I have a question about the multiparticle distribution settings in TraceWin, especially regarding the options for Input distribution type and Mask.
In the Input distribution type dropdown, there are several options such as:
(1)4 dimensional ellipse
(2)2 phase plane ellipses
(3)Kapchinsky-Vladimirsky transverse ellipsoid
(4)4 dimensional transverse hyperspace
(5)Real space (x,y) & (x’,y’) generated randomly in each phase space
(6)PARMILA distribution types
(7)and “Use distribution from input particle file (*.dst)”
Then, for each distribution, we can also choose a Mask such as:
(1)Uniform
(2)3~6 sigma Gaussian
(3)Conical
(4)Parabolic
and set the Max size / rms size ratio.
My questions are:
(1)What are the practical differences between these distribution types in terms of beam physics?
(2)How does the choice of “Mask” (e.g., Parabolic vs Gaussian vs Uniform) affect the resulting beam properties?
In my simulations, I found that the beam envelope and Twiss evolution do not change much between these options. Does that mean the choice mainly affects the internal distribution shape (e.g. halo or density profile) rather than the global envelope?
So I wanna know is there a recommended or faster way to choose between them for general matching and optimization tasks?
I’m mainly performing multiparticle simulations for a non-periodic beamline, so I want to understand which combinations of these options are physically meaningful versus just cosmetic differences in initial sampling.
Thanks in advance for your clarification!
How to choose “Input distribution type” and “Mask” for multiparticle simulations? [SOLVED]
Re: How to choose “Input distribution type” and “Mask” for multiparticle simulations? [SOLVED]
Dear Hunry,
Just a remark before answering: you cannot set the maximum size/RMS size ratio; it is only an indication given by the code corresponding to the selected distribution type.
(1) It's really not up to me to answer this question, it depends on what you think is most representative of your physical case. I can only tell you that a KV-type distribution should give similar results in terms of transport and space-charge as a first-order matrix transport.
(2) Same answer for marks. Now whatever the type of distribution or mask chosen, the particles distributon built as input beam will always have the rms value and the Twiss parameters defined in the "Main" page and if your beam transport is not too non-linear and does not produce an increase in emittance, effectively you would not see much difference in the plots, since it is the rms parameters that are plotted (multiplied by a constant factor to give total sizes, usually 2 or sqrt(5)).
To answer your last question, in general, unless your line is very non-linear, optimisation and design work shouldn't be done in tracking mode, but in 1st order matrix envelope mode, which is much faster and often sufficient. Once that's done, we do a beam simulation in tracking, usually with a Gaussian beam, and see if we need to redo some of the optimisations in tracking mode.
Regards,
Didier
Just a remark before answering: you cannot set the maximum size/RMS size ratio; it is only an indication given by the code corresponding to the selected distribution type.
(1) It's really not up to me to answer this question, it depends on what you think is most representative of your physical case. I can only tell you that a KV-type distribution should give similar results in terms of transport and space-charge as a first-order matrix transport.
(2) Same answer for marks. Now whatever the type of distribution or mask chosen, the particles distributon built as input beam will always have the rms value and the Twiss parameters defined in the "Main" page and if your beam transport is not too non-linear and does not produce an increase in emittance, effectively you would not see much difference in the plots, since it is the rms parameters that are plotted (multiplied by a constant factor to give total sizes, usually 2 or sqrt(5)).
To answer your last question, in general, unless your line is very non-linear, optimisation and design work shouldn't be done in tracking mode, but in 1st order matrix envelope mode, which is much faster and often sufficient. Once that's done, we do a beam simulation in tracking, usually with a Gaussian beam, and see if we need to redo some of the optimisations in tracking mode.
Regards,
Didier