Lecture -- Finite-Difference Time-Domain in Electromagnetics

[@markolazar2149]

Wow! Amazing video!

[@empossible1577]

Thank you! You are watching a video for a face-to-face class that I teach in Computational Electromagnetics. I recommend accessing the materials through the course websites because you will always have links to the notes, latest versions of the videos, and other learning resources. Here is a link to the Computational Electromagnetics course website: empossible.net/academics/emp5337/ If you are interested in finite-difference time-domain (FDTD) specifically, here is a link to the course website for my face-to-face class. empossible.net/academics/emp5304/ If you are just starting out, you may struggle a bit with the face-to-face class material. To help, we have created an online course for FDTD specifically intended to teach to the complete beginner. The course teaches a very powerful form of FDTD and includes all MATLAB codes. Here is a link to a video showing what it is in the online course: kzbin.info/www/bejne/q3PMoaV_g52anLssi=Wgrq_-i10rjMVDZH Here is a link to the actual courses: empossible.thinkific.com/collections/FDTD-in-MATLAB We also just published a book through Artech House on the finite-difference frequency-domain (FDFD) method. This is the frequency-domain brother of FDTD. I personally think FDFD and FDTD are the best FIRST methods to learn. With these two methods you can simulate anything. The only thing other techniques will offer you are greater speed and efficiency. Here is a link to the book website: empossible.net/fdfdbook/ I hope this helps! Good luck and have fun!

[@IggyWiggySniggy]

Thank you for your videos, very detailed.

[@FrederikFalk21]

This is great!

[@empossible1577]

If you are interested in learning finite-difference time-domain (FDTD), we have a complete free course on the method here: empossible.net/academics/emp5304/ Unfortunately, these are the notes I use when I teach face-to-face so they do not include codes and may be difficult follow. If you want incredibly high-quality instruction along with line-by-line explanation of the code in MATLAB, checkout these online courses on FDTD: empossible.thinkific.com/collections?category=FDTD-in-MATLAB Here is a link to a video showcasing the contents of the 1D FDTD and 2D FDTD courses: kzbin.info/www/bejne/q3PMoaV_g52anLssi=nrbHirpBctXyqqJ8 Hope this helps!

[@muhammedhalil8175]

Thank you very much. Your explanations are very helpful.

[@empossible1577]

Great to hear!

[@firstlast-qy6xn]

I am currently watching your videos about FDTD on Yee grids, for numerical simulations of electromagnetic waves. The scenario I'm trying to simulate is the one of the pTx system in an MRI scanner, which consists of N dipole antennas parallel to each other, arranged around the piece of human tissue that is to be imaged. The dipole antennas send out periodic waves at a constant frequency - does this mean it is sufficient to do a single frequency domain simulation at exactly that frequency? Or is it better to do a time domain simulation (which is what I have been doing so far, on a collocated grid, with second order wave equations, which appears to be unstable and led me to search for another solution)

[@empossible1577]

It sounds like a fixed frequency will work for you, but I will admit I not that aware of how MRI scanners work. A collocated grid will definitely not work for you. I don't think it will cause the simulation to go unstable so that might be a mistake hiding in your code. A collocated grid will more cause material interfaces to reflect and transmit incorrectly. I am more curious about your PML when it comes to instability. The UPML I teach in the free course can be unstable if you place a PML along all boundaries. The CPML is much better. Of course, your instability could also be just a sign error in your code somewhere.

[@firstlast-qy6xn]

@@empossible1577 thank you for your reply! i will look into it further

[@firstlast-qy6xn]

@@empossible1577 i just realized that i had to click on "read more", and that i didnt read the whole reply yet whats happening in the simulation is that on the boundaries of the "human head", modeled as a sphere with some conductivity σ and permittivity ε, at some point, the field starts growing exponentially, but still switching polarity periodically (at double the period length of the source from some reason) I havent yet watched your videos on boundary conditions, I will definitely watch them soon

[@empossible1577]

@@firstlast-qy6xn Hmmm...It is easy to make a mistake in the update equations when incorporating conductivity. You might also have a Courant stability issue and need to choose a smaller time step. Let me point you to two resources. The first I think you are aware, but it is the website for my free FDTD course: empossible.net/emp5304/ The second is a paid FDTD course that is next level in terms of how it teaches FDTD. It also includes all of the MATLAB codes with a line by line explanation of the codes. empossible.thinkific.com/collections/FDTD-in-MATLAB

[@firstlast-qy6xn]

@@empossible1577 thank you very much!

[@benjiusofficial]

Huh, that's pretty interesting. What would be the preferred alternative to FDTD for resonant devices?

[@empossible1577]

Awesome question! It is hard to say what is the "best" alternative method because it depends so much on what you want to simulate and the physics you want to incorporate. I can say that the best first method to learn is finite-difference frequency-domain (FDFD). I think it is the easiest method to learn and implement that obtains a rigorous solution to Maxwell's equations. It is incredibly versatile and can simulate just about any type of device or structure. The drawback of FDFD is efficiency and you will feel that if you ever try to simulate very large 3D structures. For everything else, it is an awesome method. I know many methods in computational electromagnetics (CEM) and FDFD is what I use most frequency by a large margin. If you are interested in this method, I actually just published a book on FDFD. The book is intended for those completely new to CEM to learn the art and science, like boundary conditions, sources, post-processing, parameter sweeps, and much more. I teach the basic concepts and techniques of CEM through FDFD, but the concepts apply to all numerical methods. Here is a link to the book: empossible.net/fdfdbook/

[@hussiendaeeh]

Thank you very much

[@abhiramcd]

Iyyalkk kali ariya