Tomohiro Endo

@tupui Thank you for your advice but, currently, I can't build scipy in my environment. So I ran a benchmark test here. https://colab.research.google.com/drive/1ZJwqlwW7n5DueNNaeOI1nWDf_4qt4_9K?usp=sharing

It became possible to build scipy in a local environment. This is the benchmark test I ran with asv in my local environment. At least for my machine, we can...

Thank you @tupui. Please correct me if I am wrong. https://github.com/scipy/scipy/blob/a2b53047a27897d6dae1587f4b88549934f759f7/scipy/stats/_kde.py#L590-L601 and https://github.com/scipy/scipy/blob/a2b53047a27897d6dae1587f4b88549934f759f7/scipy/stats/_stats.pyx#L795-L796 is the same tests. So the test in cython can be removed. In my commit, I removed...

Though I haven't written a unit test, I made a cython version `logsumexp`. It is similar to https://github.com/scikit-learn/scikit-learn/blob/ef21ca1950f9570f2f6b1ce05d24111f78f9f612/sklearn/linear_model/_sag_fast.pyx.tp#L76-L95 (https://github.com/scipy/scipy/pull/15493#discussion_r910096498) The function I made can only accept an one dimensional array...

I didn't find any mistake in the commit and it looks good to me. The code became simpler than before. Thank you.

I wrote a AWS CloudFormation template that creates infrastructures ranging from VPC to a EC2 spot instance and installs Stable Diffusion Web UI automatically. https://github.com/europeanplaice/stable-diffusion-webui-ec2-spot-instance