The OPAL Discussion Forum

[Simon Friederich <sifriede AT uni-mainz.de>]

Hi Christof and Andreas,

My latest approach was exactly how you've described it Christof, with the exception that I use also a normal distribution in the longitudinal case too. As I export the CST distribution right after the acceleration region I guess this is still a good approximation.

So these are the steps that I use in python (numpy + pandas):

1. read the exported CST distribution

    distr_df = pd.read_csv(cst_distribution, delim_whitespace=True, names=cst_cols, comment="%")

2. use the macrocharges as weights to calculate the covariance matrix:

    distr_df['w'] = distr_df.macrocharge / (-1.602e-19) 
    cov = np.cov(distr_df[selected_cols], aweights=distr_df.w, rowvar=False)

3. create a new distribution from a multivariate normale distribution using a given mean value (basically: (0,0,0,0,0,beta*gamma)) and the covariance matrix:

    distr_new = np.random.default_rng().multivariate_normal(mean, cov, size=int(N_opal))
    output_df = pd.DataFrame(data="distr_new," columns=['x', 'px', 'y', 'py', 'z', 'pz'])

I have some slight discrepancies if I look e.g. at the envelope in a drift space (i.a. I observe slightly different rms divergences) but for now I think is a good solution.

Best regards
Simon

Am 16.09.2020 um 17:54 schrieb Christof Metzger-Kraus:

Hi Simon,

couldn't you use the longitudinal charge distribution as a pdf and resample the long. particle distribution with it. In transverse direction you would then use e.g. a gaussian distribution. For the momenta of the particles you would use the moments of the distribution. 

I haven't done this myself though.

Christof 

Simon Friederich <sifriede AT uni-mainz.de> schrieb am Mo., 14. Sep. 2020, 10:40:

Dear all,

From CST I can export a particle distribution. In contrast to OPAL (and propably most of the other available simulation codes):
CST weights the macro particles by different macro charges instead of having the same macro-charge for all particles. The number of particles is (almost) equally distributed (slightly influenced by the simulation accuracy).

I am thinking of a way to convert the CST distribution to one, that can be used in OPAL (using FROMFILE).
Has anyone faced this task before and has a method/script that (s)he might share?

[I've looked into the python script by Bettina Kuske and Dieter Faulbaum (HZB), that converts an ASTRA distribution to OPAL, but it's unfortunately not suitable as ASTRA has a const. macro-charge for all particles, right?]

Thanks again in advance.

Best regards,
Simon

P.S.: I am sorry, that I've opened a second request mail, but being able to convert CST distributions to OPAL would circumvent the DC acceleration problem (see my last listing mail) for now.

-- 
-----------------------------------
Dr. Simon Friederich
Helmholtz-Institut Mainz
Kollaboration B/ACID
Johannes Gutenberg-Universität Mainz
Staudingerweg 18
55128 Mainz, Deutschland

Tel.: +49 (0)6131 39-23160
E-Mail: sifriede AT uni-mainz.de
----------------------------------------

-- 
-----------------------------------
Dr. Simon Friederich
Helmholtz-Institut Mainz
Kollaboration B/ACID
Johannes Gutenberg-Universität Mainz
Staudingerweg 18
55128 Mainz, Deutschland

Tel.: +49 (0)6131 39-23160
E-Mail: sifriede AT uni-mainz.de
----------------------------------------