User talk:Glrx

From Wikipedia, the free encyclopedia
Jump to: navigation, search


Welcome!

Hello, Glrx, and welcome to Wikipedia! Thank you for your contributions. I hope you like the place and decide to stay. Here are some pages that you might find helpful:

I hope you enjoy editing here and being a Wikipedian! Please sign your messages on discussion pages using four tildes (~~~~); this will automatically insert your username and the date. If you need help, check out Wikipedia:Questions, ask me on my talk page, or ask your question on this page and then place {{helpme}} before the question. Again, welcome! RayTalk 19:29, 2 August 2009 (UTC)

Edits to pi[edit]

re: this addition of περίμετρος and some nearby edits

Thanks for adding some real Greek. Even though you claim to know only 3 digits, you at least know more than 3 letters... Imaginatorium (talk) 17:57, 21 March 2016 (UTC)

Reed Solomon article[edit]

This section needs improvement. It's basically describing the case where a codeword is a set of values, and in the example case, includes the message unmodified as the initial set of values of a codeword. Then it goes on to state that this is somehow optimized without mentioning the change to consider a codeword as a set of coefficients. Even though there's an equivalence in the codewords as mentioned later on in the article, treating codewords as values results in inefficient decoding compared to treating codewords as coefficients (BCH type methods). Rcgldr (talk) 02:55, 30 March 2016 (UTC)

@Rcgldr:
Yes, it needs work. I looked at the edit/comment=out earlier today, but I didn't know what to do. The addition skips over the finite field issue (values of f(x) are big otherwise), but that is not a big fib. It uses the original view rather than the modern view (views are explained lower down in the article). Reverting to previous isn't better. Blowing away the section didn't seem right because it was trying to give a simple and intuitive description of how RS works. It became one of many tabs on my browser to revisit.
Right now, I'm tempted to move the text in § Basis down to § Construction and rename "Construction" to some better title. For most readers, the important thing is that it is an ECC family, it has flexible coding properties, and where it is used. After that, the article can get more involved.
Sigh. I wish User:Nageh were still around. Glrx (talk) 03:25, 30 March 2016 (UTC)
On the talk page, I mentioned that the basis section seemed redundant to the construction section. It's not quite the original view: in the original view, the message elements are considered to be coefficients of a polynomial, and used to generate the codeword values for a[1] through a[n]. The basis section shows an intermediate view, where the message itself is part of the codeword and a method like Lagrange interpolation used to determine the polynomial that generates the codeword values. In both cases, a polynomial of degree k-1 is usually involved, a[1] through a[n] are known to both encoder and decoder, and decoding involves trial and error of a method like Lagrange interpolation to try combinations of n elements taken k at a time until a match is found. The switch to treating a codeword as a set of coefficients and using a generator polynomial of degree t-1 known to both encoder and decoder is what made RSECC practical.
On a side note, just about every other article related to RSECC describes the number of redundant symbols as 2t instead of just t. However that can be awkward in the case of that the number of redundant symbols is odd (the old floppy tape drives used a matrix where each column was a (29,32) codeword. Each row had a 16 bit CRC used to detect erasures, so 3 erasures could be corrected.
Rcgldr (talk) 03:54, 30 March 2016 (UTC)
Yes, you are right about the coeffs. I just keyed on sending the values. Glrx (talk) 04:15, 30 March 2016 (UTC)

Rather than create a new section, I made this section more general. In the RS history section, it mentions an efficient decoder was not known when RS codes were first introduced (1960). It then goes on to mention an efficient decoder algorithm was determined in 1969, but fails to mention that this required changing the encoding process to use generator polynomials of degree t-1 and considering codewords as a set of coefficients. I'm wondering if the change to the encoding process predates Berlekamp and Massey (1969), or if it was part of their discovery. Rcgldr (talk) 05:06, 30 March 2016 (UTC)

The encoding process had to be changed in order for a Peterson like decoder to work. The Peterson decoder for BCH codes was developed in 1960, but I can't find when RS encoding was changed to allow a modified Peterson decoder to work.
After correspondence with Dave Forney, I cleaned up the history section. Rcgldr (talk) 03:55, 2 April 2016 (UTC)

I did find the credited author (Yasuo Sugiyama in 1975) for the adaptation of the extended Euclid algorithm and updated the history section to reflect that. Rcgldr (talk) 06:04, 30 March 2016 (UTC)

I updated the history section to reflect all of this. Rcgldr (talk) 08:47, 30 March 2016 (UTC)
  • re: theoretical decoder

