Hello Everyone,

I have modelled a group of nanospheres grouped together almost in a straight line. Can I model the scattering properties such that results obtained from the first nanosphere would become an input for the second nanosphere and so on for the whole network?

Thanks,

Hirak

in discussion DDSCAT / General discussion » How to choose number of cores?

Hi, I've compiled the ddscat with MPI enabled. When I run it, only one core's usage reaches 100% & rest of the cores usage is around 10%. In case of OpenMP, I use the OMP_NUM_THREADS variable to fix the thread number. However, in case of MPI, I don't know how to choose the number of cores.

I'm running the code in Linux Mint platform. The computer has 16GB RAM, 8 cores & I also installed 'OpenMPI.' Please kindly help me.

Thanks

in discussion DDSCAT / General discussion » Error in code when running for CONELLIPS target.

in our case, shell thickness is 0.5d and number of dipoles ~ 4840 dipoles. We took the d ~ 2 , when setting the physical size i.e. N*d^3 = (4*pi/3)aeff^3. after putting the N = 4840 dipoles and d ~ 2 and the we find out the affective radius i.e. ~ 20 nm. So in case shell thickness will be 0.5*2 = 1 nm.

Sir, am i right ?

Thanking You.

in discussion DDSCAT / General discussion » Error in code when running for CONELLIPS target.

The ddscat.par parameters 21 21 21 20 20 20 specify a target with outer sphere diameter 21d and inner diameter 20d (*not* 21nm and 20nm unless you happen to have d=1nm). A sphere with diameter 21d has N approx (4*pi/3)*(10.5)^3 approx 4800 dipoles. The shell thickness would be 10.5d-10d=0.5d.

The effective radius aeff (in physical units) is specified on a separate line of ddscat.par . The physical size of

d is then fixed by N*d^3=(4*pi/3)aeff^3 , where N is the total number of non-vacuum dipoles.

Please consult the UserGuide for further details.

in discussion DDSCAT / General discussion » Error in code when running for CONELLIPS target.

Dear Sir,

I have one more problem regarding the core-shell calculations, when we define the outer and inner size e.g. 21 21 21 20 20 20, it means that, there is 1 nm shell thickness with core size 20 nm. while we defining such parameter, these parameters also defining the dipole, So for this condition, the number of dipoles must be 10 ^ 4. But in our above considered case, the dipoles are very less ( 4 x 10 ^ 3) . therefore, i want to know that how we can define the relevance dipoles and effective size (radius) ?

I am waiting yours fruitful discussion.

I wish you ll reply as soon as possible.

Thanking You

in discussion DDSCAT / General discussion » Error in code when running for CONELLIPS target.

1. specifiying the outer (70d) and inner (30d) diameters obviously determines the thickness of the shell (20d).

2. if ddscat terminated without writing Q values into qtable, then it didn't terminate normally.

I suggest you examine whatever was written to "standard output" to determine the reason for the code terminated.

in discussion DDSCAT / General discussion » Error in code when running for CONELLIPS target.

Daer Sir,

I have one more problem regarding the core-shell calculations, when we define the outer and inner size e.g. 21 21 21 20 20 20, it means that, there is 1 nm shell thickness with core size 20 nm. while we defining such parameter, these parameters also defining the dipole, So for this condition, the number of dipoles must be 10 ^ 4. But in our above considered case, the dipoles are very less ( 4 x 10 ^ 3) . therefore, i want to know that how we can define the relevance dipoles ?

I am waiting yours fruitful discussion.

I wish you ll reply as soon as possible.

Thanking You

in discussion DDSCAT / General discussion » Error in code when running for CONELLIPS target.

Dear Sir,

I ran the calculation for Q_scat values for conellips using 70 70 70 30 30 30 shape parameters. The calculation executed with normal termination but no Q_scat or Q_ext values were shown in qtable file.

Moreoever, I am not much clear about the shape parameters. I mean, in my calculations, how to decide the thickness of core and mantle in conellips?

Thanks in advance,

Please consult the UserGuide.

The index shows 2 entries for "effective radius" — pages 9 and 24.

a_{eff}= effective radius is defined by equation 4 on page 9.

How to calculate the effective radii of a spherical particle (let say 1 nm)? I have checked in the manual but still have doubt on the calculations. can some re post the answer to this?

in discussion DDSCAT / General discussion » Regarding the Near-field Calculations

As explained in the UserGuide, wxxxryyykzzz.E1 and wxxxryyykzzz.E2 each contain the electric fields

E_sca and E_inc at each point j in the computational volume, where the total E=E_sca+E_inc,

for incident polarizations 1 and 2.

