Søren Fuglede Jørgensen @fuglede@mastodon.social

Best part about https://www.state.gov/biographies/donald-j-trump/ is how the time (but not the date) seems to be changing all the time.

Numerical instability in the simulation?

More seriously, what do these folks mean when they say "70% more contagious"? Are we talking about how it affects what the exponential rate would behave like in the wild? That e.g. what would otherwise be a situation with a 0.9 growth rate should now be a 1.53 growth rate instead?

Tænker du på det samme som mig B1? Det tror jeg faktisk, B117! Det er pandemitid!

SciPy 1.6.0 just made it in before the end of 2020! [0] I'm personally quite happy with how its new matching solver came out. [1]

[0] https://mail.python.org/pipermail/scipy-dev/2020-December/024468.html
[1] https://docs.scipy.org/doc/scipy/reference/generated/scipy.sparse.csgraph.min_weight_full_bipartite_matching.html

Yet another December well spent with #AdventOfCode! Yet, feels so good to get rid of the 5:30 alarm.

https://adventofcode.com/
https://github.com/fuglede/adventofcode

I was a bit surprised to see that comma.ai explicitly values programming competition participants in recruiting, but seeing geohot feature on https://adventofcode.com/2020/leaderboard explains a lot.

Back of the envelope: So, according to a more or less randomly picked internet source [0], 1 mink causes about 10 kg of CO2eq emissions over its lifetime. So, culling 15 million of them will save us about 150000 tons CO2eq per year (not counting production of whatever people will wear instead). Compare that with the Danish target of reducing emissions by 20 million tons CO2 per year by 2030. Did we just get about 1% closer to that target?

[0]: https://www.cedelft.eu/publicatie/the_environmental_impact_of_mink_fur_production/1131

Danish investment bank Saxo Bank recently introduced minimum brokerage fees on a 1 month notice; as such, customers with small positions are now forced to either close immediately (paying a fee to Saxo Bank) or wait (paying an even larger fee to Saxo Bank); pest eller kolera, as we'd say.

If this is legal(?), what would keep a unscrupulous investor platform from introducing sky-high fees to force the hands of smaller customers?

PSA: Check out Mars right now (east-southeast from my corner of the world). It's that yellowy-orangey thing that's way brighter than it normally is: It's closer to us now than it's going to be for some 15 years.

How to solve a well-known open maths problem in 3 simple steps:

1. pip install burau

2. >>> from burau.curve import calculate_polynomial

3. Find four positive integers a, b, c, and d so that

calculate_polynomial(a, b, c, d)

produces the empty dictionary. Or show that that's impossible.

https://github.com/fuglede/burau

Version 1.2.0 of everyone's favorite .NET assignment problem solver is now available: https://github.com/fuglede/linearassignment/releases/tag/1.2.0

Have you ever had to assign opinionated stuff to other stuff, be it workers to jobs, players to teams, students to classes, or something else, then this sort of thing might be useful to you.

I'm starting to see where the general strategy to fight COVID around here is going:

- Restriction 1 (implemented): Stay at home and don't expose it to sunlight.
- Restriction 2 (implemented): Don't feed it after midnight.
- Restriction 3 (to be implemented): Do not let it come in contact with water.

Wait what.

As Danish Defence Intelligence Service is caught spying on Danes on behalf of the NSA, let's revisit our parliament's response from when the allegations first came up: https://information.dk/indland/2014/01/ingen-grund-antage

Sometimes all you can do is marvel in disgust at the utter incompetence on display.

You can even use it to cheat on your algebra exam!

Supposedly, GPT-3 is almost so good at generating text that we humans no longer have to. At first, I was curious why we didn't see much discussion around whether it was just plagiarizing training samples or not (and why doing so wouldn't be optimal as far as "generating human-like text" goes), but seeing this Q/A with itself on the topic of entropy, it seems like that's not going to be much of a concern. I think it's 1021, by the way, but I'm sure that's still a huge range!

If you did have a Danish corona test taken and have NemID, having the material destroyed is easy enough:

1. Go to your e-Boks at https://www.e-boks.dk, click "Ny besked", choose "Statens Serum Institut" and "Hovedpostkassen"

2. Write something like "Jeg vil med henvisning til databeskyttelsesforordningen samt sundhedsloven anmode om, at alle personoplysninger, herunder biologisk materiale, fra min COVID-19-test af {{DATO}}, destrueres."

With https://news.ycombinator.com/item?id=23922733 comes another reminder that large scale DNA collection is never unproblematic. And a reminder to be wary when Danish State Serum Institute uses corona tests to gather and store biological material on a large portion of the Danish population, cf. https://www.version2.dk/artikel/statens-serum-institut-glemte-at-fortaelle-biologisk-materiale-med-dna-corona-test-gemmes/

Summer time is open source time. Warm-up: Make sure that Python graph library #NetworkX can calculate minimum weight matchings of bipartite graphs without assuming that the underlying graphs are complete. [0]

This is useful whenever "something" has to be "assigned" to "someone" where each "someone" has preferences about what they're being assigned. [1] This turns out to be a common subproblem in combinatorics and constraint satisfaction.

[0] https://github.com/networkx/networkx/pull/4057
[1] https://en.wikipedia.org/wiki/Assignment_problem