I moved this from error correction algorithms to the codeword viewed as a sequence of values section since it's not a practical decoder. I added a noted that all of the decoders in the error correction algorithms use the BCH view. Rcgldr (talk) 08:47, 30 March 2016 (UTC)

  • re: equivalent views - regarding the transform equation for p(x), I find that is working without the 1/n factor as shown in the article: . Rcgldr (talk) 03:53, 2 April 2016 (UTC)
@Rcgldr:. I appreciate your efforts on these subjects, and I envy your correspondence with Forney. There's are some nagging notions about RS/BCH that have been in the back of my mind, but I haven't had time to look because I'm buried in my real life and have other distractions in my WP life. RS, BCH, and several other coding topics are on my watchlist, so I will see them eventually. Glrx (talk) 18:00, 8 April 2016 (UTC)

Lucy spy ring[edit]

re: Xyl 54's edits inserting interlanguage links
re: my edits removing interlanguage links

Hello
You recently reverted some changes I made to this article, but your edit summary didn't say why (more of an observation, really; and one which implied you'd put the links in, rather than taking them out). I've opened a discussion at the article talk page, if you care to comment. Regards, Xyl 54 (talk) 21:22, 30 March 2016 (UTC)

See WP:Manual of Style/Linking#Linking which states, "To avoid reader confusion, inline interlanguage, or interwiki, linking within an article's body text is generally discouraged." Glrx (talk) 22:51, 30 March 2016 (UTC)

I think I probably handled that poorly[edit]

It seems to me quite probable that I owe you an apology, and if so, I do apologize. You did nothing wrong, and if anyone did, it was me. I'm actually and truly not at all sure how to move forward, but I'm sure that "slow and steady" is a part of the correct equation.  Lingzhi ♦ (talk) 12:07, 4 April 2016 (UTC)

@Lingzhi:
re: Talk:Vincent van Gogh#linking the inline citations to the reflist
re: Lingzhi's revert
re: Lingzhi's reversal
I'm not sure what to say here. Should I let it slide or get real?
Your email was appalling. The bullet is you reverted an hour's worth of my work not because you thought it was the right thing or that I'd misread the talk page consensus but rather because you thought it might please a third party. And you weren't even sure it would please that party; you were only guessing. Editors are grown ups; if they want something, then they should say so on the talk page. Your decision wasn't anything about improving WP. Your sudden reversal to {{cite book}} is even more surprising because at one time you offered to do the edit.
I understand it is difficult to offer up an apology, but I'm taken back by it's conditional nature. See non-apology apology. I read the "slow and steady" as you want me to delay even though your email suggested my edit would be the eventual result. You've accused me of stepping on another editor and misreading consensus. You blew away my work. Yet you claim that you are not sure.
I don't buy that you are "not at all sure how to move forward". I think you know exactly what you should do.
I admire User:John's diplomacy on the talk page.
Glrx (talk) 19:01, 8 April 2016 (UTC)
  • No, I don't know what I should do. If I knew what I should do, I would certainly do it. Improving WP is not always and everywhere the only or even main goal (though in dry theory, of course it should be). If improving WP were the only goal, then I would have sent you a "thank you" ping and then forgotten about it. :-) The reason that I accused you of stepping on another editor is because.. the timing was so exactly, exactly, perfectly wrong. It looked deliberate to me, and I was quite surprised when you claimed innocence... (oh, and btw, if you wanna say I didn't WP:AGF, you'd be right, but my WP:AGF filter was just overwhelmed by the timing thing)... As I have said before, if you want to think I'm an a**hole, please be my guest, but from my perspective I am (or at least, was) caught between Scylla and Charybdis and not clever enough to know how too swim through without getting wet... If this helps, think about it this way: what do I gain from anything I did? What was my motivation? Then consider the possibility that confused people do confusing things. But in any event, I am of course sorry if I made you feel unvalued. That's all.... pps, I have restored your one hour's worth of work. Best wishes in all things.  Lingzhi ♦ (talk) 02:23, 9 April 2016 (UTC)

Intel 4004[edit]

re: my revert at Transistor–transistor logic

What logic levels are the 4004 really? Bytesock (talk) 23:03, 9 April 2016 (UTC)

