#soylent | Logs for 2021-01-29
« return
[00:08:51] <TheMightyBuzzard> chromas, LFS
[00:21:50] * TheMightyBuzzard buggers off having exhausted his afternoon ass-sitting time
[00:31:04] -!- AzumaHazuki [AzumaHazuki!~hazuki@the.end.of.time] has joined #soylent
[00:59:31] <Runaway1956> LOLOL - my latest journal has a whole bunch of them triggered.
[01:02:17] <Runaway1956> Anal swabs? My eyes must be failing. Now they want to swab our anuses? Or is that ani? Is that a Greekish word? "Bend over and spread 'em!"
[01:16:03] <Bender> [SoylentNews] - Data Breaches Fell in 2020, but Identity Thieves Still Raked in Billions - https://sylnt.us - making-the-most-of-a-bad-situation
[01:28:35] <Bytram> https://www.cnn.com
[01:28:38] <systemd> ^ 03Cicely Tyson, iconic and influential actress, dies at 96
[01:29:07] <Bytram> She was how old? Zounds!
[01:29:50] <TheMightyBuzzard> #yt subterranean homesick blues
[01:29:51] <MrPlow> https://www.youtube.com -- Bob Dylan - Subterranean Homesick Blues
[01:31:50] * TheMightyBuzzard yawns
[01:31:50] * MrPlow flips a Skittle into TheMightyBuzzard's gaping mouth
[01:33:18] <TheMightyBuzzard> blarg. days keep getting shorter and the amount of shit i have to do in a day just keeps piling up. be really, really glad when that church is done and i can find a few hours to spend with my rod in my hand.
[01:34:52] <Bytram> TheMightyBuzzard: s/ro/fishing ro/
[01:35:30] <Bytram> ;)
[01:36:08] <TheMightyBuzzard> pun intended
[01:36:18] <Bytram> noted
[01:36:50] * Bytram struggled, but succeeded, in not invoking !grab
[01:37:10] <TheMightyBuzzard> if you can't find something funny to say about the person you know best, you don't have a sense of humor.
[01:39:10] <TheMightyBuzzard> ya know. i may have to add a wee bit more paid work into the yearly quota. be so worth it to get someone in to do nothing but do the dishes every night.
[01:39:44] <Bytram> get a dishwashing machine?
[01:40:18] <TheMightyBuzzard> so i can wash them before it ineffectively sprays hot water at them? nah.
[01:40:30] <Bytram> uses hotter water (more sanitary) and gets 'em cleaner than doing it by hand.
[01:41:20] <TheMightyBuzzard> the soap's the sanitizer. you can't use hot enough water to sanitize squat without melting half the stuff you own.
[01:42:49] <Bytram> nah, they're much better now. Dishes just need a bit of a rinse, stick them in, turn it on, wait... might just have a bit of bleach in the soap, too. I just know that when my folks installed a dishwasher, us kids stopped having colds and tonsilitis all the time
[01:43:04] <TheMightyBuzzard> and nothing gets em cleaner than when i do them by hand. i do not allow for missed bits cause i don't want to eat off anything with missed bits.
[01:43:43] <Bytram> so, maybe what you'd want is a... bit bot?
[01:44:17] <TheMightyBuzzard> tr's kids make lousy bit bots. the boy takes 10x as long as anyone should and the girl misses all of the spots.
[01:44:33] <Bytram> https://www.cnn.com
[01:44:36] <systemd> ^ 03US Army announces a new grooming policy in a push for inclusion
[01:45:16] <Bytram> now all the uniforms come with padding around the breasts?
[01:45:27] <Bytram> o_O
[01:46:15] <TheMightyBuzzard> feel kinda sorry for the ones who go through the gas chamber with their shiny new hairstyles. there's a reason those regs are in place and pushing the envelope will lead to a lot of pain.
[01:47:22] * TheMightyBuzzard is somewhat tired of virtue signaling theater
[01:47:36] <AzumaHazuki> i've gt a bunch of co-workers who wear their hair like this. it is not fun in the IV room for them
[01:47:38] <Bytram> I suspect that the powers-that-be are quite aware that sometimes the best way to persuade someone is to give them what they ask for and let them learn it was NOT what they wanted.
[01:47:53] <AzumaHazuki> of course, i have to stick my brain down the back of my scrubs to work in there so @_@
[01:47:58] <AzumaHazuki> braid*, ye gods
[01:48:25] <TheMightyBuzzard> AzumaHazuki, now imagine someone with nifty cornrows screwing up the seal on their mask in a room full of tear gas, where they'll be spending about 15 or 20 minutes before they can go outside.
[01:48:57] <AzumaHazuki> sounds like a bad time, m'yes
[01:49:23] <TheMightyBuzzard> and that's the friendliest gas they're ever going to need the mask for. well except the range shitters.
