-fdtd/-time: minimum / maximum time, when to store

 ##############################################################################
 # Flags: nomenu, noprompt, nomessage,                                        #
 ##############################################################################
 # section -time                                                              #
 ##############################################################################
 # firstsaved    = undefined                                                  #
 # lastsaved     = undefined                                                  #
 # distancesaved = undefined                                                  #
 # tminimum      = undefined                                                  #
 # tmaximum      = undefined                                                  #
 # decaytime =       1.0e+30                                                  #
 ##############################################################################
 # amptresh =       3.0e-3                                                    #
 # dtsafety =       0.950                                                     #
 # ndt      = auto                                                            #
 # ___stopafter=       1.0e+30          -- force stop after that time.        #
 # ___evmax    = undefined              -- dangerous.                         #
 # ___dt       = undefined              -- dangerous.                         #
 ##############################################################################
 # return, help, ?                                                            #
 ##############################################################################

• firstsaved:
  The first time where the electric and magnetic fields shall be stored on files.
• lastsaved:
  The last time where the fields shall be stored on files.
• distancesaved:
  The time-distance between fields to be saved.
• tminimum= TMIN, tmaximum= TMAX:
  The minimum and maximum time to simulate. gd1 will simulate minimum a simulation time of TMIN and maximum one of TMAX.
  • If a broadband computation is performed: If after TMIN no power is flowing through any port, the simulation is stopped.
  • If a monochromatic computation is performed (risetime /= 0): If after TMIN are power flowing through all ports is stationary, the simulation is stopped.
  No matter what the status of the fields in the computational volume is after a simulation time of TMAX, the simulation will be stopped.
• amptresh= TRESH:
  The treshold value for exiting the FDTD-Loop when prforming a broadband computation: When all port-amplitudes have decayed down to TRESH times the maximum of all monitored modes somewhere in the time, the computation stops.
• dtsafety:
  The safety factor for the time step.
  gd1 computes the largest stable timestep as
  $dtmax= \sqrt{\frac{4} {''highest\; eigenvalue\; of\; the\; closed\; volume''}}$,
  This is normally reliable and gives the minimum number of timesteps required for a wanted simulation time. If for some reason you want to use a different timestep, you may change the timestep by changing dtsafety. The used timestep is
  dt = dtsafety * dtmax
• decaytime= OOALPHA:
  Specification of artificial losses in the computational volume.
  • OOALPHA:
    Time (in seconds) that the fields (both E and B) decay by a factor of 1/e.

  The reason for this parameter is as follows: single.gd1 calculates mostly in single precision, with a relative accuray of its internal values of $10^{-7}$ for most machines. With this machine accuracy, it is not possible to calculate the decay of high Q-structures. If you would fill your volume not with vacuum but with a dielectric with a very low conductivity $\kappa$, the program would attempt to scale in every timestep the electric field with a factor of $\frac{1- \frac{\Delta t \kappa}{2 \varepsilon}} {1+ \frac{\Delta t \kappa}{2 \varepsilon}}$. For low values of $\kappa$ this factor is so near to one that it cannot be represented in single precision.

  With the specification of decaytime, the fields (both E and B) in the entire volume are scaled by such a factor, that after a simulationtime of OOALPHA the fields have decayed by a factor of 1/e.

  If you have to calculate the scattering parameters of a resonator with a known (not too low) internal Q-value, e.g. greater than 1000, you can specify artificial losses with this keyword and gd1 or single.gd1 together with gd1.pp will calculate the proper scattering parameters.

• ndt:
  This enforces the used time step "dt" to be such that "dt" is smaller than the normally used timestep and that an integer multiple of "dt" gives the value of "ndt".
• ___stopafter:
  Enforces a stop after computing so far. Not very useful. I once needed it.
• ___evmax:
  If specified, ignores the result of the estimation of the highest eigenvalue of the curl curl matrix. The largest stable timestep is then computed not from that estimate but from a highest eigenvalue being ___evmax. NOT USEFUL. This is avaiable to test that the results of parallel and serial computations are bitwise identical.
• ___dt:
  If specified, uses that value as the timestep. NOT USEFUL.

Example The following specifies that we want to simulate minimum a time span of 100 periods of a frequency of 1 GHz. We are absolutely shure that after a timespan of 1000 periods the fields have decayed sufficiently, and we want to store the fields between 10 periods and 20 periods, in a distance of 1/2 period.

  define(FREQ, 1e9)
  tmin=  100/FREQ, tmax= 1000/FREQ
  firstsaved= 10/FREQ, lastsaved=  20/FREQ
  distance=   1/2/FREQ