Inconsistency comparing to Mie for moderate x < 5

[Justin Koenig]

Hello Everyone,

this is my first time foraying into the DDA and am quite confused as to why my results are significantly different than traditional Mie-Calculation.

Here's my command for ADDA v1.4.0 run on a single CPU.

adda -eq_rad 0.175 -dpl 100 -m 1.0946 0.001  -lambda 0.637 -asym -Csca -eps 12 -jagged

This returns a asymmetry parameter of 0.51 and Q_sca of 0.035.

Comparing to the Mie-Solution using Mie-Calc (by Simuloptics) as well as comparing to https://omlc.org/cgi-bin/mie_angles.cgi?diameter=0.35&lambda_vac=0.637&nr_sphere=1.4575&ni_sphere=0&n_medium=1.333&n_angles=180&density=0.1,

g =  0.72078

Q_sca = 0.07043

Increasing -dpl from 100 to 200 does not seem to change much. Changing m to 1.4575 is closer (0.62), but still far away from the Mie-calculated g, not sure if i misunderstood how to calculate m (which I thought was the refractive index ratio of particle to medium)

I also receive significant errors / warnings, warning me of the non-convergence of the scattered field for the solid angle.

I'm not exactly sure what causes this significant difference. I would be most grateful for any information about what I overlooked.

Best

Justin Koenig

[Maxim Yurkin]

Dear Justin,

Please see
https://github.com/adda-team/adda/wiki/FAQ#how-to-calculate-scattering-by-a-particle-near-or-inside-the-infinite-plane-parallel-plane-substrate
.
In short, when doing simulations in dielectric host medium, you should divide both the particle refractive index and
the (vacuum) wavelength by the host refractive index. There was also some minor discrepancy in the refractive index (in
comparison with the reference). So the minimal example to reproduce your reference would be
```
adda -eq_rad 0.175 -dpl 100 -m 1.0933983 0 -lambda 0.477869 -asym -Csca
```
which results (v.1.5.0-alpha3 (2df9749) with default `alldir_params.dat`) in:
Qsca = 0.071017 (Mie - 0.071017)
g = 0.720748  (Mie - 0.7208)
- quite good agreement with your Mie reference. Note that I have taken the Mie values from your link (omlc), which are
somewhat different from the ones you provides (maybe due to slightly different values of refractive index).

If you want to account for absorption, you need to divide Im(m) by m_host as well. I assume, that you played with `-eps`
and `-jagged` trying to get better accuracy, but `-eps 12` would make no noticeable difference, while `-jagged` command
is incomplete as provided (lacks argument) and adding something like `-jagged 2` would only deteriorate accuracy.

Concerning integration-accuracy warnings, take a look at the produced files `log_int_*`. Csca and asym_z lead to very
good accuracies (at least 1e-11), while asym_y and asym_x have huge _relative_ uncertainty. However, the latter is
because the values themselves are zero due to symmetry, so there is nothing to worry about.

Maxim.

P.S. This answer has been forwarded to your e-mail address for your convenience. However, if you want to continue the
discussion please reply to the group's e-mail. You are also advised to check the corresponding discussion thread at
http://groups.google.com/group/adda-discuss and/or subscribe to it to receive automatic notifications, since other
answers/comments will not be necessarily forwarded to your e-mail.

[Justin Koenig]

Dear Mr. Yurkin,

thank you very much for the rapid response. I've since read the section in the manual (which I had only skimmed previously). I apologize for the confusion and appreciate the tips regarding -jagged, -eps, and the logs.

I have one more request which is only tangentially related to the scattering software itself. I am having trouble finding software to generate dense fractal aggregates (FracVAL for example struggles with generation of aggregates with D_f>2.5). I would like to generate a spherical particle with a certain porosity, essentially, where the pores have a defined pore size distribution (or just singular pore size). I would then voxelize the STL with a different software to define the dipole centers. If you have any experience or know any researchers with researchers in this field on how i may achieve this, or if this is worth pursuing, please advise.

Thank you once again for your time, and apologies for my confusions regarding such a basic problem.

Best

Justin Koenig

[Maxim Yurkin]

First, I invite other participants of the group to share their experience with generation of dense fractal aggregates.

Second, I am aware that ADDA have been used quite a lot for fractal aggregates, and even coauthored a couple of such
papers, but I was never involved in generating the corresponding shapes (do not even remember any software names).
Moreover, as far as I remember, standard applications have Df < 2.5 (down to 1.6 or 1.8), so the corresponding
generators are not necessarily applicable to higher Df.

Third, if you need to make porous spheres, you may try built-in option `-granul ...` in ADDA, see Section 6.5 "Granule
generator" of the manual. It is neither fractal nor aggregate (just a random set of spheres), but it may reach rather
large volume fraction.

Maxim.