Statistics Professor @ Sacramento State
Sacramento, CA
Joined June 2011
- Tweets2.1K
- Following437
- Followers322
- Likes1.9K
"While there is usually one way to do things right, the ways to screw something up are infinite." stackoverflow.blog/2020/07/1…
Debuggers are so nice. Blog post: introduction to #debugging in #julialang with inspiration from @matloff
webpages.csus.edu/fitzgerald…
Everything mysteriously broke when I copied and pasted some code that had a Unicode hyphen (8208) for a command line flag instead of an ASCII dash -. Argh. 😬
Clark Fitzgerald retweeted
Interested in scientific visualization, Python and fusion energy research? We're hiring for a 6 month position at @GoogleAI Applied Sciences: goo.gl/6rJ7Cy
Happy about everything that tmux can do on a SLURM cluster. Try
srun --pty bash
to open a shell in a compute node to experiment a bit so you don't tie up the head node.
Clark Fitzgerald retweeted
I left employment at Anaconda Inc. last week.
I start employment at NVIDIA next month.
Surprisingly little about my work will change.
More details: matthewrocklin.com/blog/work…
Clark Fitzgerald retweeted
Here are 6k accepted #rstats answers that use/mention #rdatatable where the question was not about data.table. (Distinct from the 8k questions tagged data.table.) Would love to see some data science on this to help guide the project. Anyone interested? stackoverflow.com/search?q=%…
USAspending.gov offers their entire database as a Postgres dump. Materializes to ~1TB! #BigData #opendata files.usaspending.gov/databa…
The caching in #rstats bigmemory package worked perfectly for an application that moved through small neighborhoods of a large matrix. Nice work by @kaneplusplus et. al. cran.r-project.org/package=b…
Clark Fitzgerald retweeted
I'm pleased to announce xarray v0.11:
xarray.pydata.org/en/stable/…
This release includes:
- file-storage refactor for performance with @dask_dev
- better support for calendars used in climate science
- lots of other miscellaneous API clean-ups, bug-fixes and enhancements
> as.character(list(numeric()))
[1] "numeric(0)"
> as.numeric(list(numeric()))
[1] NA
> as.numeric(list())
numeric(0)
#rstats🤔
Cool use case for #metaprogramming:
Given a table with many equations in text, turn them into functions and call them on observations in another table.
Clark Fitzgerald retweeted
Please file an issue/bug report if you have a problem with a package, instead of contacting the maintainer directly. This lowers the burden on the maintainer and helps others in the community know about the state of potential problems.
#rstats
Clark Fitzgerald retweeted
Goodnight columns, goodnight rows,
Goodnight kind strangers on Stack Overflow,
Goodnight factors, goodnight strings,
Goodnight overfitted things,
Goodnight humans, goodnight bots,
Goodnight inconclusive plots.
Goodnight R 😴
#rstats
TIL that parameter in a model can occupy several TERABYTES. Wow!
From Google's tensorflow paper ai.google/research/pubs/pub4…
Last week we worked on some slow code.
Parallel gave us a ~20x speed up.
Switching to the right data structure gave us a ~20,000x speedup.
#parallel #programming
Clark Fitzgerald retweeted
We're developing a new protocol (__array_function__) that allows for alternate implementations of NumPy functions (e.g., for GPU, autodiff or units support): numpy.org/neps/nep-0018-arra…
It's now implemented on the master branch of NumPy -- please give it a try and report back!