kurlberg

8 ms·

61.
by kurlberg 11y ago
A case of life imitating AI koans: Uncarved block In the days when Sussman was a novice, Minsky once came to him as he sat hacking at the PDP-6. "What are you doing?", asked Minsky. "I am training a randomly wired neural net
62.
by kurlberg 11y ago
The main point is not about generating skewed distributions, it's about "universality" - the same distribution starts popping up in seemingly unrelated settings. E.g., (large) eigenvalues of random matrices is connected to t
63.
by kurlberg 11y ago
If you know emacs I recommend taking a look at org-mode. It's an emacs mode that let's you write in "markdown" (but including tex-syntax for math) and then lets you export to html or pdf (via tex/latex.)
64.
by kurlberg 12y ago
Scary scenarios regarding "faulty ram destroys volume" can be found in various places, e.g., https://pthree.org/2013/12/10/zfs-administration-appendix-c-... However, for a refutation of this scenari
65.
by kurlberg 12y ago
This is indeed where I read it. When googling to find it again I found a similar story at: https://pthree.org/2013/12/10/zfs-administration-appendix-c-... Any good ideas on how one might fix this incorrect &
66.
by kurlberg 12y ago
Thank you! I was always a bit puzzled why the Sun engineers would not verify against the on disk checksum, but it made some sense that in a server setting ECC can be assumed. By the way, the above "insane" scenario comes from the
67.
by kurlberg 12y ago
As far as I've understood, the gist of the "ZFS without ECC is scary"-mantra is that one bad memory location (say a flaw that results in a word always getting zeroed) might corrupt all your on disk data due to scrubbing (&q
68.
by kurlberg 14y ago
>The root of both bugs was a design flaw: the decision that >git.kde.org was always to be considered the trusted, >canonical source. It seems that an even bigger design flaw is that they (still) aren't doing regular backups. Th
69.
by kurlberg 14y ago
I don't know if you want a hint, but... If you do it the right way, no tweaking is needed at all. (The puzzle can be viewed as a very nice illustration of ergodic trajectories.)
70.
by kurlberg 14y ago
The answer to "Who's fastest when?" is a bit complicated - you can find a nice picture of the "champion algorithm" for different ranges at http://gmplib.org/devel/log.i7.1024.png More context and explanation can be found at: http://gmpli