loading of waveforms

Programming things:

The CPU FFT library was switched to PocketFFT. This leads to much better performance on the AMD systems I tested; slightly worse on Intel. However, it also makes maintenance and compiling much easier.

Bug fixes:

Fix a bug where setting a non-colinear angle would cause pulse injection to fail in FDTD simulations
Fix a bug where a failed load of a custom FDTD grid wouldn't cause the simulation to throw an exception
The Windows version is now built by a Github action for better reproducibility
The Intel Mac version is deprecated since Apple stopped updates for them. If someone still wants a build, tell me. That means that to use it on Mac, you should follow the compilation directions.

Bug fixes + some crystals:

Bug fix release:

Added new beam visualization mode, where you can make color images of the light field that results from the simulation
Add new controls for the appearance of plots, which you can access by clicking the ↔️ button on the bottom of the window.
Fixed a bug where a FROG file loaded for pulse 2 wouldn't be used.
Fixed a bug where very large grids whose dimensions were divisible by 4 but not by 8 would look weird.
The LightwaveExplorer python module is now on pip! i.e. you can run "pip install LightwaveExplorer" to get the latest version.

Fixed a bug introduced in 2025.0 where on CUDA the plasma would not affect fields in the y-direction
Fixed a bug where cancelling a batch job caused the subsequent batch to cancel itself
Axes and scales are now shown on the space and momentum images, and they're also saved when an SVG is produced.
Plot font size scaling works better with high-DPI displays
Refactored device structures to more efficiently and consistently allocate memory
NOTE: to read the files produced with 2025.0 and later in python, you need to update the LightwaveExplorer module (now available on pip!), many thanks to Florian Lindinger for the improved version that works with new and old files!

FDTD mode is more accessible: besides the options for crystals in the Propagation pulldown menu, you can also load a custom grid containing different nonlinear materials at different positions using the fdtd() function
FDTD reflection mode for nonlinear reflection from a surface, using the fdtdReflection() sequence command.
Improved convergence for the nonlinear absorption/plasma response. Note that the changes I made here will re-scale the values of the nonlinear absorption parameter (now they are consistent with what was in the paper). The conversion factor for older values is $\left(\frac{1}{2\pi\epsilon_0}\right)^{\frac{1}{N_p-1}}$ where $N_p$ is the number of photons in the multiphoton transition (divide the old value by this number).