The i4004 uses PMOS logic; see http://www.intel.com/Assets/PDF/DataSheet/4004_datasheet.pdf. Sadly, the WP PMOS article has many problems. Internal voltage swings could be close to 15 V, but I don't know.
Logic families are not determined by logic levels; it's the technology used to implement the logic that counts. TTL uses bipolar transistors and an unusual multiple emitter input circuit. The classic logic line is TI's 7400 series. There are speed-power variations such as 74L00 and 74H00. There are variations of TTL such as 74LS00 and 74S00 that are TLL circuits with Schottky clamps. Other logic lines (such as 74HC00) implement the same logic functions with compatible logic levels but use a different logic family (such as HCMOS).
Glrx (talk) 23:39, 9 April 2016 (UTC)
I improved the PMOS article. Interesting that families are defined by process technology and not voltage levels. Which are the ones that will dictate what a designer can do without cumbersome signal conversions. And then what functions that a voltage compatible family offers. The WP revert message kind of shocked me :-) so had a look at the datasheet on the 4004 and it sunk in that it was something completely different than one expects, those voltage ranges are really asymmetrical and high. It must been messy to design with PMOS parts. Bytesock (talk) 03:00, 10 April 2016 (UTC)
By the time of the i4004, TTL was the dominant logic family (RTL and DTL were disappearing; ECL was the high-speed market and CMOS the low-power low-performance market).
There's some symettry in the RTL, DTL, and TTL names; the first letter is the device responsible for the logic, and the second letter is the device for inversion and gain.
Voltage levels are not enough. Current source/sink are also important (loosely fanout). There are also issues with OC outputs vs. totem pole outputs. It is not a simple abstraction. Transition times are also important; many TTL designers came to grief because they ignored transmission lines. Glrx (talk) 03:42, 10 April 2016 (UTC)
The only factor that really mess up designs tend to be the voltage levels however. And the transition time. The rest is more about paying attention than anything else. Ie, less fanout, use less inputs without amplification. If using amplification calculate propagation times etc. Bytesock (talk) 14:02, 10 April 2016 (UTC)

Regarding the PMOS voltage levels. Are they like this?

Symbol Min Max Unit
VSS +15-5% +15+5% V
VIL VDD VSS−5.5 V
VIH VSS−1.5 VSS+0.3 V
VOL VSS−12 VSS−6.5 V
VOH VSS−0.5 VSS V

4004 datasheet Is it reasonable generic to interface with random PMOS logic chip? Say if you wanted to read out a Intel 4001 ROM or control 4003 I/O chips? Bytesock (talk) 14:02, 10 April 2016 (UTC)

I doubt PMOS logic levels were standardized. A 15-V i4004 probably would not want to be connected to a 24-V metal-gate PMOS part. The Intel 4004 family of parts could be connected together. It looks like the i4008/9 were used to translate the i4000 series logic levels to standard TTL levels. I think TI (and other companies) had logic level translators; it's not hard to do with bipolar transistors. Glrx (talk) 17:36, 10 April 2016 (UTC)

Good catch[edit]

I accidentally selected "Central America" when I meant "Central Asia". It's fixed now. – Muboshgu (talk) 18:08, 10 April 2016 (UTC)

Music_OCR revert[edit]

re my revert at Music OCR

Hi there!

Can you please help me understand why you removed my addition to the Proprietary software section? I tried to follow the existing structure and don't see how the newly added line is different to the old ones, e.g. the PhotoScore one. — Preceding unsigned comment added by DrAlienDragon (talkcontribs) 18:11, 26 April 2016 (UTC)

