... electron's13.1
positrons are a rarity in the code, and so all references to electrons in this section should be taken to refer to electrons or positrons 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... travel13.2
EGS4 only updates the directional cosines of the electron at the end of each step, and this can give rise to a subtle artifact in which all the photons for one step are emitted on a cone oriented with respect to one direction. To overcome this problem, the electron's directional cosines are linearly interpolated between initial and final directions throughout the step. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... flat13.3
This is actually an approximation - the distribution is only flat if the refractive index is assumed to be constant, which is not quite true. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... boundary13.4
defined such that $\underline{\hat{a}}.\underline{\hat{n}}\ge 0$ for the purposes of the calculations in this section. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... SNNS18.1
Stuttgart Neural Network Simulator - contact Oxford for more information. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... zero.A.1
The 250 figure is the current best estimate of the optimal time window, in the final system this may be revised up or down slightly. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... delayA.2
Since the delays will not all be exactly the same, they will have to be measured and included in a calibration bank in the database. Currently SNOMAN sets the cable delay to zero, although this will change. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.