Moritz Schauer

Ah, so my first guess for the memory layout was actually right according to ``` dyadicdetailrange(j::Integer) = (2^j+1):(2^(j+1)) ``` I just did not believe it: i started with smooth signal...

Ah, I found my mistake, the time points `t = linspace(0,1,128)` were not at equal distance on a circle of length `1`, this works. ``` julia> t = (1:128)./128 julia>...

Bump :-)

It should just say "success probabilities`a/(a+c)` and `b/(b+d)`" then...

I think this is caused in https://github.com/JuliaStats/Distributions.jl/blob/master/src/univariate/continuous/beta.jl#L36, perhaps file an issue with the Distributions project?

I would separate model assumptions, e.g. equal variance from hypotheses (eq. equal mean.) Something like: > For samples from two populations perform a two sample `t`-test for the null-hypothesis that...

Maybe something for https://discourse.julialang.org/ ?

Top! https://discourse.julialang.org/t/classical-hypothesis-tests-wald-lr-lm/39575

Reopen, it makes sense to discuss this here

Ah that’s where the order comes in, thanks. But the stack like hash table: I think a symbol table with scope context has this behavior