sega_sai

36 ms·

661.
by sega_sai 5y ago
It seems to happen again after the first incident was fixed few hours ago (with github actions)
662.
by sega_sai 5y ago
No, I'm from 'Vtoraia Shkola'. And I wouldn't be surprised at all if there is a bias towards russian math school alumni here.
663.
by sega_sai 5y ago
It's a weird article. I think this idea of 'Russian math' is very nebulous, and is probably used as marketing tool nowadays by some schools abroad. I have graduated from one of the top physics/maths high schools in Russi
664.
by sega_sai 5y ago
Obviously everybody (Branson/Bezos) is entitled to spend their cash in whatever way they want. But, if they are making a spectacle of it, publicise it and promote it like some big achievement, then those people should be prepared to ta
665.
by sega_sai 5y ago
It's not final-final published PDF, because the copy-edited version is owned by the journal, but the accepted versions are available at places like this https://www.research.ed.ac.uk/en/publications/ (on top
666.
by sega_sai 5y ago
I don't think scihub is the future. OpenAccess is the future. I know that in the UK you basically have to post your accepted papers in publicly accessible repositories if you want your papers to count for the Research Excellent Frame
667.
by sega_sai 5y ago
I don't think professional mathematicians think highly of this https://www.galoisrepresentations.com/2019/07/07/en-passant-...
668.
by sega_sai 5y ago
That reminds me of this story https://www.nytimes.com/2019/06/17/us/daniel-beckwitt-bethes... Being non-expert it's hard to judge the safety of the tunnel, but still it doesn't sound to me like
669.
by sega_sai 5y ago
It's an interesting illustration of 'be careful what you wish for' and that the definition of the proper loss function is a very important part of the solution to any problem.
670.
by sega_sai 5y ago
I think the issue is in both: seeding with time and using a generator that does not have an additional source of entropy (i.e. is fully deterministic given seed).
671.
by sega_sai 5y ago
I certainly understand some arguments against rainbow, but in the same time the new default colour map in matplotlib (viridis) is (IMO) worse for many applications. It doesn't give enough contrast/doesn't define contour-lines
672.
by sega_sai 5y ago
I think the problem is that the complexity analysis of adding is often easier, as opposed to subtracting. I.e. if I added an if statement, one can argue that you may need to only analyse that statement, while removing a bit of code may requ
673.
by sega_sai 5y ago
So far we lost x) Freedom of movement x) The medical insurance coverage when traveling in Europe x) Access to many European research grants We gained x) roaming charges x) postage duties for packages I'm sure I'm missing many oth
674.
by sega_sai 5y ago
Your link is from March 2021...
675.
by sega_sai 5y ago
I (and many in science) switched from C/PDL/IDL to Python in last 10 years because it was is the best tool, not for zealotry. Python vs Perl for science has nothing to do with c/c++/fortran. C/C++/fortran have
676.
by sega_sai 5y ago
I'm afraid this is ~ 10-15 years too late. I've used PDL somewhat in ~ 2005-2010 when python didn't have that many packages/numpy, and it did the job and could substitute IDL to some degree. But realistically, right now
677.
by sega_sai 5y ago
I had a backup JBOD with 36 of these drives. I thought since it was a backup machine that rarely will be used I could buy non-enterprise drives... Big mistake. There were 47 disk failures from 2013 to 2018.
678.
by sega_sai 5y ago
When there is no evidence, the discussion is purely in the sphere of priors. And that doesn't make it very fruitful to discuss, as different people will likely have different priors. I think I agree that the likelihood of lab-leak was
679.
by sega_sai 5y ago
You are not quite correct. The deconvolution only works well if you know exactly the convolution kernel and you have infinite signal to noise. In practice neither is true and you'll have artefacts if you try to do deconvolve. And the r
680.
by sega_sai 5y ago
Epicycles are predictive, because they just represent a Fourier approximation to an ellipse. They predict the future, just with finite precision, so to increase precision you need to add more parameters. Regarding dark matter, I'm not
681.
by sega_sai 5y ago
There are different simulations focusing on various aspects of the universe. All the CDM simulations give (predict) a dark matter density profile close to the so called Navarro-Frenk-White (NFW) profile (i.e. here https://ned.ipa
682.
by sega_sai 5y ago
I'll point you here on the topic of the cluster speed https://ui.adsabs.harvard.edu/abs/2007MNRAS.380..911S/abstra... I.e. it's not a problem.
683.
by sega_sai 5y ago
Copernicus model was not that much of a change from a model complexity point of view, as it just removed some common epicycles by putting Sun in the center. The key changes were Kepler's results and Newton's, where one had provide
684.
by sega_sai 5y ago
Epicycles were a good theory that approximated a ellipses by combinations of circles (essentially through a Fourier transform). When a simpler theory came that was easier to compute, was more predictive and came from a first principles calc
685.
by sega_sai 5y ago
It's a bit of the same from Sabine. Dark matter hypothesis allowed us to simulate a vast range of processes in the Universe, from large scale cosmological simulations, predicting clustering of galaxies, cosmic microwave background to s
686.
by sega_sai 5y ago
This is a nice one: “It is not to be supposed that the death of the Archduke Francis Ferdinand will have any immediate or salient effect on the politics of Europe.” I think the quality of predictions in geo-politics or economics hasn&#x27
687.
Millions at security risk from old routers, provided by EE, Sky and Virgin(bbc.co.uk)
23 points by sega_sai 5y ago|0 comments
688.
by sega_sai 6y ago
Here is the top 20 files by the number of commits 12846 MAINTAINERS 4167 drivers/gpu/drm/i915/intel_display.c 3330 drivers/gpu/drm/i915/i915_drv.h 2360 drivers/gpu/drm/i915
689.
by sega_sai 6y ago
What I meant is when whole systems go down (i.e. canvas, blackboard, office365 or similar) as opposed to the internet for one person, the problem is the amount of stress and extra-work inflicted on thousands of people is (I think) can be qu
690.
by sega_sai 6y ago
I appreciate this view, but I'm in academia, and with covid19 we are teaching remotely, doing exams remotely, etc. If the systems are down that can have a real disrupting effect on students not being able to submit homeworks/exams

More ›