[01:49:37] <AzumaHazuki> or in the mess hall on taco tuesday
[01:49:56] <TheMightyBuzzard> very handy that they made us carry them out to the firing range in summer. i was the only one smart enough to make use of mine.
[01:50:39] <TheMightyBuzzard> smelled very like being in the monkey house at an ill kept zoo, pre-mask
[01:51:28] <TheMightyBuzzard> anywho, sigh, dishes. yall enjoy.
[01:51:33] <Bytram> As for the men in the Army who are wondering if beards will be allowed, Grinston had a short answer: "No."
[01:51:48] <Bytram> Bbbbut, the women can?
[01:52:07] <Bytram> Klinger!!!!!!
[01:52:14] <Bytram> break time
[02:06:22] -!- c0lo [c0lo!~9f99d30a@159.153.yqr.yr] has joined #soylent
[02:15:21] -!- Alfred [Alfred!~Agorophob@66.115.unl.mx] has joined #soylent
[02:17:58] -!- SwahiliWilli has quit [Ping timeout: 265 seconds]
[02:33:28] -!- AzumaHazuki has quit [Remote host closed the connection]
[02:35:51] -!- lld has quit [Ping timeout: 265 seconds]
[02:41:16] -!- SwahiliWilli [SwahiliWilli!~Agorophob@154.3.jv.orp] has joined #soylent
[02:43:35] -!- Alfred has quit [Ping timeout: 265 seconds]
[02:53:47] <c0lo> =smake bots
[03:07:38] -!- lld [lld!~lld@2001:f40:jms::k:nkzt] has joined #soylent
[03:21:02] <c0lo> #submit https://www.dailymail.co.uk
[03:21:03] <MrPlow> Submitting. There is a mandatory delay, please be patient.
[03:21:28] <MrPlow> Submission successful. https://soylentnews.org
[03:29:47] <c0lo> #submit https://www.thesun.co.uk
[03:29:48] <MrPlow> Submitting. There is a mandatory delay, please be patient.
[03:30:14] <MrPlow> Submission successful. https://soylentnews.org
[03:33:45] <c0lo> #submit https://scitechdaily.com
[03:33:45] <MrPlow> Submitting. There is a mandatory delay, please be patient.
[03:34:10] <MrPlow> Submission successful. https://soylentnews.org
[03:38:42] <c0lo> #submit https://www.sciencealert.com
[03:38:43] <MrPlow> Submitting. There is a mandatory delay, please be patient.
[03:39:08] <MrPlow> Submission successful. https://soylentnews.org
[03:41:19] <c0lo> #submit https://www.universetoday.com
[03:41:19] <MrPlow> Submitting. There is a mandatory delay, please be patient.
[03:41:45] <MrPlow> Submission successful. https://soylentnews.org
[03:44:10] <c0lo> #submit https://www.nanowerk.com
[03:44:11] <MrPlow> Submitting. There is a mandatory delay, please be patient.
[03:44:36] <MrPlow> Submission successful. https://soylentnews.org
[03:45:41] <Bender> [SoylentNews] - FreeBSD/i386 Demoted to Tier 2 for FreeBSD 13.x - https://sylnt.us - remember-when-32bit-was-the-new-hotness?
[03:48:38] <c0lo> #submit https://www.livescience.com
[03:48:39] <MrPlow> Unable to find a summary for that page
[03:51:22] <c0lo> #submit https://www.sciencemag.org
[03:51:22] <MrPlow> Submitting. There is a mandatory delay, please be patient.
[03:51:47] <MrPlow> Submission successful. https://soylentnews.org
[03:53:41] <c0lo> #submit https://appleinsider.com
[03:53:41] <MrPlow> Submitting. There is a mandatory delay, please be patient.
[03:54:06] <MrPlow> Submission successful. https://soylentnews.org
[03:59:01] <c0lo> #submit https://www.gizmodo.com.au
[03:59:03] <MrPlow> Submitting. There is a mandatory delay, please be patient.
[03:59:28] <MrPlow> Submission successful. https://soylentnews.org
[04:00:01] <c0lo> #submit https://www.theguardian.com
[04:00:02] <MrPlow> Submitting. There is a mandatory delay, please be patient.
[04:00:27] <MrPlow> Submission successful. https://soylentnews.org
[04:00:57] <c0lo> #submit https://www.engadget.com
[04:00:58] <MrPlow> Submitting. There is a mandatory delay, please be patient.
[04:01:23] <MrPlow> Submission successful. https://soylentnews.org
[05:44:29] <SwahiliWilli> MrPlow plow yo mama!
