reduced_beam_characteristics.0.0 file dispersion functions

[OLuckyG]

Hello,
I have a question about the step file that ImpactX generates. In the step file, there are the dispersion coordinates. How are these calculated?

I have designed a ring where the initial position starts from a straight section where there shall be no dispersion, the first cell is: Drift-Solenoid - Drift kind of straight cell. I give my own generated distribution using:

import amrex.space3d as amr
import pandas as pd
from impactx import Config, ImpactX, elements,distribution,twiss
file = pd.read_csv("input_impact.dat",sep='\s+',skiprows=1).values

xt = file[:,0]
xpt = file[:,1]
yt = file[:,2]
ypt = file[:,3]
Tt = file[:,4]
Tpt = file[:,5]
qm_eev = 1.0 / 938.0/ 1e6  # proton charge/mass in e / eV

###================ This piece of code works as input own distribution

dx_podv = amr.PODVector_real_std()
dy_podv = amr.PODVector_real_std()
dt_podv = amr.PODVector_real_std()
dpx_podv = amr.PODVector_real_std()
dpy_podv = amr.PODVector_real_std()
dpt_podv = amr.PODVector_real_std()

for p_dx in xt:
     dx_podv.push_back(p_dx)
for p_dy in yt:
     dy_podv.push_back(p_dy)
for p_dt in Tt:
     dt_podv.push_back(p_dt)
for p_dpx in xpt:
     dpx_podv.push_back(p_dpx)
for p_dpy in ypt:
     dpy_podv.push_back(p_dpy)
for p_dpt in Tpt:
     dpt_podv.push_back(p_dpt)

 pc.add_n_particles(dx_podv,dy_podv,dt_podv,dpx_podv,dpy_podv,dpt_podv,qm_eev,bunch_charge_C)

When I check the step file, the zeroth step, where it is at the beginning of the cell, the dispersions in x and y reads: (0.2037306687794294,-0.2772598763007251) I have tried using the distribution function generation defined with 6D gaussian using the **twiss helper function, and it give the same result. I realized in the manual, for defining the Gaussian distribution from $\lambda$ coordinates, there are dispersion parameters.

I wonder if this is due to the way I defined the coordinates and need some more parameters to define or there is some other reason?

The distribution I generate is a specific coupled distirbution in $x-y$ plane, however I see the same effect while using the incode distribution generation as well:

distr = distribution.Gaussian(
    **twiss(
        beta_x=10.0,
        beta_y=10.0,
        beta_t=60000.0,
        emitt_x=0.5e-6,
        emitt_y=0.5e-6,
        emitt_t=5e-05,
        alpha_x=0.0,
        alpha_y=0.0,
        alpha_t=0.0,
    ),
)
sim.add_particles(bunch_charge_C, distr, 40000)

I wonder why the "dispersion" functions in the step file gives non-zero entries for a straight cell at the beginning of the ring...

Thanks in advance!

[ax3l]

Hi @OLuckyG,

Sorry for the delay, there was some travel and vacation on my end.

The values are calculated from particles and dispersion calculations tend to be non-converged and noisy with low particle numbers. Do your values get better when you add more particles and converge to expected values?

Their calculation is done here:
https://github.com/BLAST-ImpactX/impactx/blob/25.09/src/diagnostics/ReducedBeamCharacteristics.cpp#L293-L297