For large mean multiplications, significant deviations are observed, especially at the low end high ends of the distribution. These can empirically be modeled with the Polya distribution which can be obtained by using POLYA-FIXED or POLYA-TOWNSEND
Although the exponential distribution has a non-zero probability at 0, the multiplication factor used in Garfield is always equal to at least 1: if a factor less than 1 is drawn, it is rejected and a new number is drawn.
The mean multiplication factor has to be strictly positive.
[Default: you have to specify the mean multiplication.]
The factor has to be larger than or equal to 1, the latter being equivalent to specifying NONE.
[Default: none, you have to specify the multiplication factor.]
The multiplications obtained with this option are constant for a given drift path. They may vary when the Monte Carlo drift line integration routines are used since these introduce variations in the drift path.
This option is accessible only if Townsend data has been entered in the gas section.
Please note that the 2nd argument is a relative, not an absolute, standard deviation. If you use for instance
avalanche gaussian 20000 0.5
then you'll end up with a normally distributed multiplication with a mean of 20000 and a sigma of 10000.
The Gaussian distribution has non-zero probability for factors less than 0. Since the multiplication factor as defined here has to be equal to 1 at least, Garfield will if needed repeatedly draw a random number until an acceptable number is found.
The mean is allowed to be negative, but the width must be strictly positive - a width of zero would be equivalent to using the FIXED option.
[Default: none, you have to specify both the mean and the relative standard deviation.]
The Polya distribution is a \Γ-distribution that matches reasonably well the fluctuations in a cylindrically symmetric amplification region. A physical interpretation of the parameter \θ is given in G.\ D.\ Alkhazov, NIM\ 89 (1970) 155-165.
When \θ is set to 0, an exponential distribution is obtained. If \θ\ <\ 0, the distribution is "concave", while for \θ\ >\ 0 the distribution assumes the more usual shape with a maximum.
The properties of this distribution can be examined using the RND_POLYA function, as shown in the following example:
Say "Please enter theta" Parse Terminal theta Call book_histogram(ref,100,0.0,5.0) For i From 1 To 50000 Do Call fill_histogram(ref,rnd_polya(theta)) Enddo !options log-y Call plot_histogram(ref,`Multiplication`,`Polya distribution`) Call plot_end
Although the Polya distribution has a non-zero probability at 0, the multiplication factor used in Garfield is always equal to at least 1. If a factor less than 1 is drawn, it is rejected and a new number is drawn.
The mean multiplication factor has to be strictly positive,
[The default value for the mean multiplication is 1, the default setting of the \θ parameter is 0.5.]
See POLYA-FIXED for comments on the Polya distribution.
Additional fluctuations are obtained when the Monte Carlo drift line integration routines are used since these introduce variations in the drift path.
Although the Polya distribution has a non-zero probability at 0, the multiplication factor used in Garfield is always equal to at least 1. If a factor less than 1 is drawn, it is rejected and a new number is drawn.
This option is accessible only if Townsend data has been entered in the gas section.
[The default value for \θ is 0.5.]
This is similar to the EXPONENTIAL option, but here the mean of the distribution may be different for each ionisation cluster. Additional fluctuations are obtained when the Monte Carlo drift-line integration routines are used since these introduce variations in the drift path.
Although the exponential distribution has a non-zero probability at 0, the multiplication factor used in Garfield is always equal to at least 1. If a factor less than 1 is drawn, it is rejected and a new number is drawn.
This option is accessible only if Townsend data has been entered in the gas section.
Formatted on 21/01/18 at 16:55.