[05:53:47] <Runaway1956> https://youtu.be
[05:53:49] <systemd> ^ 03Attempted home break in caught on doorbell camera; one suspects shoots himself in leg ( https://www.youtube.com )
[06:18:17] <c0lo> chromas, where are those two bots supposed to be running? On your computer at home or what?
[06:18:30] <c0lo> I mean, systemd ans sedbot.
[06:21:34] <chromas> Yeah; they run on the box with the failing hdd
[06:22:49] <chromas> scamazon canceled by replacement parts order because it suddenly doesn't like my payment method that I've been using for awhile now
[06:23:00] * chromas blames crutchy
[06:23:29] -!- SedBot [SedBot!~SedBot@603-724-736-251.res.spectrum.com] has joined #soylent
[06:25:32] -!- SoyCow7066 [SoyCow7066!~d8c4e185@925-300-434-342.ubr6.dyn.lebanon-oh.fuse.net] has joined #soylent
[06:25:44] <Bender> [SoylentNews] - Billions of Cicadas Set to Invade Parts of USA for First Time in 17 Years - https://sylnt.us - stop-bugging-me!
[06:28:00] -!- SedBot has quit [Changing host]
[06:28:00] -!- SedBot [SedBot!~SedBot@pid1] has joined #soylent
[06:43:12] <c0lo> Can they be installed on those virtual chemical boxen that support S/N?
[06:47:11] -!- SoyCow7066 has quit []
[06:48:38] <chromas> they could
[06:48:51] <chromas> although I wouldn't put systemd on there because there's a bug that causes it to hog ram sometimes
[06:49:28] <chromas> and linux is smart enough to keep handing out memory, no matter how much is unavailable
[06:57:34] <c0lo> What lang are they coded in?
[06:59:45] -!- Ingar has quit [Ping timeout: 265 seconds]
[07:01:29] <chromas> https://babylonbee.com
[07:01:30] <systemd> ^ 03GameStop Announces That Due To Skyrocketing Stock, They Can Now Afford To Pay Up To 25 Cents For Your Used Games
[07:01:50] <chromas> systemd is d lang while sedbot is an awk script
[07:02:05] <chromas> sedbot should be fine running on a soybox. I should ask about that
[07:02:06] * SedBot is a 54-line awk script, https://github.com
[07:17:52] -!- Ingar [Ingar!~ingar@2001:bc8:xsrm:iors::i] has joined #soylent
[07:22:28] -!- Ingar has quit [Ping timeout: 265 seconds]
[07:23:44] -!- Fnord666_ has quit [Read error: Connection reset by peer]
[07:24:04] -!- Fnord666_ [Fnord666_!~Brian@925-300-434-342.ubr6.dyn.lebanon-oh.fuse.net] has joined #soylent
[07:24:04] -!- Fnord666_ has quit [Changing host]
[07:24:04] -!- Fnord666_ [Fnord666_!~Brian@Soylent/Staff/Editor/Fnord666] has joined #soylent
[07:24:04] -!- mode/#soylent [+v Fnord666_] by Aphrodite
[07:24:28] -!- Fnord666_ has quit [Read error: Connection reset by peer]
[07:24:44] -!- Fnord666_ [Fnord666_!~Brian@925-300-434-342.ubr6.dyn.lebanon-oh.fuse.net] has joined #soylent
[07:24:44] -!- Fnord666_ has quit [Changing host]
[07:24:44] -!- Fnord666_ [Fnord666_!~Brian@Soylent/Staff/Editor/Fnord666] has joined #soylent
[07:24:44] -!- mode/#soylent [+v Fnord666_] by Aphrodite
[07:29:38] -!- Fnord666_ has quit [Quit: No Ping reply in 180 seconds.]
[07:30:53] -!- Fnord666_ [Fnord666_!~Brian@925-300-434-342.ubr6.dyn.lebanon-oh.fuse.net] has joined #soylent
[07:30:53] -!- Fnord666_ has quit [Changing host]
[07:30:53] -!- Fnord666_ [Fnord666_!~Brian@Soylent/Staff/Editor/Fnord666] has joined #soylent
[07:30:53] -!- mode/#soylent [+v Fnord666_] by Aphrodite
[07:34:57] -!- Fnord666_ has quit [Read error: Connection reset by peer]
[07:35:06] -!- Fnord666_ [Fnord666_!~Brian@925-300-434-342.ubr6.dyn.lebanon-oh.fuse.net] has joined #soylent
[07:35:06] -!- Fnord666_ has quit [Changing host]
[07:35:06] -!- Fnord666_ [Fnord666_!~Brian@Soylent/Staff/Editor/Fnord666] has joined #soylent
[07:35:06] -!- mode/#soylent [+v Fnord666_] by Aphrodite
[07:38:18] -!- Ingar [Ingar!~ingar@2001:bc8:xsrm:iors::i] has joined #soylent
[07:44:00] <FatPhil> Buy DOGE and HODL!!!!!
