Q about about orientation average
Junum
I am planning to simulate scattering properties of atmospheric ice
crystals.
They are usually arbitrary shapes, so I am making their structure for
ADDA.
What I want to calculate is scattering phase function (Mueller
matrix), especially for intensity and linear polarization, Qext, Qabs,
single-scattering albedo, and asymmetry parameter (g).
Since an asymmetry parameter is not allowed with orientation average,
I need to calculate it from P11.
For random orientation or orientation average, I want 1152
orientations as it was used in Penttila et al (2007).
Penttila et al. (2007), "Comparison between discrete dipole
implementations and exact techniques".
16 scattering planes, 9 beta, and 8 gamma orientations were chose for
1152 orientations.
My question is about input to reproduce this orientations.
My target scatterer is arbitrary shaped and not axisymmetric.
1. Are 16 scattering planes 16 alpha angles?
2. Is following avg_params.dat ok for 1152 orientations?
alpha:
min=0
max=360
Jmin=2
Jmax=4
eps=0
equiv=true
periodic=true
beta:
min=0
max=360
Jmin=2
Jmax=3
eps=1e-3
equiv=false
periodic=false
gamma:
min=0
max=360
Jmin=2
Jmax=3
eps=1e-3
equiv=false
periodic=false
3. Could you let me know any example avg_params.dat for irregular non-
axisymmetric shaped particle with random orientations?
4. Since Qext, Qabs, and Mueller matrix are my interests, I think that
alldir_params.dat and scat_params.dat are also needed to modified
corresponding to changes in avg_params.dat. Is it correct?
5. Do you have any recommendation for ADDA simulations for irregular
non-axisymmetric shaped particle with random orientations?
Thank you very much.
Sincerely,
Jun
Maxim Yurkin
> 2. Is following avg_params.dat ok for 1152 orientations?
>
> alpha: min=0 max=360 Jmin=2 Jmax=4 eps=0 equiv=true periodic=true
>
> beta:
>
> min=0 max=360 Jmin=2 Jmax=3 eps=1e-3 equiv=false periodic=false
>
>
> gamma:
>
> min=0 max=360 Jmin=2 Jmax=3 eps=1e-3 equiv=false periodic=false
Yes, correct.
> 3. Could you let me know any example avg_params.dat for irregular non- axisymmetric shaped particle with random
> orientations?
The default file, provided in input/ folder is a nice example. The one above is obtained by decreasing the values of
Jmax (to decrease the simulation time). Please look at the obtained estimates of the relative errors (similar as we
discussed for alldir_params.dat) to decide whether such values of Jmax are sufficient.
But if you have in mind a certain value of the final accuracy - you can type this value into eps field. Then ADDA would
reach the required eps by using as small Jmax as possible (this adaptability does not apply to alldir_params.dat).
Moreover, I recommend using alpha_Jmax at least 5 since calculations for different scattering planes are relatively cheap.
> 4. Since Qext, Qabs, and Mueller matrix are my interests, I think that alldir_params.dat and scat_params.dat are also
> needed to modified corresponding to changes in avg_params.dat. Is it correct?
No, when doing orientation averaging only the basic scattering quantities are calculated: cross sections and Mueller
matrix in one plane. So other *.dat files are irrelevant. The only thing you can do is to specify number of angles for
Mueller matrix (from 0 to 180 degrees) by command line option '-ntheta ...'. The value that ADDA choses by itself is not
always adequate.
> 5. Do you have any recommendation for ADDA simulations for irregular non-axisymmetric shaped particle with random
> orientations?
My personal experience with such scattering problems is very limited. So, apart from issues already mentioned above, I
can recommend you to test different Jmax for a typical particle (from the set that you plan to simulate) to identify the
optimal values.
Maxim.
Junum
Thank you very much Maxim.
By the way, one thing from your answer is not clear.
You said that when doing orientation averaging only the basic
matrix in one plane.
So, for above mentioned 1152 (16*9*8) orientations, calculated cross
section and Mueller matrix are actually averaged over 72 (9*8) angles?
Please let me know.
Sincerely,
Jun
Maxim Yurkin
After orientation averaging, the Mueller matrix does not depend on phi (azimuthal scattering angle), hence it need to be
calculated only in a single plane (for any fixed phi).
16 scattering planes used for averaging is just a mathematical trick to simulate 16 different values of Euler angle
alpha (much faster than 16 independent simulations) - see manual for details. So in this case orientation averaging (for
all scattering quantities) is done over 1152 (unique) orientations.
Maxim.
Junum
Sincerely,
Jun