Talk:Octonion

Anticommutativity
The lead says that they are noncommutative and nonassociative. They are actually anticommutative and antiassociative, right? Bubba73 You talkin' to me? 03:45, 16 November 2012 (UTC)
 * No: ℝ and ℂ are both subspaces which are commutative, while any two ei and ej for i ≠ j, i, j > 1 anticommute. So like quaternions they are neither commutative nor anticommutative. I don't even know what antiassociative means; there's not a simple negative form of associativity, as there is of commutativity.-- JohnBlackburne wordsdeeds 04:25, 16 November 2012 (UTC)


 * From the top of the properties section:
 * $$e_ie_j = -e_je_i \neq e_je_i\,$$
 * Isn't that what anticommutative means? Bubba73 You talkin' to me? 04:36, 16 November 2012 (UTC)
 * It is not true if i or j =0 ($$e_0e_i=e_ie_0$$) It corrected the affirmation in the article.--Cbigorgne (talk) 07:20, 16 November 2012 (UTC)

Thank you, that helps clear it up. Bubba73 You talkin' to me? 15:51, 16 November 2012 (UTC) Hello. I do not think this is resolved. Actually, it is true that $$e_ie_j=-e_je_i$$ if $$i,j\neq 0$$ and $$i\neq j$$. For the non-associative part, the conditions are even harder to write...2A01:E35:8A15:73A0:6134:310F:F279:7C8E (talk) 22:20, 21 March 2013 (UTC)

I assume "antiassociative" would mean (ab)c = −a(bc). This isn't always true for the octonions: it must be so since the octonions contain copies of the quaternions within them. But it will be true if {a,b} generate an algebra isomorphic to the quaternions that is orthogonal to c (essentially what the article says). Double sharp (talk) 17:00, 12 March 2016 (UTC)

Multiplication table
What do you guys think about slightly increasing the width of each column in the table? What about replacing $e _{5}$ with e 5 for the duration of the table?

Both of these changes would only be for aesthetic purposes. It would match the corresponding table for sedenions. BradBentz44 (talk) 06:33, 28 May 2014 (UTC)


 * I'd suggest replacing the var tags with italics quotes ( →  ), since the semantics are not of variables.  Controlling column width (and making them a little wider) in the table would be good. Ordinarily, I would not agree with removing the   template, but here the subscripts end up too large on my browser.  An alternative is removing the subscript from the font modification: $e$5 – does this achieve what you're after aesthetically? —Quondum 13:37, 28 May 2014 (UTC)


 * I do think that $e$5 looks better than $e _{5}$. There isn't much difference between $e$5 and e 5. Do you think we should go ahead with the change? BradBentz44 (talk) 20:30, 28 May 2014 (UTC)
 * I don't think it's needed. The formatting matches that of the inline math above and below; changing it would make it far less consistent and so visually jarring. And a subjective reason such as 'for aesthetics purposes' is not a good reason to change formatting from one style to another; an article should not be changed just for such reasons. See e.g. MOS:STABILITY.-- JohnBlackburne wordsdeeds 20:49, 28 May 2014 (UTC)
 * I see the redirect at the link MOS:STABILITY that you provided does not go anywhere but WP:MOS – any idea which section it should go to?
 * The formatting is not consistent: the inline math does not in general use math, whereas the table does. This is sufficient reason to change the article to a consistent format.  Perhaps your browser setting result in very similar rendering, but with mine the subscript size gets really messed up with one.  But I agree that this is not a reason for any choice.  So really, should we go with math throughout except for the standalone &lt;math&gt;? —Quondum
 * There's no heading but MOS:STABILITY goes to the third para of the lead: "Style and formatting..." with links to relevant arbitration cases. MOS:MATH has a similar guideline at MOS:MATH which refers back to the main manual of style.-- JohnBlackburne wordsdeeds 11:49, 29 May 2014 (UTC)
 * Thanks, I missed the anchor. Anyhow, I'm not trying to dispute the principle that style should not be changed without reason, I only thought it was a link that needed fixing. That still leaves us with my observation that this article mixes HTML with and without : the table uses it, but the only place outside the table where it is used is in the sentence immediately after the table. We should change the article to use and   uniformly, or remove the  usage. —Quondum 13:25, 29 May 2014 (UTC)

Invention of "associative"
I removed a remark: "Hamilton invented the word associative so that he could say that octonions were not associative."

In this was credited to Cayley, then  and moved before the citation, which does not involve the term "associative" at all. This makes me suspect it wasn't meant very seriously.