[08:56:39] <Bender> [SoylentNews] - Favicons May be Used to Track Users - https://sylnt.us
[08:59:04] -!- Alfred [Alfred!~Agorophob@154.3.jv.orp] has joined #soylent
[09:00:30] -!- SwahiliWilli has quit [Read error: Connection reset by peer]
[09:21:22] <Ingar> all your memory are belong to us
[09:42:29] -!- boru [boru!~boru@euje-95-549-084-092.dynamic.mnet-online.de] has joined #soylent
[10:20:12] <FatPhil> ~gday boru
[10:22:59] <FatPhil> We got a little package from one of our German clients last week - she was complaining that the local supermarket seems to have been taken over by a Russian chain, and you can't get a good Stollen any more (in particular, one with marzipan). Have you noticed any Russian interference?
[10:24:10] <FatPhil> You can tell Estonia's become posh - as Lidl has decided that there's now space at the bottom end of the market to muscle in.
[10:33:54] <boru> ~GrüßGott FatPhil
[10:34:08] <boru> And no, I can't say I've noticed that at all.
[10:34:22] <boru> ~grüßGott FatPhil
[10:34:26] <boru> Oh come on.
[10:34:48] <boru> The only supermarket where I've seen more non-German food products is Edeka.
[10:35:21] <boru> And I'd argue that real Germans (or at least, Bavarians) would go and get them at a bakery, rather than crap from a supermarket.
[10:40:35] <FatPhil> I think she's is rhine/hesse way
[10:52:50] <boru> Degenerates.
[11:22:16] -!- alexbst has quit [Quit: i know a guy who goes to shows, when he's at home and he blows his nose, he don't use tissues or his sleeve, he don't use napkins or any of these, he uses magazines, magazines]
[11:26:30] <Bender> [SoylentNews] - Doomsday Clock Will Stay at 100 Seconds Away from 'Midnight' Apocalypse - https://sylnt.us - Tick-Tock-Tick-Tock-Tick-Tock-Tick-Tock-Tick-Tock-Tick-Tock-Tick-Tock-Tick-Tock-Tick-Tock-Tick-Tock
[12:05:05] -!- alexbst [alexbst!lysander@thmnovybb.internationalconspiracy.org] has joined #soylent
[12:55:53] -!- AzumaHazuki [AzumaHazuki!~hazuki@the.end.of.time] has joined #soylent
[13:01:16] <c0lo> https://www.theguardian.com
[13:01:17] <systemd> ^ 03‘The perfect target’: Russia cultivated Trump as asset for 40 years – ex-KGB spy
[13:01:52] <c0lo> oh, wow, chromas, you got your hdd finally?
[13:02:51] <AzumaHazuki> all just a conspiracy, c0lo! It"s TDS from violent antifa liberals who hate america
[13:04:11] <c0lo> =submit https://www.scmp.com
[13:04:13] <systemd> Submitting "China’s frozen food imports being left out in the cold over virus fears"...
[13:04:34] <systemd> ✓ Sub-ccess! "03China’s Frozen Food Imports Being Left Out in the Cold Over Virus Fears" (0p) -> https://soylentnews.org
[13:04:48] <c0lo> =submit https://www.japantimes.co.jp
[13:04:50] <systemd> Submitting "Japan Times"...
[13:05:12] <systemd> ✓ Sub-ccess! "03Japan Times" (18p) -> https://soylentnews.org
[13:06:48] <c0lo> "China tests frozen fish for covid" and then "many Chinese shoppers grow reluctant to buy overseas food products after COVID-19 infections were reported among people handling such items"
[13:23:11] <FatPhil> any stats nerds out there - is there a name for a distribution that isn't uniform, but has a linear probability density over a range?
[13:25:35] <Bytram> FatPhil: Like: ___------------------__
[13:25:37] <Bytram> ??
[13:27:14] -!- Fnord666_ has quit [Read error: Connection reset by peer]
[13:28:40] <FatPhil> that's uniform - I want a slope to be possible in the range.
[13:28:48] * Bytram thinks of a band-pass filter
[13:29:03] <FatPhil> the closest I've found is a triangle: https://en.wikipedia.org
[13:29:04] <systemd> ^ 03Triangular distribution - Wikipedia ( https://en.wikipedia.org )
[13:29:16] <c0lo> "linear probability density"... the only thing I remember is the probability density represents the relative probability and its integral needs to be 1.
[13:29:43] <FatPhil> linear := p(x)=ax+b
[13:29:44] <Bytram> c0lo++
[13:29:44] <Bender> karma - c0lo: 24
[13:30:07] <FatPhil> uniform := p(x)=b
[13:30:15] <c0lo> So, get a segment over your domain and pick the constants so that the integral is 1.