You have 3 edits on Wikipedia. The first two edits are apparent advertisements for the "Sheet Music Scanner" phone app; your third edit is to this talk page.
I see the mention of "Sheet Music Scanner" as being WP:UNDUE. WP wants secondary sources that say the application has merit. I googled for SMS and only found its iTunes listing.
WP does not want to be a buyers' guide.
SMS does not fit the Music OCR topic. It reads and plays from cameraphone input. There's no indication of how much music notation it needs to parse to achieve its goals. There's no indication that the OCR output is available in machine-readable or even printable form. I googled PhotoScore; that application has been around for 22 years and offers editable output; it also reads lyrics. Compare also with Audiveris.
I'm only a lowly editor around here. If you still think SMS should be included, then you are free to follow WP:BRD by creating a new topic on Talk:Music OCR that advocates the inclusion of SMS in the article. If there is a WP:CONSENSUS to include it, then it goes into the article.
Glrx (talk) 19:01, 26 April 2016 (UTC)
Thank you for your answer. With all due respect, I don't see how the number of edits is relevant to the fact that I've built an OMR engine as proprietary software that I'm trying to list here.
I am not sure what you mean by not fitting the topic - to play something from camera input means doing full OMR on the image the same way as any other software on this page.
"There's no indication that the OCR output is available in machine-readable or even printable form. "
I don't see how the OMR topic implies that the output should be machine readable or printable (but my app does enable export to MIDI as you can see from the website). Even the OMR page says: "... to interpret sheet music or printed scores into editable or playable form." - my app plays the music, it is playable.
"There's no indication of how much music notation it needs to parse to achieve its goals."
The list of recognized symbols is listed on my website: treble, bass, and alto (viola) clefs, notes, duration dots, rests and accidentals, which is enough to play most scores. Btw. there are some youtube videos of the app in action.
Regarding being a buyer's guide, the link to PhotoScore leads to their webshop, the same way as my link, I really don't see the difference. If you consider the link to the app's website advertisement, you should remove the other links too IMHO. My app wants to be an affordable alternative to fairly expensive professional software that is linked on this page. Currently, it is the only mobile alternative to NotateMe (Neuratron) that actually works without a scanner.
— Preceding unsigned comment added by DrAlienDragon (talkcontribs) 20:10, 26 April 2016‎
An editor with few edits that focus on a particular subject is a WP:SPA. That's not necessarily a bad thing, but it does raise some questions about the editor. In particular, it suggests the editor may not be familiar with Wikipedia policies yet. It also suggests that the editor may not have a neutral point of view.
Above, you indicate that your purpose here is to list your software on Wikipedia. Your goal is to advertise your product. Wikipedia does not want to be an advertising forum. See also conflict of interest. Editors need to be very circumspect when writing about their work.
I see a large difference between the capabilities of your app and PhotoScore's. But my view is not important. Wikipedia wants WP:reliable sources to show a product's significance. PhotoScore has a colorable claim because it is in Byrd's list.
There may well be other links that should be removed from the list (notice the section is tagged as using primary refs), but the existence of other, similar, links is not an argument to include your link: WP:OTHERSTUFFEXISTS. If an entry is challenged, then reliable sources (rather than similar links) are needed to support the entry.
Glrx (talk) 01:14, 27 April 2016 (UTC)

Lucy spy ring, again[edit]

Hello
I am just checking to see whether you still wish to be part of the discussion on this page; I replied to your comments there a while ago but have not yet seen a response. I left it a while, as you were busy elsewhere, but I would like to resolve this before moving too far on myself. Regards, Xyl 54 (talk) 22:31, 30 April 2016 (UTC)

