robochat · HackerTrans

6 か月前議論

I think they mean that the model should have sleep period where they update themselves with what they learnt that day.

9 か月前議論

I remember being amazed when I learnt about summing arithmetric series.

10 か月前議論

They took a multiplier of 5x (4 indirect deaths for every direct death) and stated that this was conservative given studies of previous conflicts.

robochat

11 か月前議論

"Valuable Humans in Transit and Other Stories" by Qntm has some good (harrowing) stories about human uploads too.

robochat

昨年議論

We actually use the twitter detections to launch analyses of the seismic data in order to get confirmed results for events that aren’t reported yet [1] but there are some statistics for the twitter detections in the supplementary material of that article [2]. Basically, in 2016-2017 (wow, so long ago), we detected 893 earthquakes via twitter, with a median delay of 67s and a median separation from the published epicentre of 94km. Note that estimating earthquake epicentres is nontrivial anyway and so, for comparison, 10km accuracy would often be considered ok. So the twitter, I mean X, method isn’t optimal but it gets you down to the right region. Partly it’s because geocoding the tweets is inaccurate and partly it’s because people live clumped together in cities rather than smoothly spread over the surface of the earth.

[1] https://www.science.org/doi/10.1126/sciadv.aau9824 [2] https://www.science.org/action/downloadSupplement?doi=10.112...

robochat

昨年議論

There is also the Earthquake Network (EQN) app that works on very similar principals to Google’s system - phones monitor their MEMs accelerometers, when they are left charging with their screen off, and when enough neighbouring phones detect vibrations simultaneously an earthquake is detected and apps nearby are alerted. It’s been running since 2012.

[1] https://sismo.app

robochat

昨年議論

The USGS created a system to do exactly this about 15 years ago. I’m not sure whether they’re still running it but at the EMSC, we've been running a similar system for many years to highlight earthquakes important to the public and improve our messaging. Twitter doesn’t give access to geotags anymore but we do manage to roughly estimate an earthquake’s location by analysis of the tweets. Estimating magnitude is much more difficult. Naturally there are some false positives but it works well overall.

[1] https://www.usgs.gov/media/audio/shaking-and-tweeting-usgs-t...

robochat

3 年前議論

I have used Thunderbird for years and my work account has to handle large numbers of emails. In general it is fast for me on both Windows and Linux. I have had some issues over the years but they normally turned out to be due to corrupt mail folders.

robochat

3 年前議論

Can anyone explain the maths that he did because my calculation give different results. Isn't there only a 1 in 2588 chance of a student 'guessing' 19 'answers' out of 45 where there are 5 options for each answer? Whereas the article states that it is 1 in 100 ? (actually he writes 1:100 which is odds, so it is 1 in 101?) Don't we just use the Binomial distribution, so: Prob = 45!/(19!26!) * (1/5)*19 * (1 - 1/5)*26 It's really bugging me that I can't get the maths to work.

robochat

3 年前議論

Actually, I can't seem to get the maths to work. Isn't it just the Binomial distribution? Each question has 5 options so the probability of 'success' is 1/5 and so to get 19 questions 'specifically wrong' out of the 45 planted questions by chance is just (1/5)*19 * (1-1/5)*26 * 45!/(19!26!) = 1 in 2588 but in the article it is 1:100. What am I doing wrong ?

robochat

3 年前議論

This feels harsh because if the tests were similar but with small differences then getting the right wrong answers seems likely to happen anyway. Also, he went the extra step and outright failed the student but I'm not sure why that extra step was justified, surely the cheaters were automatically going to get very poor marks.

robochat

3 年前議論

It was an online course and the exam was online too.

robochat

3 年前議論

For all of my major exams in the UK, we were encouraged to study by practising on old papers and my score would have been abysmal if I hadn't. So not being allowed to look at past exams feels strange to me too.

robochat

3 年前議論

There is a strong connection between the incandescent lightbulbs and the sun because the sun is almost a blackbody emitter too. Studying incandescent lightbulbs allowed researchers to discover the blackbody spectrum and Kirchhoff's law of thermal radiation, which was then applied to the Sun to discover its apparent temperature. The spectral lines in the Sun's spectrum (the frequencies where it deviates from the blackbody spectrum) then allowed researchers to study elements present in the Sun's photosphere.

robochat

3 年前議論

The paper is written in scathing tones, by page 11 I was laughing at each new report of QED's 'perfect' agreement with whatever was the latest experimental result for the electron's anomalous magnetic moment although as it continued as far as page 14, it became sobering. However, the recent discrepancies are way down in the decimal points of the result now; when these calculations require 1000s of complicated Feynman diagrams and pages and pages of calculation, errors seem inevitable. My hope is that an alternative approach can be found that avoids these complicated calculations and in fact, researchers already seem to be working on this [1].

[1] Nima Arkani-Hamed - The End of Spacetime (Sep 2022) https://www.youtube.com/watch?v=GL77oOnrPzY

robochat

3 年前議論

I agree with this sentiment but in this particular case, with the personality that Putin projects to the world, it seems like appeasement would just encouragement him to try to take more.

robochat

3 年前議論

I like nim but I've never really gotten used to the slicing syntax. I guess that I'm just too used to python, plus you have to be a bit careful to leave a whitespace when doing backwards indexing so that the operators work correctly.

So python:

    a = [0,1,2,3,4,5]
    a[0:5] => [0,1,2,3,4] - half closed interval
    a[0:-1] => [0,1,2,3,4] - negative indexing
    a[:5] => [0,1,2,3,4] - skipping start index 
    a[3:] => [3,4,5] - skipping end index

nim:

    var a = [0,1,2,3,4,5]
    a[0..5] = [0,1,2,3,4,5] - closed interval
    a[0..^1] = [0,1,2,3,4,5] - closed interval with negative indexing
    a[..4] = [0,1,2,3,4] - skipping start index
    a[..<5] - Error - doesn't work
    a[3..] - Error - doesn't work
    a[0..<5] = [0,1,2,3,4] - half closed interval
    a[0..< ^1] = [0,1,2,3,4] - half closed interval with negative indexing 
                              (Important - the white space is required, it will crash otherwise!)

I'm sure that there are good reasons for this behaviour and that if you use nim a lot, you just get used to it but I find the python syntax just nicer.

robochat

3 年前議論

You can easily get bash to do that by adding these lines to your .inputrc file.

  "\e[A": history-search-backward
  "\e[B": history-search-forward
  "\e[C": forward-char
  "\e[D": backward-char

robochat

4 年前議論

Yes this is why many high frequency traders have installed laser or microwave links to their nearest stock exchanges, simply to gain a few microseconds advantage in their trades. Fibre isn't as fast as laser going through the air. It's 50% slower (as the parent comment states) (unless you are using special hollow core fibres which are uncommon for now). More importantly, fibres rarely go in a straight line between 2 points, they wind their way through buildings, down into basements, through buried pipes etc and this all adds extra distance to the route and hence more delay to the signals. A line of sight link is the shortest route between 2 points.

robochat

4 年前議論

Actually, Einstein's 1935 paper along with Boris Podolsky and Nathan Rosen basically highlighted the existence of entanglement in Quantum Mechanics. Of course Einstein was trying to show that QM was wrong via the EPR paradox but although reality has since proven him wrong, it was still a major paper written decades after his more famous work.

コメント