[13:30:41] <FatPhil> with the mode at the bound, the triangle degrades to p(x)=ax
[13:30:50] <Bytram> it has been too many decades since my stats and probability classes in college -- not gonna be much help here, sorry!
[13:31:40] <c0lo> FatPhil: s/mode/node/
[13:31:40] <SedBot> <c0lo> <FatPhil> with the node at the bound, the triangle degrades to p(x)=ax
[13:31:52] <FatPhil> Yeah, I'm way past my prime on this topic. I sucked at it at uni,and that was decades ago.
[13:32:14] <FatPhil> c0lo: mode = maximum p
[13:32:14] <c0lo> a=2 would satisfy then
[13:32:36] <FatPhil> on [0,1], yes, but I want b>0, not b=0
[13:33:26] <c0lo> And you don't remember the formula for a trapezoidal area?
[13:33:28] <FatPhil> The equation relating a and b is easy! It's the properties, things like moments, that I don't want to have to work out myself.
[13:35:03] <Bytram> my textbook was SO bad that it would give examples and neglect to point out that it worked because there were [unstated] factors in the equation that made it work out that way, and that other values would cause things to be entirely different. I ended up buying a stats textbook for a *different* course which was more illustrative and rigorous and between the two was able to get a passing grade (maybe even a "C")
[13:35:14] <FatPhil> The end problem is that I'd like to know how to measure how different a distribution of random points is from such a distribution. My hypothesis is p(x)=ax+b for unknown a,b, and I want to know what a and b should be, and also whether I'm making a good approximation with this assumption
[13:35:29] <Bytram> FatPhil: So, back up a minute... what are you trying to do?
[13:36:24] <Bytram> Sounds like you are way down into a possible solution to something, but maybe not. So what is the situation at hand?
[13:37:43] <Bytram> Like: How do I set up a MySQL database? When all one has are 10 value pairs, and a simple array or hash table would suffice.
[13:38:09] <boru> You mean an X/Y problem?
[13:38:38] <Bytram> boru: you who?
[13:38:44] <Bytram> ;^)
[13:39:10] <boru> You, Bytram; I was trying to discern if that's what you meant, when trying to classify this sort of problem.
[13:39:37] <boru> Oh, it even has its own website now, apparently: https://xyproblem.info
[13:39:38] <systemd> ^ 03XY Problem
[13:40:31] <boru> It predates ESR, but it's also referred to in: http://www.catb.org
[13:40:32] <systemd> ^ 03How To Ask Questions The Smart Way
[13:41:45] <Bytram> oh. I'm just trying to understand what FatPhil was doing, and pointing out that he might be trying to (analogy time) solve a problem using MySQL (because that's what he heard/though about) when the actual problem at hand could be better solved with another approach... So what is the *actual* problem underlying all this that needs solving?
[13:42:04] <boru> Yeah, that's the gist of the XY problem.
[13:42:11] <Bytram> kewel
[13:42:31] * Bytram wanders off to process some stories for the site
[13:42:32] <boru> That said, FP is a smart chap, so I doubt it applies to him in this case.
[13:45:29] <FatPhil> I have O(1) stats on univariate O(N) data from an external source in buckets. I *presume* that almost all buckets will have approximately the distribution I mention (because they are slices from a larger distribution). I just want to know if I can measure how likely/accurate that presumption is.
[13:45:52] <FatPhil> I do have the ability to ask for more stats, but it must remain O(1).
[13:47:28] <FatPhil> at the moment I can easily get count, sum(x), and sum(x^2)
[13:48:08] <FatPhil> so I'm extrapolating from the density of a 1-d scatter plot to the actual density. But I never get to see the O(N) scatter plot, only its stats.
[13:48:16] <boru> Perhaps I am misunderstanding something, but how does big-O apply to statistics?
[13:48:31] <boru> Do you mean distribution in some sort of hash map?
[13:48:35] <FatPhil> That's the size of the data.
[13:48:42] <Bytram> FatPhil: So, maybe, code it with that assumption built in, as a *research* AND add a test to report when that assumption fails? Having gathered more data, the result should become apparent. no?
[13:49:09] <FatPhil> There's lots of 1-D data, but I only get a few values to describe that data.
[13:49:23] <FatPhil> So "just ask for all the data" is *not* a solution.
[13:50:20] <c0lo> "I have O(1) stats on univariate O(N) data from an external source in buckets." eff me sideways, I understand nothing of what you have.
[13:53:05] <c0lo> Build a histogram on those values with a bucket width of your choice, normalize the histogram, look at the normalzed histogram if it appears linear?
