GSvit documentation

open source FDTD solver with GPU support

User Tools

Site Tools


docs:become

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
docs:become [2020/04/23 08:21]
pklapetek
docs:become [2020/04/24 12:27] (current)
pklapetek
Line 40: Line 40:
 {{:​docs:​methods_explanation_humanized.png?​600|}} {{:​docs:​methods_explanation_humanized.png?​600|}}
  
-//Technical remarks: To calculate single aperture or three apertures in one computational domain correctly it is important to add the side boundaries. It is also crucial to use CPML throughout all the+//Technical remarks: To calculate single aperture or three apertures in one computational domain correctly it is important to add the side boundaries. It is also crucial to use the best available absorbing boundary condition (CPMLthroughout all the
 calculations,​ otherwise we get everything dependent on the computational volume. calculations,​ otherwise we get everything dependent on the computational volume.
-If we manually setup the few apertures ​grating, result corresponds to the theory. +If we manually setup the grating with few apertures, result corresponds to the theory. 
-However, ​using periodic boundary conditions and virtual repetitions ​are valid only if we sum many virtual repetitions for far field calculation,​ at least about 7. +However, ​if we use periodic boundary conditions and virtual repetitions ​NFFF, this works only if we sum many virtual repetitions for far field calculation,​ at least about 7. 
-Using this approach for only few apertures (e.g. 3) leads to wrong result (even if it still looks fine at the diffraction order angle, is it only by chance?)+If we would use this approach for only few apertures (e.g. 3) it would lead to wrong result (even if it still looks fine at the diffraction order angle, is it only by chance?). 
-On the other hand, summing virtual repetitions without periodic BCs does not lead to anything reasonable - we have to use periodic BCs, to get the +When higher numbers of apertures are evaluated ​in this way (e.g. 11), the values of the central maximum can drop down and using a smaller time step helps to correct this effect.
-correct fields in the near field+
-When higher numbers of apertures are evaluated (e.g. 11), the values of the central maximum can drop and using a smaller time step helps to correct this effect.+
 This might be related to the far field integration (linear interpolation works better when signal is not changing so rapidly), ​ This might be related to the far field integration (linear interpolation works better when signal is not changing so rapidly), ​
 however it is still unclear if this is the only effect. The backup files for this calculation are {{ :​docs:​2d_transmission_aperture_backup_files.tar.gz |here}}.// however it is still unclear if this is the only effect. The backup files for this calculation are {{ :​docs:​2d_transmission_aperture_backup_files.tar.gz |here}}.//
  
 Accuracy aspects (evaluated on the single aperture transmission): ​ Accuracy aspects (evaluated on the single aperture transmission): ​
-   * effect of time step is small, prolonging it from dt=0.5 Courant limit to 0.25 does not change the shape of the curve in a detectable way. +   * effect of time step is small, prolonging it from dt=0.5 ​of Courant limit to 0.25 does not change the shape of the curve in a detectable way. 
-   * effect of absorbing boundary type smaller ​or distance to the boundary much smaller ​(mostly invisible, below percent?). The transmission around zero order is larger slightly when Liao condition is used instead of CPML.  +   * effect of absorbing boundary type or distance to the boundary much is not very large (mostly invisible, below percent?). The transmission around zero order is larger slightly when Liao condition is used instead of CPML and CPML is recommended to prevent impact of resonances in the computational domain in longer calculations.  
-   * effect of NFFF box size is up to 6 percent for the maximum difference of intensities at the first diffraction order (NFFF box sizes from 100x20 to 340x130, when y size is evaluated from the aperture plane and even for maximum NFFF the boundary in x is about 200 voxels far). Smaller NFFF box seems to be better; when it is enlarged artefacts can be seen. There is always a difference between analytical model that does not go to zero for maximum angles, and numerical model that goes to almost zero.+   * effect of NFFF box size is up to 6 percent for the maximum difference of intensities at the first diffraction order (NFFF box sizes from 100x20 to 340x130, when y size is evaluated from the aperture plane and even for maximum NFFF the boundary in x is about 200 voxels far). Smaller NFFF box seems to be better; when it is enlargedartefacts can be seen.  
 +   ​* ​There is always a slight ​difference ​for the single aperture calculations ​between analytical model that does not go to zero for maximum angles, and numerical model that goes to almost zero.
  
 ---- ----
Line 127: Line 126:
  
 The voxel size could impact the overall performance in more general way than only to wrong The voxel size could impact the overall performance in more general way than only to wrong
-reflectance. The very preliminary tests (with less accurate boundaries and other imperfections) however did not show large dependence on the voxel size, at least for reasonable voxel sizes, in the range of 3-12 nm. However, this could be still studied ​with the present setup+reflectance. The very preliminary tests (with less accurate boundaries and other imperfections) however did not show large dependence on the voxel size, at least for reasonable voxel sizes, in the range of 3-12 nm. Also, scaling the present model twice (using 5 nm instead of 10 nm voxel spacing and doubling all voxel distances), leads to the diffraction maximum of 0.142, which is almost negligible increase. Similarly, voxel spacing of 3.33 nm leads to diffraction maximum of 0.143 at already much higher computational costs. However, this could be still studied ​more in detail
  
  
docs/become.1587622897.txt.gz · Last modified: 2020/04/23 08:21 by pklapetek