2 ms·
> The trick is that a normal distribution occurs when you sum independent, symmetric distributions. It's not neccesary that the distribution is symmetric. You
by gus_massa 3y ago
> The trick is that a normal distribution occurs when you sum independent, symmetric distributions.
It's not neccesary that the distribution is symmetric. You can try for example with 500 genes witn 10% of 0.00 extra height and 90% of 0.01 extra height. You will get again a gaussian but the center will be in 500*(.10*0.00+.90*0.01).
- remolacha 3y agoInteresting, will try this. I imagine "symmetric" is not the correct term, but what is? I imagine there are constraints on types of random variables for which this will work, but I haven't looked into it deeply.
- gus_massa 3y agoIt's weird/interesting/ammazing, that there are almost no constrains. The distribution can be as weird as you want. The only condition is that the variance must be finite.