[13:54:37] <Bytram> try again. "All the data" exists *somewhere*. Gather that as RESEARCH data, ALONG WITH the the data you would have ordinarily provided to you. Play around with the O(1) data and test your method against the full data set. From the sounds of it, you are in a GIGO situation. I can well imagine the mapping of the scatter plot is many-to-1 to what you see in your O(1) extract.
[13:55:05] <FatPhil> The histogram is probably "too much data". Mean & SD are fine, histogram not.
[13:55:23] <Bender> [SoylentNews] - Chemists Settle Battery Debate, Propel Research Forward - https://sylnt.us
[13:57:15] <Bytram> In the histogram, for any given X... how many values of Y are possible? Could there be no value? Is it at most one value? Could there be multiple values?
[13:57:16] <FatPhil> HOwever, I do only need an approximation, maybe even 3 buckets would be enough.
[13:57:41] <Bytram> is X integer or continuous?
[13:58:00] <c0lo> https://en.wikipedia.org
[13:58:01] <systemd> ^ 03Standardized moment - Wikipedia
[13:58:02] <FatPhil> can be considered continuous, as it has a very wide range
[13:59:01] <Bytram> "range" as in a function that maps values from a domain to a range?
[13:59:33] <c0lo> The moments not gonna help you to decide if the prob function is linear.
[14:00:33] <FatPhil> the values can range up to the millions, easily. it's time some operation took in milliseconds.
[14:01:32] <FatPhil> What I want to do is similar to measuring skew, but that is for a bell-curve, which I know to be an inappropriate model.
[14:02:18] <c0lo> Bytram, histogram is "frequency of occurence within bucket intervals". so, sorta "count(*) where v between x and x+delta"
[14:02:36] <FatPhil> probably all I need is to know if the data's "peaky" or "spread out", so the SD probably is enough.
[14:02:59] <Bytram> so... there could be millions of values for any given X?
[14:03:28] <Bytram> =g rubber ducky debugging
[14:03:29] <systemd> https://en.wikipedia.org - Rubber duck debugging - Wikipedia
[14:04:07] <c0lo> FatPhil, do look on https://en.wikipedia.org You won't get what you want only with count, sum(x), sum(x^2).
[14:04:08] <systemd> ^ 03Standardized moment. - Wikipedia
[14:04:19] <FatPhil> yeah, I'll check there
[14:04:30] <Bytram> gtg, laters, and good luck!
[14:12:07] <FatPhil> I do have a little bit of extra knowledge that will help me and I think that might me enough to get by with just the current 2 moments. Each of the current triplets (count,mean,SD) I have corresponds to a sub-range of X, so I can see how each bucket relates to its neighbours too.
[14:23:17] -!- AzumaHazuki has quit [Remote host closed the connection]
[15:15:56] <Fnord666> FatPhil If you're expecting an even distribution across the buckets then I would think the spread between the high and low values would be useful.
[15:47:45] <chromas> Maybe try a belle-hooks curve :)
[16:03:27] * Runaway1956 shudders at the things he has learned about vacuum cleaners
[16:03:30] <Runaway1956> https://www.realclearscience.com
[16:03:31] <systemd> ^ 03The Dangers of Masturbating With a Vacuum Cleaner
[16:04:24] * Runaway1956 wonders how many of those vacuum cleaners were of the age of consent.
[16:10:04] <chromas> Ah yes, the dark side of the meat packing industry
[16:13:23] <chromas> =yt doofy cleaning my room
[16:13:24] <systemd> https://youtube.com - Deputy Doofy Scary Movie (06:25; 419,085 views; 👍1,713 👎72)
[16:27:40] <Bender> [SoylentNews] - New Biosensors Quickly Detect Coronavirus Proteins and Antibodies - https://sylnt.us - shining-a-light-on-coronavirus-detection
[16:42:01] <FatPhil> Fnord666: I don't have the spread, alas. The raw data as it comes in is ephemeral, I can only accumulate very limited stats on it.
[16:48:44] <chromas> Did you consider a spherical cow?
[16:53:57] <Fnord666> LOL
[17:10:12] <FatPhil> all cows are presumed to be trapezoid
[17:16:57] <FatPhil> #weather
[17:16:59] <MrPlow> Apteegi, 10123 Tallinn, Estonia - Today: "Possible light snow (1–3 in.) in the afternoon." 32/24F, Humidity: 93%, Precip: 54%, Wind ~12mph. Sat: "Snow (2–4 in.) in the morning." 25/22F, Humidity: 94%, Precip: 54%, Wind ~12mph. Sun: "Foggy throughout the day." 25/21F, Humidity: 95%, Precip: 39%, Wind ~8mph.
