| ECE532 Biomedical Optics
© 1998
Steven L. Jacques, Scott A. Prahl
Oregon Graduate Institute |

# Monte Carlo simulations

## Sampling a Gaussian laser beam profile

### Consider an example.

Suppose we have a Gaussian laser beam profile of relative irradiance E(r) [mm^{-2}] where b is the 1/e radius, i.e., E(b) = (1/e)E(0):

The probability density function describing the beam profile as a function of radial position r is:

The first term is the Gaussian function exp(-r^{2}/b^{2})/(b^{2}) that describes the radial dependence of the beam irradiance along one radial dimension. The second term 2r describes the circular rings of integration over the cross-sectional plane of the beam. The product of these two terms yields the probability density function p(r) whose integration over r = 0 to infinity equals unity. The distribution function F(r) is the integral of p(r) for r = 0 to r_{1}:

Now apply the same logic to the probability density function for a random number, p(rnd) = 1. The integral of p(rnd) for rnd = 0 to 1 equals 1, so p(rnd) is properly normalized. The distribution function F(rnd_{1}) is the integral of p(rnd) for rnd = 0 to rnd_{1} which simply equals rnd_{1}:

Equating the two distribution functions, F(rnd_{1}) = F(r_{1}), yields:

Rearrange the above equation to solve for r_{1} as a function of rnd_{1}:

The above equation is the rule for sampling p(r).

The two figures below illustrate the generation of p(r) and E(r) using the above equations. The dots indicate the histograms for p(r) and E(r) created using 10,000 random numbers for rnd_{1}. The lines indicate the analytic expressions for p(r) and E(r). The figures were generated using a MATLAB program listed here.

p(r)

E(r)

to time-resolved Monte Carlo pages |
to steady-state Monte Carlo pages

Chapter 4 |
Course |
Home