The files wxxxryyykzzz.En are intended to be read by the program DDPOSTPROCESS.f90.

If you compile and run this on the same architecture that was used to run DDSCAT, DDPOSTPROCESS should be able to read the binary files wxxxryyykzzz.En created by DDSCAT.

Please see section 30 of the UserGuide for instructions on using DDPOSTPROCESS.

The user needs to create a file ddpostprocess.par to tell DDPOSTPROCESS what output is desired.

An example of ddpostprocess.par (as well as output files) can be found in examples_exp/ELLIPSOID_NEARFIELD

in discussion DDSCAT / General discussion » Regarding the Near-field Calculations

Dear Sir

I am doing the near-field calculations, as given in user guide of DDSCAT 7.3 on 29, 29.1 & 29.2 sections. when, we calculate the near-field calculation after choosing important parameter i.e. 1 = NRFLD (=0 to skip nearfield calc., =1 for nearfield E, =2 for E and B) and 2 = IORTH (=1 to do only pol. state e01; =2 to also do orth. pol. state), and 1 = IWRKSC (=0 to suppress, =1 to write ".sca" file for each target orient.; After doing these changing in ddscat.par file and run in the ddscat. we got the extra file like w000r000k000.E1 & w000r000k000.E2 and so on….w000r000k000.E200. Till now, i followed these steps.

Now, my Question is how to use .E2, .E2, .E3 and so on .E200 files for further calculation ?

Q1. Is it single point calculation.

Q2. If it is not single point calculation, then how to add these file into E1+E2…….=En

Q3. is it run directly through ddpostprocess or we have to make ddpostprocess.par file. if yes, then How to make and which parameter are use in ddpostprocess.par file. and how to run after preparing the ddpostprocess.par file.

I wish you will reply soon.

Thanking You

in discussion DDSCAT / General discussion » openmp seems do not work for ddscat

OpenMP does appear to be working. However, only some easily-parallelized parts of DDSCAT have been set up to use OpenMP directives; part of the computation is still done in serial mode. As a result, the OpenMP speedup is typically only factors of 2 or 3. This seems consistent with your report of ~200% cpu usage.

One probably doesn't gain much from using more than 4 OpenMP threads. In my own calculations I typically set OMP_NUM_THREADS=4 .

In general, users are strongly encouraged to use the current release: DDSCAT 7.3.2

in discussion DDSCAT / General discussion » openmp seems do not work for ddscat

I am testing DDSCAT7.3.0 on a linux machine. The compilation tools are: gcc5.4.0/gfortran, mpich-2.1.4. And the Openmp is enabled in Makefile by setting : "DOMP=-Dopenmp" and "OPENMP=-fopenmp". The compilation process is successful.

The testing problem is the scattering problem of a sphere: the geometry parameters of the target is

'ELLIPSOID'

80 80 80

Then I set OMP_NUM_THREADS=12 and run the executable ddscat.

Here is the problem: it seems the program is not running in openmp mode. The command "top d1" shows only about 200%, instead of 1200%, of CPU is used. No matter what value is given to OMP_NUM_THREADS, the usage of CPU is always about 200%.

Did I miss anything in compiling, or running the program ?

Thank you very much!

correction: SHPAR_1, SHPAR_2, SHPAR_3=(A/d, B/d, C/d) where d=dipole spacing.

The effective radius is *defined* to be the radius of an equal volume sphere (see eq. 4 of the UserGuide). An ellipsoid with "diameters" A,B,C along the three principal axes has volume V=(pi/6)*A*B*C and a_eff=(3*V/4*pi)^{1/3} = 0.5*(A*B*C)^{1/3}.

The physical value of a_eff is specified in ddscat.par.

The shape of the ellipsoid, and the number of dipoles that will be used, is determined when you set parameters SHPAR_1, SHPAR_2, SHPAR_3 (=A,B,C)

Respected Prof. Draine,

Thank you for your elaborated answer. Now I can see it and can be able to play with it.

Wishing Merry Christmas and Happy New Year, 2018 to you and family.

Thanks and regards,

Hirak

in discussion DDSCAT / Report errors » Help with orientation issues

respected sir, how we calculate effective radii in case of ellipsoid. pls help

in discussion DDSCAT / General discussion » Error in code when running for CONELLIPS target.

sir,

I am using DDSCAT 7.3 code in LINUX and facing problem for calculating absorption cross section for core- shell nanorods. Please help me how i calculate effective radii , and also tell how shape parameter and effective radii are correlated with each other.

in discussion DDSCAT / General discussion » How to fix lattice spacing 'd' value?

Hello Dr. Draine,

Thanks for the answers & clear explanation.