Geometric distribution instead of Binomial in Forest Fire Sampler

[rishabh-ranjan]

https://github.com/benedekrozemberczki/littleballoffur/blob/9cc8da02e2362a8355c0342c0702c3eb6f4530c3/littleballoffur/exploration_sampling/forestfiresampler.py#L76

Shouldn't this be using the binomial distribution instead, as we want to simulate the effect of each neighbor burning with probability p independently? Am I misunderstanding something?