Talk:Mode-k flattening

"Mode-k flattening" listed at Redirects for discussion

The redirect Mode-k flattening has been listed at redirects for discussion to determine whether its use and function meets the redirect guidelines. Readers of this page are welcome to comment on this redirect at Wikipedia:Redirects for discussion/Log/2023 March 8 § Mode-k flattening until a consensus is reached. Onel5969 ^{TT me} 22:33, 8 March 2023 (UTC)

Requested move 19 March 2023

The following is a closed discussion of a requested move. Please do not modify it. Subsequent comments should be made in a new section on the talk page. Editors desiring to contest the closing decision should consider a move review after discussing it on the closer's talk page. No further edits should be made to this discussion.

The result of the move request was: Merge to Tensor reshaping. Rough consensus to merge to tensor reshaping; pinging [[User:jacobolus|]] and [[User:Quondum|]], who suggested and supported the merge, to implement it. (closed by non-admin page mover) BilledMammal (talk) 15:37, 28 March 2023 (UTC)

---

Mode-k flattening → ? – Opening a move request per closure of this discussion. Felix QW (talk) 16:22, 19 March 2023 (UTC)

@Onel5969, StarryGrandma, SunDawn, and Alexmov: Pinging participants of the RfD discussion. Felix QW (talk) 17:13, 19 March 2023 (UTC)

Vasilescu and Terzopoulos introduced the mode-m or mode-n flattening and matrixizing terminology in the early 2000s. Elden and Savas gives them credit which was the original reference on this page. The letter m is the first letter of the word mode and it has values $1\leq m\leq M$ where M is the upper bound. [Google search] of mode-n flattening shows that other people have adopted that terminology, while some have changed it to mode-k flattening.

In tensor algebra there are lots of variables. Employing the first letter of the word that you are trying to denote makes it easier to keep track of variales. For example, when refering to a specific mode, we use m. An index is denoted by i, $1\leq i\leq I$. The index of a mode $m$ is denoted by $i_{m}$ where $1\leq i_{m}\leq I_{m}$.

Alexmov (talk) 17:45, 19 March 2023 (UTC)

Probably a better title for the page is "Flattening" or "Mode Flattening" without the letter m or k designation.

Alexmov (talk) 18:46, 19 March 2023 (UTC)

•  COMMENT Flattening (edit | talk | history | protect | delete | links | watch | logs | views) is a different topic article. This article cannot move there per the suggestion that this article move to "flattening" provided above. -- 65.92.244.151 (talk) 21:34, 21 March 2023 (UTC)

  Why was the name changed from mode-m flattening to mode-k flattening? The name mode-k flattening is clearly incorrect given the references listed. Alexmov (talk) 19:02, 24 March 2023 (UTC)

• This article really seem to be too niche to have an article of its own in WP. Anyone who is involved with a conflict of interest should declare it. —Quondum 01:06, 22 March 2023 (UTC)
• Merge to tensor reshaping. This seems like one specific example of tensor reshaping that could be sufficiently covered by a section at that article, with redirects from both mode-k and mode-m flattening names, or whatever others seem necessary, and probably a link from the disambiguation page flatten. –jacobolus (t) 14:15, 22 March 2023 (UTC)

  I agree with this, perhaps specifically to Tensor reshaping § Matricization. —Quondum 15:00, 22 March 2023 (UTC)

Why was the name changed from mode-m flattening to mode-k flattening? The name mode-k flattening is clearly incorrect given the references listed. Alexmov (talk) 18:32, 24 March 2023 (UTC)

You seem to be far too frenetic about this. The original move to 'Mode-m flattening' was apparently contested on procedural grounds (see Wikipedia:Redirects for discussion/Log/2023 March 8). If you can't figure this out for yourself, maybe you should relax and not assume that everyone else just doesn't understand. Besides, sources that refer to "mode-k flattening" are easy to find. —Quondum 19:24, 24 March 2023 (UTC)

My last interaction with this page was five days ago on March 19. To suggest that I am being "far too frenetic" is absurd. Alexmov (talk) 19:32, 24 March 2023 (UTC)

Let me see if I understand this. The references listed and the literature indicate that most people use the terminology mode-n flattening. However, you want to delete the page and this is the process by which people that have contributed to the page feel like they've been "heard".

Is that it? Alexmov (talk) 20:16, 24 March 2023 (UTC)

Calm down. I suspect that few people will "hear" you if your tone continues. —Quondum 20:30, 24 March 2023 (UTC)

Best of luck with the decisions regarding this page. Alexmov (talk) 20:34, 24 March 2023 (UTC)

I think you (Quondum) are misinterpreting the intended tone. It doesn’t seem frenetic or aggressive to me (cf. WP:AGF). But Alexmov, sarcasm doesn’t always come through very well in plain text, and it is probably better to stick to straight-forward phrasing. –jacobolus (t) 00:44, 27 March 2023 (UTC)

Note that a merge with another page is not the same as deleting anything; all of the specific content claims can be preserved and the titles can be redirected to point to the same content. It just seems to me that most of the context/basic notions are the same between "mode k/m flattening" and "tensor reshaping", and the former seems like one specific example of the latter? Is my impression wrong? I don’t have any special expertise on this topic, beyond having done plenty of Matlab/Python array reshaping in the process of implementing various numerical algorithms. –jacobolus (t) 00:48, 27 March 2023 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

potential merge to tensor reshaping

I don’t feel like the above discussion really reached a consensus; it seems like it was closed prematurely. We only had a few participants, and I don’t feel particularly strongly about this – I was mostly throwing out one possible alternative.