[17:18:46] <c0lo> https://www.nytimes.com
[17:18:48] <systemd> ^ 03Opinion | Marco Rubio Deserves Ivanka Trump
[17:52:12] -!- Alfred has quit [Quit: Leaving]
[18:42:55] -!- SwahiliWilli [SwahiliWilli!~Agorophob@172.83.sv.vho] has joined #soylent
[18:56:33] <Bender> [SoylentNews] - The Cloak-and-Dagger Tale Behind This Year's Most Anticipated Result in Particle Physics - https://sylnt.us
[19:52:57] <Teckla> Does anyone here check Usenet using Google Groups?
[19:55:13] <FatPhil> googoogroups stopped working in my browser about 15 years ago, so no. Proper news server for me.
[20:00:57] <Teckla> FatPhil: If you don't mind saying, what news server do you use?
[20:01:24] <Teckla> Google Groups won't let me access comp.lang.c due to it allegedly having malware/spam/etc.
[20:01:39] <Teckla> There's no even an, "I know what I'm doing; let me in anyway" option.
[20:01:45] <Teckla> I suppose I should just use a proper news server.
[20:02:10] <Teckla> Like in the old days. And I could even use tin, like the old days, for a proper retro nostalgia. :)
[20:05:06] <halibut> I agree with what some others have said. A nonuniform-yet-linear distribution is an odd thing to try for a distribution.
[20:05:11] <halibut> However, some basic properties of one:
[20:05:11] <FatPhil> I suspect this is the config file it's reading when it starts: (setq gnus-select-method '(nntp "reader.eternal-september.org"))
[20:05:34] <halibut> Let p(x) = (ax + 1) / 2 for -1 <= x <= 1, and p(x) = 0 elsewhere.
[20:05:51] <Teckla> ^ s/no even/not even/
[20:06:01] <halibut> (For a symmetric distribution, b must be 1/2 for p(x) to be normalized.)
[20:06:16] <Teckla> FatPhil: That's a popular one. Thanks.
[20:06:18] <halibut> <x^(2n)> = 1/(2n + 1)
[20:06:29] <halibut> <x^(2n+1)> = a/(2n + 3)
[20:06:34] <FatPhil> halibut: I know that it should ahve that distribution, as it's a subsample of a larger very messy multimodal distribution.
[20:06:39] <halibut> |a| <= 1/2 (or p(x) becomes negative)
[20:06:47] <Teckla> As a pre-Eternal September Usenet user, I remember the glory days fondly...
[20:07:12] <FatPhil> did you celebrate teh 10000th of September a week or so back?
[20:07:28] <FatPhil> alias aoldate='echo $((`date +%s`/86400-8643))'
[20:07:43] <FatPhil> $ aoldate
[20:07:43] <FatPhil> 10013
[20:08:59] <Teckla> Heh, nope.
[20:09:24] <FatPhil> halibut: yup, that's the mode at limit case for the triangle distribution I mentioned at the start.
[20:09:33] <Teckla> I haven't derived any value from Usenet in a very long time. I'm not even sure why I still periodically check it.
[20:09:58] <FatPhil> rasf1's pretty active (formula 1 chat)
[20:10:20] <FatPhil> lots of old-school types there, "forums" are just not their thing.
[20:10:46] <halibut> The |a| = 1/2 case is a maximally-asymmetric triangle distribution. The a=0 case is uniform. The rest of the range is something in between.
[20:11:19] <FatPhil> there are absolutel fucking idiots there, my killfile's hits about 80% of the posts, but there's no spam or trolls in the conventional sense.
[20:11:45] <FatPhil> halibut: ah, yes, sorry, yeah, the limit cases are the triangles.
[20:12:52] <FatPhil> did you use a symbolic algebra package to calculate the expectations, or are they from some published table?
[20:12:55] <halibut> The standard triangle distribution is symmetric. The |a| = 1/2 case here is not symmetric.
[20:13:08] <halibut> It's a pretty simple integral. I just did it by hand.
[20:13:15] <FatPhil> hence "mode at limit"
[20:13:33] <FatPhil> jebus, I can't even integrate a constant nowadats
[20:13:40] <halibut> p(x) = (ax + b) / 2. Compute int(x^n * p(x), x=-1 to 1)
[20:13:59] <halibut> Polynomials are very simple. x^n => x^(n+1)/(n+1)
[20:14:29] <FatPhil> why x^n*p(x)? shouldn't it be p(x)^n?
[20:14:50] <FatPhil> what's your <> notation represent?
[20:15:11] <halibut> It's pretty rare to need to take a power of a probability distribution.
[20:15:32] <halibut> <X> is the notation for the expected value of X.
[20:15:37] <FatPhil> herp derp, yup
[20:16:03] <halibut> Sounds like you do not need further explanation.
[20:16:05] <FatPhil> yeah, p(x) * x^n does make sense
[20:17:09] <halibut> For large N, sum(x^n)/N is a decent estimator of <x^n>.