@Xyl 54: My position hasn't changed from 31 March. I don't view the Bureau Ha or Deutscher Herrenklub links as notable, but if you want to include them as {{ill}} links I won't object. Glrx (talk) 19:07, 5 May 2016 (UTC)
Firstly, my apologies for the lateness of this reply.
I appreciate you haven't changed your position; the purpose of my reply over there was to point out our position is contradictory. You cannot argue that HK and BH are non-notable on the en WP (which would suggest using plain text and an inline link) while at the same time insisting the Ill template be used (which would of necessity create a redlink) Nor can you reasonably reject adding useful information (to an encyclopaedia!) that has been requested, because you personally object to the formats (neither of which has any absolute prohibition in the guidelines) to be used.
So, do you have a better way of presenting this information (one that achieves what I was intending, while at the same time ticking all your own boxes), or are you prepared to accept a remedy that is good enough, in the absence of that something better.
Otherwise it feels like you are simply reserving the right to object at a later stage, and I would like to resolve this matter as amicably as possible. Regards, Xyl 54 (talk) 22:50, 12 May 2016 (UTC)
I'm lost.
I don't think the topics are notable on en:WP, so they don't need links. The article explains Bureau Ha as an intelligence front. Herrenklub seems unimportant to the Lucy article.
I don't do much editing on de:WP, but
Consequently, I'm for no links. I don't see the links as additions of "useful information" but rather links to irrelevant details in a foreigh language. Encyclopedias are not intended to include all knowledge.
You don't place either topic high on notabily and seem to accept {{ill}}: "As for the Herren Klub, I wouldn't have rated it any more notable than BH, (and if there's one thing the en WP doesn't need it's yet another article on another bunch of Nazis) but if we are going to redlink it I would favour the format you used above".
Glrx (talk) 17:58, 15 May 2016 (UTC)
So, here we are again.
Allow me to recap: I want to put a couple of links into this article, in order to "increase readers understanding of the topic at hand" and to "help the reader find related information", and because the terms in question are "proper names that are unlikely to be familiar to readers". And as there is no appropriate article on the en WP, I intended to use an interwiki link to somewhere that does have the information (in this case the de WP). These could be inline links (viz Herren Klub (de), Bureau Ha (de) [or Bureau Ha (de), using the redirect there]) which you have reverted before; or we could use the Ill template, which you suggested (viz Herren Klub (de), Bureau Ha (de) [or Bureau Ha (de)]) but have now objected to, though none of them are prohibited.
I've also wanted to remove the link currently in the article as a pipe from Bureau Ha (per WP:EGG), as the piped article (Swiss intelligence agencies) has not, and won't have, any mention of the Bureau; You've insisted on keeping it, despite your stated objection to EASTEREGGs.
I have also asked you for suggestions on how to link in a way that you don't object to; your only suggestion was not to have any links at all.
So, do these objections of yours extend to an intent to revert any of these links if they are put in? Because if so, we will need to get a third opinion, to resolve this (and I cannot believe this issue is having to go the full 15 rounds and then to a judges decision!) Xyl 54 (talk) 23:24, 31 May 2016 (UTC)
I had to go back and reread the article.
By and large, Deutsche Herrenklub and Bureau Ha are irrelevant to the Lucy spy ring. The article seems to mislead when it suggests that Rudolf Roessler ran Lucy. Roessler was a mailman selected by Thiele and others to disseminate the information. Roessler was not recruiting agents in Germany or elsewhere. Thiele, Gersdorf, Fellgiebel, or others were running Lucy. Rado was running a Soviet ring and hooked up with Roessler, but it's not clear whether Rado was an open intel officer (Switzerland was full of them) or clandestine. It also sounds like the British used Roessler as a mailman to reach Rado.
None of that makes DH or BH interesting. Maybe the principals of Lucy knew each other through DH, but Lucy is still a well-kept secret, so we don't know. Roessler's ties to BH/Swiss Intell would mean that the Swiss got copies of the information, but the article suggests they were passing it to the British. Maybe the Swiss also passed stuff to the Germans.
Lucy's known successes helped the Soviets more than the British or the Swiss.
I still oppose links to foreigh language wikis. The foreign targets do not offer insight into Lucy. DH is all over the map. Hitler met principals in private but denounced in public. DH seemed to be against Marxism but Lucy benefitted the USSR. The article used a friend as a cutout to Bureau Ha. The German article on Masson does not mention BH. Masson was head of Swiss Military Intelligence during WWII, he was getting Lucy's output, so the details of the Roessler to Masson link is a minor detail. The de.WP article on Roessler equates Swiss Military Intelligence to BH and is done with it. BH was run by Hausamann, but Hausamann is not mentioned in the current article. The de.WP for Hausamann just labels Bureau Ha as a conduit, so it is not more informative than the current en.WP article. Masson, Hausamann, and BH were just conduits for Lucy's information.
You are welcome to open a 3O, but please get a good idea of what the de.WP articles say so you can tell people why the links would be important.
Glrx (talk) 20:56, 4 June 2016 (UTC)
And again!
I see you have re-read the article, though if you think
.a) Roessler was simply the mailman
.b) Lucy was a British op to pass information to the Soviets
.c) the Swiss were passing Lucy product back to the Germans
then I suggest you read it again. (tho' if the article says Roessler was running Lucy in Germany then that needs changing, but I didn't put that there; I only re-wrote the History section. It's also incorrect that Roessler worked for Masson at Bureau Ha; that's been added since)
And there may be a lot of stuff we don't know, but what we do know is there, and is well-documented in reliable sources.  If you disagree, I would suggest you find some sources that say different.
Also, you can dismiss the detail as minor if you like; all details are minor if you have no interest in a subject.
As for getting a good idea of what the de articles say, I'm well aware of that. And I've already said why I want to put the links in; there is (marginally) more information there than there is here, and more than needs to be added to this article. Contariwise, are you clear your objection isn't simply a case of not liking something?
Anyway the 3O request is in now, if you care to comment. Xyl 54 (talk) 22:36, 27 June 2016 (UTC)

Ionization[edit]

re: AWB edit1 and AWB edit2
earlier, 29 May, edit that I also reverted

In order for author names to be incorporated into metadata, they need to be separated. One way of doing this is use Vancouver style, since the 'vauthor' parameter is automatically parsed. You should not be removing that parameter, unless you intend to replace it with alternative means of indicating individual authors. --Dcirovic (talk) 19:28, 4 June 2016 (UTC)

It is inappropriate to introduce a new citation sytle onto a page that is already using other styles. WP:CITEVAR. The page was not using Vancouver style. Although I recognize the value of splitting authors, you are running a script that does not follow the set style on the page. I see a lot of that on my watch list. Glrx (talk) 19:48, 4 June 2016 (UTC)

June 2016[edit]

You currently appear to be engaged in an edit war. Users are expected to collaborate with others, to avoid editing disruptively, and to try to reach a consensus rather than repeatedly undoing other users' edits once it is known that there is a disagreement.