Graves wrote to Hamilton about his "octaves" on December 26, 1843 (according to Hamilton's account: but Hamilton was already using the term "associative character" to talk about quaternions in his paper presented November 13, 1843:

Hopefully this is sufficient explanation to remove the remark rather than sticking a citation request on it. Agashlin (talk) 12:16, 2 May 2017 (UTC)


 * Just wanted to add that I regret having written an uncharitable summary for that edit, I don't think there was malicious intent and I should have assumed good faith anyway. Agashlin (talk) 12:44, 2 May 2017 (UTC)
 * Since this ia a Talk page, references stay in sections. — Rgdboer (talk) 21:16, 11 October 2017 (UTC)

Matrix representation
first it was stated: Just as quaternions can be defined as pairs of complex numbers, the octonions can be defined as pairs of quaternions

and next time you wrote: Because of their non-associativity, octonions do not have matrix representations, unlike quaternions.

but if a quaternion can be represented by a matrix, and octonions can be defined as 4 quaternions then it could be represented uniquely by 4x4x4 (3D) matrix.. No? — Preceding unsigned comment added by Xakepp35 (talk • contribs)
 * You should look at the Cayley–Dickson construction. It is how each of the algebras is derived from the previous one, so how octonions is derived from quaternions. And because of how it works each time the new algebra loses certain properties. As you move from quaternions to octonions you lose associativity, so there is no matrix representation (as matrices are associative).-- JohnBlackburne wordsdeeds 06:28, 11 October 2017 (UTC)


 * I guess the question is "what do you mean by matrix representation?" Obviously it is not a subalgebra of a matrix ring because any such algebra is associative. But the elements of any finite dimensional algebra can be viewed as a linear transformation and therefore a square matrix.  This is, at least, the spirit of Jacobson's structural description of general finite dimensional algebras.  That being the case, I'm rewording to be more explicit about the assumption that was previously made: that a 'matrix representation' meant "subalglebra of a matrix ring". Rschwieb (talk) 15:01, 8 January 2021 (UTC)

Contemporary Octonion Research
Snark aside, I fail to see why this isn't serious contemporary research worthy of note on Wikipedia. Flesh it out for me.

Than again: https://www.math.columbia.edu/~woit/wordpress/?p=3665 Still not even wrong or no? kencf0618 (talk) 19:41, 28 July 2018 (UTC)


 * Sorry, my mastery of the English language may not be sufficient to reign any snark, apologies in advance.
 * Well, I am not scared away by pentagrams, not even with candles in their corners, so why should I shy away from a minimal projective plane. Associating selected objects from the physicists' particle zoo with distinguished elements of an algebra is free, at least for a while. Projections of predictive power of these associations are nevertheless missing. Promising is all there is that I am aware of, for the time being. I am, however, really scared away by the alleged multiple tensor(!) product of groups, or is it rather vector spaces, or division algebrae, being myself really unsure about a valid outcome.
 * The redactrice of the Quanta Magazine evidently knows a lot about puffery and qualification of her readership, too (see the comments section there!), but since I am neither female, nor playing an accordeon (not even a bandoneon while doing a tango), and not at all a martial artist, I won't comment on her authority in estimating the scientific weight of her starring protegee, nor on that of the latter's article. The article in Quanta Magazine has no scientific weight to my measures.
 * The linked remarks from 2011 by J. Baez at al., meanwhile ripened during a period of seven years and however prophetic, may also be applicable, but is there any reliable source (not talking about secondary), that is beyond speculative expectations, since then?
 * Please, feel free to edit according to your estimation what is best for the encyclopedicity of this article. Purgy (talk) 09:32, 29 July 2018 (UTC)


 * We'll just have to keep an eye on it. kencf0618 (talk) 00:18, 31 July 2018 (UTC)

A problematic reference to the multiplication table and mnemonics used based on $$e_1e_2 =e_4$$ mod-7 construction
It seems beginning on July 10, 2019 User:MatthewDougherty correctly added a tag that some text related to a 7-cycle construction based on e1e2 = e4 was confusing and the reference did not match the multiplication table above, i.e. $$e_1e_2 = e_3$$. On Apr 8, 2020 User:Pianostar9 deleted some of that text, but not the tag.

Then on May 3 User:CFjohnny1955, with some clarification by User:XOR'easter, deleted the tag by adding text that inferred the multiplication table and the mnemonic were different octonions: "...by using the triangular multiplication diagram, or Fano plane, below (and instead of the multiplication table, above)."

To be clear, the table and both 2D and 3D Fano mnemonics are (and always have been) consistently based on the triads 123-145-176-246-257-347-365. The reference to the 124-137-156-235-267-346-457 octonion Fano plane is not present in this article (yet). FYI - my triad lists are generated algorithmically and sorted in ascending order.

So this incorrectly implies there is a difference in the octonion triads used in the table and the mnemonics. While I understand the motivation to clarify and explain how to perform the interesting 7-cycle math based on the triangular rotation of the Fano plane $$e_1e_2 = e_4$$, the current comment either needs to be removed, changed, or a new Fano plane mnemonic needs to be added for that purpose. I guess we could replace the table and the mnemonics to all be based on the 124 triads, but that would be disruptive to the historical use of, and references to, the 123 triad representation in the article. Also, that representation putting the IJK quaternions in the upper left quadrant is also very important for understanding Cayley-Dickson constructions, so changing it would not be a good idea.

