you have some normal distribution with mean 0 and variance $^2$ and that you want the interval [,] such that the uniform distribution on the interval has mean 0 and variance $^2$ ?
This is my variance equation when the numbers are drawn from a normal distribution $$2/(n_{in}+1/n_{out})$$ The answer for the uniform distribution with the same variance is $$6/(n_{in}+1/n_{out}),-6/(n_{in}+1/n_{out})$$ How exactly does the 6 come into play is it from the integral of the uniform distribution formula and how exactly does that work ? I understand why you need a range for a uniform distribution but I don't understand where the 6 comes from ?
for more context: n_in and n_out are the number of neurons into a layer and out of a layer of a neural network. Here is the paper where I am getting the formulas from https://proceedings.mlr.press/v9/glorot10a/glorot10a.pdf on page 5