Please be particularly aware that Wikipedia's policy on edit warring states:

  1. Edit warring is disruptive regardless of how many reverts you have made.
  2. Do not edit war even if you believe you are right.

If you find yourself in an editing dispute, use the article's talk page to discuss controversial changes; work towards a version that represents consensus among editors. You can post a request for help at an appropriate noticeboard or seek dispute resolution. In some cases it may be appropriate to request temporary page protection. If you engage in an edit war, you may be blocked from editing. ElKevbo (talk) 21:35, 4 June 2016 (UTC)

Good grief. A template for two reverts that restored material that had been around for 5 months? The first removal was done without comment by an IP.
On 28 December, Mackey el Capitain introduces sourced text into article on University of Phoenix that says UoP gives credit for corporate training at some companies. The edit gives a long list of companies that includes Fortune 500 companies. It is reasonable that such companies would have corporate training programs that are serious and could carry academic weight.
ElKevbo edits article several times since that edit, but does not remove that material.
On 1 June, an IP edits the addition by deleting the example companies; the IP does not give an edit comment about why the list was deleted. The IP's edit makes the article's text vague: it says 300 companies but without specific examples, the reference loses meaning.
On 4 June, I revert to restore the company list.
Four hours later, Orangemike reverted me with the edit comment "this list is cherry-picked by the subject's PR people for maximum prestige; see WP:NONCONTAGIOUS" [sic]. The WP:NOTCONTAGIOUS is not apropos. UoP is not claiming WP:N by using these company names. UoP is already WP:N.
I revert Orangemike stating "Undid revision 723713809 by Orangemike (talk) Not abt notability; I don't like UoP, but the Q Is whether list of examples is accurate."
ElKevbo reverts with the comment, "please don't edit war" (devoid of any comment about the content) and templates a regular.
Glrx (talk) 23:14, 4 June 2016 (UTC)

your goodness of fit question[edit]

RE: Talk:Goodness of fit#Error in example?

Hi, sorry it took me more than 1 year to get back to answering your reply to my question about the goodness of fit article. I can no longer find any missing reference, but the statement that chi squared red< 1 means a model is over-fitting (although it is stated as a rule of thumb) in the article is clearly false without some assumptions. You correctly said there can be measurement error, but on the talk page I mentioned that, for instance, for the ideal gas law, the measurement error due to something like brownian motion would be very small, and chi squared red would be very near zero. For another example, dropping a stone from a particular height in gravity and measuring when it lands, would give a chi squared near 1 if you purposely give the wrong value of the acceleration of gravity (so a purposely slightly low or slightly high value of g gives a better model, but the correct value of g is an over-fit?). Maybe it is a good 'rule of thumb' for particular types of statistical models, and particular assumptions about randomness of data, but if these aren't included in the discussion then the article only makes sense to people already using a particular type of statistics, and is wrong to include in a general encyclopedia in that form. Createangelos (talk) 09:35, 28 June 2016 (UTC)

To try to clarify the example, suppose someone decides to fit the distance an object falls as a function of the time, and wishes to fit it to a function of the form distance=1/2 a t^2. Here a is unspecified, they are going to make ten measurements to try to find the value of a. I'm not sure what they will say is their number of degrees of freedom, but anyway that is fixed. They will, if they do the measurements carefully, find a value of chi squared red very near zero. But if they change the exponent, trying distance = 1/2 a t^2.01, say, then since it is no longer possible to fit the parabola exactly no matter how accurately their measurements are done, they will get a larger chi squared reduced. If it is not near enough to 1, they can try d = 1/2 a t^2.02 and so-on. Eventually, the model will be a bad enough fit that they have achieved chi squared red =1. They could also get chi squared red to be 1 by decreasing the exponent using d = 1/2 a t^1.9 and so-on. So that if you know what you are doing, you can intentionally introduce errors, making the model intentionally inaccurate, to achieve chi squared red =1. Now, you might say that you shouldn't do it in this situation, where you are doing it by varying a parameter away from what it should be. But this issue certainly could arise in ways disguised, where different models of falling objects are somehow available with different exponents 1.8, 1.9, 2.0, 2.1, .. where the experimenter does not realize that this is what differentiates his models. He will choose not the best model, but one which is distorted.Createangelos (talk) 09:48, 28 June 2016 (UTC)

