i9e1

regarding the KiCad mirroring: I assume they had symbols in mind in the first place. Consider OpAmps, some draw the inv above the non-inv input, others draw it vice versa....

As Ed said its useless to discuss workflows, but I think the two extremes here have different levels of communication between the circuit and pcb designer (who is allowed to...

I came accross this when calculating the entropy of single values (like entropy(8,2) = -24.0 Bit, nonsense). Currently my uncertainty model includes certainty and robust bounds like Union{Real, Interval, UnivariateDistribution}....

Ok it's not that easy after playing with the tests. If we don't want in-place normalization, the additional allocations (and thus performance) are an issue.

There is no problem imho returning `-0.0`. It's a float so there are signed versions of zero, `0.0` and `-0.0`. ``` julia> iszero(0.0) true julia> iszero(-0.0) true julia> 0.0 ==...

Complex probabilities may be a thing [complex probability and information theory](https://doi.org/10.1080/21642583.2017.1367970). Entropy extensions are in Sec 8 but I dont get it. Constrain argument on `AbstractVector{

!close This can be closed as the documentation (now) clearly states that default ordering is min (Forward).

common example for `crosscor`: searching the most likely lag (max crosscor) does not need to normalize. The user maybe needs to do some normalization before anyways (so that data matches...

maybe related #1748 I assume now that `Poisson` is not supported as stated in doc of `fit_mle`.