Numerical stability and error analysis of transformation optics for electromagnetic simulation in time-domain

Jian Shiung Hong, Wei Ming Cheng, Ruei Cheng Shiu, Yung Chiang Lan, Kuan Ren Chen

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)


Applying transformation optics (TO) for converting non-uniform physical grids into uniform numerical grids with an additional derived anisotropic medium and an analysis of numerical stability and error in an electromagnetic finite-difference time-domain (FDTD) simulation are presented. Traditionally, the maximum time-step is limited by the smallest cell size. However, due to the invariance of time, when smaller physical grids are converted into a numerical system, the phase velocity of light can be greater than that in free space. The conventional Fourier analysis of numerical dispersion to seek numerical stability is no longer applied, but rather there is further consideration of the properties of the derived medium. Thus, we derive the numerical dispersion relation to seek a more general numerical stability condition for a converted numerical system. It is found to be consistent with that of a non-uniform physical system. Also, an error analysis is performed by comparing the differences in the electric fields simulated in the converted numerical system with those in the uniform physical system. We show that they are almost of the same order; that is, there is almost no extra error induced when applying the proposed numerical simulation method.

Original languageEnglish
Pages (from-to)1742-1748
Number of pages7
JournalJournal of Computational and Theoretical Nanoscience
Issue number8
Publication statusPublished - 2013 Aug 1

All Science Journal Classification (ASJC) codes

  • Chemistry(all)
  • Materials Science(all)
  • Condensed Matter Physics
  • Computational Mathematics
  • Electrical and Electronic Engineering


Dive into the research topics of 'Numerical stability and error analysis of transformation optics for electromagnetic simulation in time-domain'. Together they form a unique fingerprint.

Cite this