You are seriously confused about the topic. I have commented on the article's talk page.
The ChiS_R statement is sourced to Bevington, a reliable source published by McGraw Hill; the requisite assumptions are stated.
You do not understand measurement error. After doing careful measurements, "they" should not find a ChiS_R "very near zero" even if they perform the measurements in a vacuum.
Nobody should claim that choosing a worse fit to get ChiS_R equal to one is a good idea.
If ChiS_R is less than one, then the fit exceeds the measurement variance. A fit cannot be better than the measurements, so something is wrong. That is not superstition. Maybe the measurements are more accurate than believed or maybe the model has enough degrees of freedom to eliminate some of the measurement error.
Glrx (talk) 20:03, 28 June 2016 (UTC)

Yes, I think you are right. Since I don't know the technical definition of 'measurement error' I think that is what I was misunderstanding.Createangelos (talk) 22:03, 28 June 2016 (UTC)

About your revert of my edit to Quicksort[edit]

RE: my revert

First, I believe your indicated reason for the revert of the previous version "being good enough" does not meet the Wikipedia guidelines for a revert.

Secondly, what I was trying to point out is that in the usual analysis of sorting algorithms, the distinction between comparison sorts and the rest is not as much about them being capable of comparing arbitrary items with each other. That is a given as soon as you can sort anything at all, in any way at all. The salient point is that if you have to compare them *pairwise*, suddenly your sorting power is going to be limited in many models of sorting computation. As such, what you reverted in my addition, was in fact the whole *point* of why we point to the article handling comparison sorts, in the first place.

I hope you'd reconsider your revert, and after that undo it. Or perhaps work with me, or others, to make the point I was trying to raise even clearer. Decoy (talk) 03:02, 8 July 2016 (UTC)

Your insertion, "and contrariwise it only relies on pair-wise comparisons", is confusing. A comparison sort relies on comparisons. There is nothing "contrariwise" about it. The Quicksort article is not the place to draw generalities about other sorting models. Glrx (talk) 03:31, 8 July 2016 (UTC)

Thanks[edit]

RE spelling

The most obvious typos are often the hardest ones to spot. I don't mind at all, and I doubt Steve does either:-) (You know, I didn't name him that. The people down at Chili's Restaurant did. After his mother was killed by a car, they kept him alive through the winter by feeding him scraps.) Anyhow, thanks for the assistance Glrx. Zaereth (talk) 19:01, 15 July 2016 (UTC)

Undoing mistakes vs. wholesale revert[edit]

RE Gas tungsten arc welding

Next time you find an edit of which some was wrong, only undo the wrong parts. Not the entire edit. Okay? --bender235 (talk) 20:56, 18 July 2016 (UTC)

Why? So you can remain blissfully ignorant of your mistakes? You didn't take the hint with my first revert ("hyphenated pages not a range"); instead you repeated the mistake; I reverted again ("hypenated pages"), and you repeated your mistake while professing that you did not understand what was wrong ("may I ask what was wrong this time?"). Mindless. I don't want you to repeat the mistake, but I doubt you have learned anything from this episode. Your comment above suggests that you wish to continue to make errors and require other editors to fix the messes that you make. Why don't you learn not to make the mistakes in the first place? Actually look at the edits you're making and decide if they are absolutely correct before you commit. Although you may use scripts to make changes, it falls on you to make sure those changes are accurate. The other changes that you made to the page (such as http: to https:) are edits that a bot can do eventually; there's no pressure for me to do them right now; I'm happy to let the bot do it. Furthermore, I don't see why I should spend any time saving the good parts of your 2 minute automated edit campaign. Glrx (talk) 21:33, 18 July 2016 (UTC)
So you intentionally reverted my edits just to make a point rather than just fix mistakes. I see you understand the pillars of Wikipedia...
And just so you know: the edit that you complain about (hyphens replaced by en dashes) was done by AutoWikiBrowser automatically. --bender235 (talk) 22:39, 18 July 2016 (UTC)

Blowback[edit]

