Frequently Asked Questions¶
What are the units used in the simulation?¶
We assume the following physical units:
Length: micron (μm, \(10^{-6}\) meters)
Time: Second (s)
Frequency: Hertz (Hz)
Electric conductivity: Siemens per micron (S/μm)
Thus the user should thus be careful for example to use the speed of light in μm/s when converting between wavelength and frequency.
Currently, only linear evolution is supported, and so the output fields have an arbitrary normalization proportional to the amplitude of the current sources, which is also in arbitrary units. In the API Reference, the units are explicitly stated where applicable.
How do I run a simulation and access the results?¶
Submitting and monitoring jobs, and donwloading the results, is all done
through our web API. After a successful run,
all data for all probes can be downloaded in a single file
probe_data.hdf5 using tidy3dclient.web.downloadResultsFile(), and the
raw data can be loaded and analyzed directly e.g. using h5py. The data
for each probe is stored in a separate group, with the following datasets:
E: E-field, if requested.H: H-field, if requested.xmesh: Mesh along x over which the fields were stored.xmesh: Mesh along y over which the fields were stored.xmesh: Mesh along z over which the fields were stored.(for a
tidy3dclient.TimeProbe)tmesh: Mesh in time over which the fields were stored.(for a
tidy3dclient.FreqProbe)freqs: List of frequencies over which the DFT was computed.
Each of E and H that was requested has the shape (3, Nx, Ny, Nz, Np),
where Np is either the number of time steps or the number of frequencies. Note
that the fields are interpolated to the center of the Yee grid cell.
The data file can also be loaded into the corresponding tidy3dclient.Simulation
object that was used for the run using tidy3dclient.Simulation.load_results(). The Simulation
object has as attributes a list of time and a list of frequency probes, Simulation.tprobes and
Simulation.fprobes, respectively, and each Probe in those lists has the list of datasets attributes
datasets above as attributes. Loading the results in this way also allows you to use our in-built
visualization functions.
How many layers of PML should I add in a given simulation?¶
Tidy3D uses a complex frequency-shifted formulation of the perfectly-matched layers (CPML), for which it is more natural to define the thickness as number of layers rather than as physical size. In many simulations, using e.g. 15 layers in a given direction will be sufficient. If the fields in a simulation diverge, the first thing to try is increasing the number of PML layers, followed by decreasing the Courant stability factor. Finally, some simulations might work better with a setting of the underlying CPML parameters that is different from the default. In a future release, we will also allow the (experienced) user to set these parameters.
Are dispersive and lossy materials supported?¶
Currently, dispersive material models are not supported, but this is high on our priority list. However, for a sufficiently narrow frequency of interest, lossy (i.e. conductive) materials can be included, either directly through a value for the real part of the relative permittivity \(\mathrm{Re}(\epsilon_r)\) and electric conductivity \(\sigma\), or through a real and imaginary part of the refractive index at a given frequency, \(n(\omega) = n + ik\). In the case of (almost) lossless dielectrics, the dispersion could be negligible in a broad frequency window, but generally, it is importat to keep in mind that such a formulation of the losses is best suited for single-frequency results. The relationship between the two equivalent models is