Since I was the one who generated these mnemonics from my Mathematica codebase using any of the 480 possible triad combinations, I created a reference to the different sets (yes, I can create just the 124 Fano plane for reference w/o the triads and tables, if you wish):

124

123

What is the best approach to correcting the issue?

Jgmoxness (talk) 21:38, 20 July 2021 (UTC)
 * I believe my edit had simply moved and reformatted text. Pianostar9 (talk) 14:17, 23 July 2021 (UTC)

I think the current diagram and the right table are incorrect. The yellow picture in Octonion-124-137-156-235-267-346-457.svg suggests e3 = ij, while the table and text suggest the convention e4 = ij. The yellow picture also doesn't match the convention that the product of two points on a line is the third point (up to sign). For example, look at the line 137. One should have e1*e3 = e7 (which in the picture is lk), but instead, e1*e3 = ik. This seems to be incorrect in both picture and right table. I think the fix should be to remove the ij in the middle point of the picture, leaving it as e3 = k, changing the bottom right point to e7 = ik (instead of e7 = lk), changing the top point to e5 = jk (instead of li), and changing the bottom left point to e6 = kl (instead of lj). The right table should be adjusted accordingly. Sadly, this is not an edit I can make; User:Jgmoxness is the one who generated this picture. 2A02:A454:920E:1:8141:EF58:A3B3:4002 (talk) 13:10, 6 October 2021 (UTC)
 * I now see you moved the text (vs. deleting it), but this was not the crux of my point - which was just trying to outline the complex history of the changes related to the confusing set of points being made in the article.
 * You are correct about a IJKL notation issue between the Fano diagrams' association of e# format to the header rows in the IJKL multiplication tables and your suggested fix for e3=k is the right approach. I should also remove the =lij on e7=lij=lk. These two changes remove the underlying problem with trying to link IJKL headers to the e# on all 480 diagrams.
 * The problem with the IJKL notation is that in trying to have the headers be consistent with the table math sequences for all 480 representations is very confusing and no longer value added. I tried to make it simple with a consistent header structure linking to e#, but included the mistake of adding the math elements in the e3 and e7 Fano plane nodes. My apologies.
 * So I have corrected those WP Fano diagrams.
 * IJKL is all just a holdover from extending imaginary to quaternions with only 2 possible representations where ij=k (up to sign). In extending it to octonions, the IJKL convention was largely dropped for readability and the practicality since there are now 480 possible representations (many where ij≠k in all but the 123 based tables).
 * Having said this, maybe dropping the all of the IJKL references in the Fano diagrams and deleting the IJKL tables would be better.
 * Thoughts?
 * Jgmoxness (talk)

Huh???
English next time please??? Carlimited (talk) 03:29, 16 December 2022 (UTC)


 * Could you be specific, please? It would be easier to address your issue if you made it clear.—Anita5192 (talk) 06:34, 16 December 2022 (UTC)

UI problem
One of the graphs stick out past the Tools bar. I don't know how to fix this, could someone else do it? UnqaidIntern (talk) 17:13, 16 May 2023 (UTC)


 * They look fine to me. Which graph are you referring to?—Anita5192 (talk) 17:21, 16 May 2023 (UTC)
 * It's the IJKL graph, I'm on a Chromebook so that might be the reason. UnqaidIntern (talk) 17:23, 16 May 2023 (UTC)


 * I see what you mean. The diagram should be replaced by separate, smaller diagrams, but I don't have the facilities for doing that. Hopefully someone else will.—Anita5192 (talk) 18:10, 16 May 2023 (UTC)
 * Much easier to simply reduce the image size - fixed.Jgmoxness (talk) 15:44, 7 August 2023 (UTC)

Consistent representation of the Fano plane please
The Fano plane in this image and don't match. For instance, e3 is the center of one, but the other has e4. Also the inner circle is counterclockwise in one while the other is clockwise.

It's likely these are identical under relabeling, but it would be a lot easier if a consistent definition was used. I'm just learning this for the first time and got a little confused! Pstetz (talk) 20:08, 4 August 2023 (UTC)


 * These are not the same diagram because they are not the same octonion (which is addressed in the article text). There are 480 possible and different octonion multiplication matrices. These are typically identified through a set of triads 123-145-176-246-257-347-365 or 124-137-156-235-267-346-457.
 * Consistency in diagrams means forcing a selection of a single unique triad set. This would provide clarity in the visual representation, but would limit the understanding of what octonions really are. These two diagrams are likely the most commonly used among about 1/2 dozen across the literature and their differences are interesting (IMO). For the complete set of 480 diagrams and matrices (which includes split octonions by triad), see splitfano.pdf
 * Jgmoxness (talk) 15:35, 7 August 2023 (UTC)

Alternativity implies power associativity
At the top of the page, it says of the octonions "they are alternative. They are also power associative." This is technically correct, but the language suggests that alternativity and power associativity are independent. They aren't - power associativity is a weaker form of (and thus prerequisite to) alternativity. This should probably be rephrased to something like "they are alternative, and thus power associative", but I can't find the best way to phrase that sentence. Cyversch (talk) 19:04, 13 May 2024 (UTC)