[20:18:08] <halibut> For smaller N, there are frequently some subtle correction to make it unbiased.
[20:18:38] <Teckla> FatPhil: Which newsgroups do you subscribe to? Again, if you don't mind saying. :)
[20:18:39] <FatPhil> I was looking at some info about bimodal cases, and it looks like skew and kurtosis are useful.
[20:18:54] <FatPhil> Teckla: the only ones I'm "active" in are rasf1 and alt.anagrams
[20:19:18] <halibut> Yes. You can either compute them directly or, if you only have sum(x^n)/N, you can typically compute them from those sums as well.
[20:19:55] <halibut> mu = sum(x) / N, sigma^2 = sum(x^2) / N - sum(x)^2 / N^2, and so on.
[20:20:24] <Teckla> FatPhil: Ah, neither of those are in my wheelhouse...
[20:20:52] <FatPhil> I read sci.crypt and comp.compression too, but haven't posted for years
[20:21:04] <FatPhil> again, killfile is at about the 95% level!
[20:21:06] <halibut> The formulas for skew and kurtosis may be expanded into polynomials, and then <x^n> just gets replaced with sum(x^n)/N, and the mu and sigma in the formulas are given above.
[20:21:15] <FatPhil> in particular in sci.crypt
[20:22:25] <halibut> Correction: sum(x^n)/N is always an unbiased estimator of <x^n>. I was getting that confused with variance estimators.
[20:23:41] <FatPhil> I'm not stats-savvy enough to know if I should be doing /n or /(n-1), as i am just sampling, so I think the latter should be used.
[20:25:07] <halibut> Divide by N. Dividing by N-1 is the corrected way for estimating standard deviation, not <x^n>.
[20:26:16] <FatPhil> then again, I only need a rough estimation of how the points are distributed. I think I have enough ideas to build a prototype now. I just need to find an appropriate source of noisy data (which I can fake if I can double-guess what the real world will provide)
[20:27:58] <halibut> I think the best way is to make sure N is large enough that the difference between N and N-1 are not important.
[20:28:51] <halibut> If you are trying to estimate <x^n>, then dividing by N is correct. Unfortunately, if you are trying to estimate variance, skew, kurtosis, or any function of <x^n>, then the formulas are more complex.
[20:29:11] <halibut> I only happen to know the variance, because it is simple enough that it reduces to ``divide-by-(N-1).''.
[21:27:00] <Bender> [SoylentNews] - 635 Million-Year-Old Fungi-Like Microfossil that Bailed Us Out of an Ice Age Discovered - https://sylnt.us
[22:36:50] * TheMightyBuzzard yawns
[22:36:50] * MrPlow flips a Skittle into TheMightyBuzzard's gaping mouth
[22:38:15] <TheMightyBuzzard> paint sprayer today. i couldn't see skin on my left arm from t-shirt sleeve to wrist when i got home.
[22:54:05] -!- SoyCow2656 [SoyCow2656!~bdae75e6@189.174.grz.yng] has joined #soylent
[23:02:01] <pinchy> The hedge fund manager cries out as he shorts you
[23:06:41] -!- SoyCow2656 [SoyCow2656!~bdae75e6@189.174.grz.yng] has parted #soylent
[23:27:23] <c0lo> =w Triangular distribution
[23:27:24] <systemd> In probability theory and statistics, the triangular distribution is a continuous probability distribution with lower limit a, upper limit b and mode c - https://en.wikipedia.org
[23:28:22] <c0lo> =w Trapezoidal distribution
[23:28:23] <systemd> statistics, the trapezoidal distribution is a continuous probability distribution the graph of whose probability density function resembles a trapezoid. Likewise - https://en.wikipedia.org
[23:30:15] <c0lo> FatPhil, you have ^^^ the formulae for the moments (and triangle is a special case of trapezoid)
[23:32:45] <halibut> Those distributions are have a slope zero at the top. I think he was looking for one that had a non-zero slope at the top, closer to a triangular distribution, but shifted vertically.
[23:32:54] <halibut> (and then renormalized)
[23:46:25] <FatPhil> yup, a vertical slice from a very multimodal distribution
[23:48:11] <c0lo> halibut, just make the extent of the zero-slope get to zero. In their notation, make b == c
[23:48:47] <c0lo> =submit https://slate.com
[23:48:49] <systemd> Submitting "General Motors Says It Will Stop Making Gas-Powered Vehicles by 2035"...
[23:49:11] <systemd> ✓ Sub-ccess! "03General Motors Says It Will Stop Making Gas-Powered Vehicles by 2035" (5p) -> https://soylentnews.org
[23:55:29] <Bender> [SoylentNews] - As More Bacteria Grow Resistant To Antibiotics, Scientists Are Fighting Back - https://sylnt.us