Computing Wakepotentials

In order to compute wakepotentials, we have to perform a time domain computation with a line charge as excitation. The line charge travels with the velocity of light in z-direction. Since there is only one charge traveling in positive z-direction, we loose a symmetry plane. We change the borders of the computational volume to:

 -mesh
define(STPSZE, InnerRadius/15)
     spacing= STPSZE
     pxlow= -1.1*OuterRadius
     pylow= -1.1*OuterRadius
     pzlow = -(GapLength/2+TaperLength+9e-2)
     pxhigh= 0
     pyhigh= 0
     pzhigh=  (GapLength/2+TaperLength+9e-2)

For the linecharge, we have to specify its total charge, its length, and the (x,y)-position where it shall travel. We also have to say that we do not want to compute eigenvalues, but we want to perform a time domain computation. We specify that at the lower and upper z-planes absorbing boundary conditions shall be applied. In the section -time, we specify that we want to have saved the fields at 90 equidistant times between the time that the line charge has traveled 0.1 m and it has traveled 1 m.

We edit our inputfile, such that the end of it looks as:

 -eigenvalues
    solutions= 15
    estimation= 2e9   # the estimated highest frequency
#     doit

 -fdtd
    -lcharge
        charge= 1e-12
        sigma= 4*STPSZE
        xposition= 0, yposition= 0
        shigh= 1.5
        showdata= yes

    -ports
        name= beamlow , plane= zlow, modes= 3, npml= 40, doit
        name= beamhigh, plane= zhigh, modes= 3, npml= 40, doit

    -time
        firstsaved= 0.1/@clight
        lastsaved= 1/@clight
        distancesaved= 0.1/@clight

 -fdtd
    doit

The so edited inputfile can be found as "/usr/local/gd1/Tutorial-SRRC/doris05-wake.gdf". We start the computation by feeding gd1 the inputfile:

 gd1 < doris05-wake.gdf | tee out

The computation only takes some minutes, since we compute a short range wake. When the time domain iteration starts, gd1 detects that the specified wake path is tangential to two magnetic walls. gd1 spits out:

 ## I am iterating Yee's algorithm..
 ###################
 # wake-computation:
 # (x,y)-position of the line charge:
 # specified (x,y)-position : (    0.00000000    ,     0.00000000    )
 # used      (x,y)-position : (    0.00000000    ,     0.00000000    )
 # ix, iy         :                60,                60
 # min. distances :    0.00000000      ..    0.00000000
 ############ I am checking the beam-path..
 #-- charge travels at upper x-plane.
 #-- charge travels at upper y-plane.
 #########################
 # Wake computation:
 # Since the charge travels along one or two symmetry-planes,
 # only 25 % of the charge is considered traveling through
 # the computational volume.
 # The excited fields in the subvolume will be the same as if
 # you were computing without the symmetry planes.
 # The lossfactors as computed by the post-processor will be
 # the same also.
 #########################

The end of the output of gd1 (on a reasonably fast machine) is:

 timestep=      800, simulated time=    6.2198e-9  s
 wakepotentials are known up to s=    1.1353     m
 cpu time/sec: used:    64.11, since last call:     7.41, MFLOPs/s:   88.69
 Wall clock time:    71.00     s,                         MFLOPs/s:   80.08
 timestep=      900, simulated time=    6.9973e-9  s
 wakepotentials are known up to s=    1.3672     m
 cpu time/sec: used:    71.52, since last call:     7.41, MFLOPs/s:   89.45
 Wall clock time:    79.00     s,                         MFLOPs/s:   80.98
  The highest simulation time is reached .., I am stopping
 ################################
 # cpu-seconds for FDTD :       75
 # start date           : 30/11/2002
 # end date             : 30/11/2002
 # start time           : 14:00:07
 # end time             : 14:02:14
 ## This is the normal end. Don't worry.
 ## Start the postprocessor to look at the results.
 stop FDTDLoop