If you're interested, I've heard of this "blowback" phenomenon used in forensics from several different sources. The book Ballistics: Theory and Design of Guns and Ammunition talks about what happens as the bullet exits the barrel in great detail. As the bullet is accelerating out of the barrel it's followed by a pressure wave of expanding propellant, burning in an under-oxidized reaction. The expanding gas has mass of its own, therefore it resists acceleration and deceleration. At the speeds of a bullet this becomes a big factor. When the propellant comes into contact with the outside air, it finds more oxygen and flashes in a second burst. When this happens, the moving mass of propellant still in the barrel rushes out with the pressure wave and burns. This does create a momentary partial-vacuum in the barrel, creating a second surge of a tiny volume of inrushing air, which in turn burns much of the remaining propellant left in the gun. I hope that helps. Zaereth (talk) 23:41, 19 July 2016 (UTC)

Thanks for the ref; I'll track it down sometime.
There's a huge confusion about what the term means and the vigor of any "vacuum" in the barrel or behind the bullet. The wikipedia article suggests the feature is some sort of vacuum-driven "suckback" rather than "blowback", and I, being the naive fool that I am, just don't see that as a major contributor. Source definitions also vary.
The simple version of blowback is combustion products at the muzzle (and some other ports) are at high pressure and diffuse in all directions (including blowing back). A CRC Press book about GSR has pictures of weapons being discharged. The mean free path of the propellant gases is short.
A more complicated version of blowback involves gas injection into a close-by target and its subsequent rearward ejection. That can throw material all over the place -- including into a still pressurized gun barrel. One does not need a barrel vacuum as long as material has enough energy to move against the tide.
There will be an inrush around the muzzle cloud. At first (whether or not more combustion occurs outside the barrel) there will be a large volume of hot gas around the muzzle. That volume will expand until the pressure equalizes, but then it will cool and contract. A silly version is to imagine a balloon at the muzzle that captures all the combustion gases. The balloon inflates to a large volume: PV = nRT; after a time, P = atmospheric and T is still hot. As the gas cools, the balloon deflates -- but the balloon never deflates so much that it gets sucked into the barrel. At 25 liters/mole, there should be enough moles of propellant gas to fill the barrel.
There will be a barrel vacuum, but when it occurs and its strength is not clear. The combustion gases are hot. They've transfered heat to the barrel walls while propelling the bullet. After the bullet leaves, the pressure decays to atmospheric eventually (not instaneously; it takes about 8 milliseconds for a 48-inch 20mm barrel to decay to atmospheric. The gas in the barrel is hot, so it is expanded. As the gas cools, it will contract and allow outside air into the barrel. If the contraction is fast, it will be sucking in gas that was just radiated; that will be the back end of the combustion gas pressure wave (and would not be rich with other contaminants that are being blown out of the way). If the contraction is slow, it won't have the power to suck much in.
The displacement vacuum behind the bullet will be pulling material toward the bullet's path (and not toward the interior of the barrel). Near the muzzle, the gases are traveling faster than the bullet and have no trouble filling the displacement vacuum: the bullet is still getting pushed by the propellant.
Glrx (talk) 00:47, 20 July 2016 (UTC)
Shockwave.jpg
Yeah, I get what you mean. So many variables involved. I see a lot of these shows like Forensic Files and so on, and it's all just standard science (collecting facts and creating theories), except they often present the theories as facts. Since it's often the cops and not the scientists making the theories, I notice a lot of "target fixation" (a dangerous habit in air combat), where the detectives work to make the evidence fit the suspect rather than the other way around.
A friend on mine showed me that book, because he is really into ballistics. I was asking him about this photo, because (being curious about supersonic flows) I wanted to know why the mach-cone was not at the end of the bullet where it should be. Instead I found out it was at the end of a thin layer of incompressible flow surrounding both the bullet and the column of hot gas behind it, which in the photo is still expanding. Anyhow, I saw your user page so I just thought I'd share that with you. Zaereth (talk) 01:25, 20 July 2016 (UTC)
I'm not into ballistics, but my father is, so I've picked up some details. Your photo answers something that has bothered me: why doesn't the propellant momentum dominate the projectile momentum due to the propellant's faster velocity? Books say propellant momentum is only about 30% of the total momemtum. Your picture shows the propellant velocity is constrained because it must push on the atmospheric wall. I'm also wondering if adiabatic expansion will cool the barrel gases below the temperature of the inside of the barrel -- so the barrel vacuum cannot happen until the barrel wall cools down. If there were prompt barrel vacuum, there would not be a "smoking gun". Glrx (talk) 19:20, 20 July 2016 (UTC)