Replies: 3 comments
-
Hi @arka378, I don't fully understand the plots you've attached. Could you provide an example input deck, and tell me output you're expecting? Cheers, |
Beta Was this translation helpful? Give feedback.
-
Hey @Status-Mirror! I am expecting, for example, the The x-axis in the figures is inverse resolution. Moving to the left implies increasing resolution. An example input deck is attached in a zip. Best, |
Beta Was this translation helpful? Give feedback.
-
Hey Arka, The EPOCH field-solver calculates Maxwell's equations, which involve spatial derivatives of field quantities. When you increase the field order, you increase the number of cells used to calculate these gradients - hopefully smoothing out any local fluctuations. No corrections are made to the laser or outflow boundaries as far as I can see. With this in mind, I'm under the impression field-order is acting as a smoothing function. I'm not sure you should expect the 2x and 4x error reduction here. An increased field-order may help in some problems, but you may find it unhelpful in others. I should also note that I wasn't personally involved in coding up the higher-order Yee field-solvers. Cheers, |
Beta Was this translation helpful? Give feedback.
-
Hi,
I have been using EPOCH recently for my simulations but ran into some difficulty regarding the Maxwell solver.
I am using the 'yee' solver. To test this solver, I created a setup using the latest EPOCH (EPOCH2D version 4.19.3 (commit v4.19.3-0-g724471f-clean)) for a EM wave, polarized along y and travelling along x. This simulation has no particles. The grid is 2D, with dimensions of 1 wavelength in x and 10 wavelengths in y.
I evolved this setup for 1 waveperiod and measured the error in the electromagnetic fields, compared to the initial fields. I expected a secon-order behaviour from setting field_order=2, 4th order behaviour from setting field_order=4 and so on, by varying the resolution. But I get a strange behaviour, see figures attached.
For my simulation, I need the fields to be stable for about 2000 waveperiods. But with such behaviour with resolution, I cannot develop my simulation. How can I get better behaviour from the Maxwell solver, i.e better behaviour with resolution and higher accuracy over longer duration?
Beta Was this translation helpful? Give feedback.
All reactions