It seemed like some editors aren’t satisfied with a merge to tensor reshaping. But I didn’t really understand what their concerns were. Perhaps we can discuss a bit more before implementing a merge. –jacobolus (t) 16:04, 28 March 2023 (UTC)

Were there any concerns mentioned about a merge? Alexmov may be unhappy with the current name, but that would become a terminology use decision within a section, unlikely to be contentious. Extending the time window may achieve nothing; this has been running a while. A merge seems a reasonable fresh point to move onto. Of course, others may disagree. —Quondum 16:40, 28 March 2023 (UTC)

I definitely agree with the merge idea, although I suspect the section title might still be contentious. For the record, I find the verb form matrixize as a section title less-than-ideal myself. Felix QW (talk) 07:56, 29 March 2023 (UTC)

I concur. The MoS, in Article titles (Normally use nouns or noun phrases), and by reference, Section headings, indicates that verbs are not the preferred style. A merge will at least move what are top-level issues from article level (where notability is a policy and naming is an issue, but undue emphasis is still significant) to content level (where a choice of say a variable name is essentially arbitrary, and hence less contentious). However, one periodically sees an editor who holds strong opinions, which may manifest as contention, especially if these do not align well with or adequately consider the community input. Maybe I should have said that I think that others are less likely to notice or spend as much time worrying about "local" article things that don't bubble up into other forums. —Quondum 12:34, 29 March 2023 (UTC)

Alexmov, do you wish to do the merge? If I do it, obviously I'll to use my judgment about what is just duplication and thus leave it out; I also evidently have very different views on how to choose terminology from you. —Quondum 20:34, 2 April 2023 (UTC)

I have no objection to merging the articles. Since the data may be vector-ized, matrix-ized or tensor-ized, I consider "mode-m flattening" or "mode-m matrixizing" as being correct terminology. Of course if you would like me to use a different mathematical definition, then it does not matter what you name it. Alexmov (talk) 08:13, 3 April 2023 (UTC)

@Quondum @jacobolus Why can't I see the title of the flattening article? Someone reverted it to mode-k flattening before there was a quorum. I would like to fix it. Alexmov (talk) 23:45, 3 April 2023 (UTC)

I don't follow your meaning. Merging the content of this article into Tensor reshaping will leave both Mode-k flattening and Mode-m flattening as redirects to a section on Tensor reshaping, so changing article titles would be counterproductive. —Quondum 01:11, 4 April 2023 (UTC)

Matrixizing?

Having read through the above discussions about the title of this article, both the "redirect for discussion" and the "requested move" and seeing the current state which is an edit war/conflict between the article title and article body, and having no idea why this is called mode k, mode m, mode n or mode fukall, I was wondering if this might be settled by moving the article to the title Matrixizing given that's an alternative name listed "also known as" in the lead sentence. wbm1058 (talk) 15:58, 7 April 2023 (UTC)

It seem to me that there is a failure of understanding of the process of establishing consensus. By my understanding, the current content should be merged to Tensor reshaping, which Alexmov seemed to be happy to do. Whoever moved the content recently to Mode-m flattening seems to be ignoring or failing to understand the process; it looks I may have to do the merge. IMO, this is no longer a matter of resolving an "edit war". —Quondum 16:13, 7 April 2023 (UTC)

Yes, fine but there may still be a question about what to merge.

• The mode-k matrixizing of a tensor ${\mathcal {A}}\in F^{I_{1}\times ...I_{K}},$ is defined as the matrix ${\bf {A}}_{[k]}$

  -or-

• The mode-m matrixizing of a tensor ${\mathcal {A}}\in F^{I_{1}\times ...I_{M}},$ is defined as the matrix ${\bf {A}}_{[m]}$

Is there any specific meaning to k or m or are these just arbitrary random letters representing variables? wbm1058 (talk) 17:59, 7 April 2023 (UTC)

Since I merged this document with the tensor reshaping article, I chose "mode-m flattening" and "matrizing" terminology. The terminology was coined by Vasilescu and Terzopoulos and given credit to by the original reference.

Multiway data analysis has a lot of variables. The letters $m,M$ where $1\leq m\leq M$ are the first letters of the word "mode". We say that we perform a mode-$m$ flattening, and that there are $M$ possible mode-$m$ flattenings.

I will redirect the article to the tensor reshaping article when I am done. Alexmov (talk) 19:48, 7 April 2023 (UTC)