marvinmoyco / fdtd-code

Codes for FDTD implementation for EEE 190/ECE 198

FDTD Codes for ECE 198

1-D FDTD

• FEATURES
  1. Multiple Boundary Conditions - Dirichlet Boundary Condition or Perfectly Absorbing Boundary Condition.
  2. Multiple Sources - Gaussian Pulse, Sine Wave, Square Wave, Modulated Sine Wave.
  3. Multiple Source Excitation Method - Hard Source, Soft Source, and Total Field/Scatter Field
  4. Basic FDTD Algorithm - Conventional algorithm (leap-frog method)
  5. FDTD-Schwarz Method - Combined algorithm of FDTD-Schwarz in serial and parallel mode (using OpenMP).

Notes:

• Libraries used are header-only type so no need to compile them. Just include the directories of the necessary libraries in the makefile before compiling.
• Main code uses C++ for generating simulation data but data visualization uses Python 3 (matplotlib/plotly)

Current Issues:

FDTD

1. Sinusoidal source propagate backwards (towards the left) when using TF/SF.

FDTD-Schwarz Method

1. Sinusoidal source propagate backwards when using TF/SF.
2. Algorithm limited up to 8 subdomains per simulation only because it causes CPU bottleneck if increased to more than 8 subdomains.
3. Reflections generated in the internal boundary of each subdomain. This causes errors in the simulation where some energy is generated out of nowhere.

Libraries Used:

C++ Libraries:

• xtensor - Main scientific computation library in C++ (very similar to Numpy).

• xtl - Dependencies of all xtensor libraries.

• xtensor-io - For I/O operations like saving simulation data into HDF5 files.

  • HgihFive - C++ library for manipulating HDF5 files (header-only library)

    • libhdf5 - can be installed using sudo apt install libhdf5-dev

• xtensor-fftw - Used in a FFT operation in the simulation.

  • FFTW 3 Library - can be installed using sudo apt install libfftw3-dev

• xtensor-blas - Used in computing L2 norm (using linalg module)

  • OpenBlas - can be installed using sudo apt install libopenblas-dev

  • LAPACK - can be installed using sudo apt install liblapack-dev

Python Libraries:

• numpy - Main numerical library for linear algebra and scientific computing in Python.
• matplotlib - Basic plotting library based on MATLAB.
  • ffmpeg - Required library when using matplotlib for creating animation (video) of simulation
• plotly - Interactive graphics library for Python.