Nima Sarajpoor

@seanlaw We just got an error caused by slight imprecision in identical case in a about-to-be-merged PR. (see [error](https://github.com/TDAmeritrade/stumpy/actions/runs/3045826420/jobs/4907870570)) I was wondering if we should partially (?) solve this as...

@seanlaw Thanks for sharing the links. I took a look at them. > Hmm, according to [this](https://stackoverflow.com/a/26264307/2955541), `np.allclose` uses `all(A == B)` behind the scenes I think you are right....

**side note:** It might be worthwhile to note that even small error in $\rho$ can result in large error in z-normalized Euclidean distance. Example: ``` m = 3 # size...

> should we have a `config.STUMPY_RHO_THRESHOLD` instead of the `config.STUMPY_CORRELATION_THRESHOLD`? Did you mean `config.STUMPY_DISTANCE_THRESHOLD`? (I mean.... aren't `RHO` and `CORRELATION` the same thing?) --- Also, note that the following numbers...

> Doesn't this mean that `config.STUMPY_CORRELATION_THRESHOLD` is too small? Maybe we set it to `1 - 1e-06` instead? > > Can you please clarify if the issue is in the...

> > Did you mean config.STUMPY_DISTANCE_THRESHOLD? (I mean.... aren't RHO and CORRELATION the same thing?) > > Oops, you're right I am thinking about adding something like `config.STUMPY_DISTANCE_THRESHOLD=1e-3`. I will...

> > config.STUMPY_DISTANCE_THRESHOLD=1e-3 > > What is this for? What will it be used to control or compare against? Where would you use it in the codebase? So, this is...

@alvii147 @seanlaw So, I took a look at the code and worked with it a little bit. I tried different things but didn't see improvement! ~I have a couple of...

@seanlaw @alvii147 Okay...so I had some comments but apparently I was wrong. I am doing some investigation. Will update you if I find something.

This is what I got:  And,  So, it seems there is improvement when length of `T` is small. However, when `T` gets longer, the gap becomes more and...