## Not signed in

Want to take part in these discussions? Sign in if you have an account, or apply for one below

## Site Tag Cloud

Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.

• CommentRowNumber201.
• CommentAuthorUrs
• CommentTimeNov 2nd 2011

There’s some unpleasant issue going on: I perfectly see the Doriah version, but I see the version at Leinster 2001 badly corrupted, with many “Could not include SVG pullback” and the like. Presumibly there are commands used that are only defined and accessible from the Doriah page.

We may need to wait for Andrew to take care of this.

to me it has not been immediate to see that published papers were under “Arxive”. maybe a more explicit “nJournal issues” or “Published papers” or the like could be better.

Okay, I have changed it to “Published articles”

it is not clear where to get the referee’s reports for Tom’s article. I think we should link them from the 2011 page, together with the “submitted on/accepted on” dates

Okay, I have added more to 2011 (journal)

At HomePage (schreiber) I have made the link color much paler. Is it better now?

looks nice to me

Okay, thanks. Should we do something similar for the $n$Journal web?

1. Okay, thanks. Should we do something similar for the $n$Journal web?

maybe just using the darker blue we’re now using for visited links also for still unvisited ones could do the job there (I think it would anyway better not to distinguish visited and unvisited links on nJournal to avoid too many colours there)

the referee’s report page is still pointing to this version of Tom’s paper

• CommentRowNumber203.
• CommentAuthorUrs
• CommentTimeNov 2nd 2011

the referee’s report page is still pointing to this version of Tom’s paper

Thanks, I have fixed this (at An Informal Introduction to Topos Theory (nlabreviewed))

• CommentRowNumber204.
• CommentAuthorAndrew Stacey
• CommentTimeNov 2nd 2011

We should have a naming convention for articles. I feel that Leinster2011 is not quite precise enough. The volume and issue numbers should be in there as well to make it unique.

I’ll fix the missing stuff.

• CommentRowNumber205.
• CommentAuthorAndrew Stacey
• CommentTimeNov 2nd 2011

The “chunk” stuff was because the journal had the CamelCase wikilinks enabled, so stuff like FinSet was being considered as a wikilink. I’ve turned that off.

I’ve uploaded the arrows. For future reference (probably mine), here’s how I did it.

1. Selected (via the mouse, not the checkboxes) the list of SVG pages from the “All Pages” list at doriath; then xclip -o > arrows saves it.
2. To get the source of the arrows, I did something that should be equivalent to

cat arrows|perl -lne 's/^ *//; chomp; $_ or next; s/ /+/g; print "wget http://ncatlab.org/doriath/show/$_";' | zsh

3. Then to get the actual source text, I did (something equivalent to)

perl -i -lne 's/^ *//; /div/ && ($p = 0);$p and print; /div id="revision"/ && ($p = 1);' SVG*  4. Finally, to get them into the journal web pages, I did: for f in SVG+*; do xclip$f
firefox http://ncatlab.org/journal/show/\$f
done


This reads the files one-by-one into the clipboard, launches a firefox page where I can paste it in, and then waits for me to hit return before moving on to the next one.

• CommentRowNumber206.
• CommentAuthorAndrew Stacey
• CommentTimeNov 2nd 2011

Is the year of publication 2011 or 2010? The top of the page currently says:

An informal introduction to topos theory , Journal of the nLab vol. 1 no. 1 (2010)

• CommentRowNumber207.
• CommentAuthorAndrew Stacey
• CommentTimeNov 2nd 2011

Minor comment. I keep seeing things like:

An informal introduction to topos theory , Journal of the $n$Lab vol. 1 no. 1 (2010)

That space before the comma is completely unnecessary (and looks equally yucky). I think that this is remembered behaviour from some previous incarnation of the Markdown renderer where the space was needed for the parser to parse it correctly. I’ve just experimented on the Sandbox and it can go. Please write:

An informal introduction to topos theory, Journal of the $n$Lab vol. 1 no. 1 (2010)

• CommentRowNumber208.
• CommentAuthorUrs
• CommentTimeNov 2nd 2011
• (edited Nov 2nd 2011)

Thanks, Andrew, for all this.

Yes, let’s change the entry name to something better than “Leinster20xy”.

I think that this is remembered behaviour from some previous incarnation of the Markdown renderer where the space was needed for the parser to parse it correctly.

Yes, underscore followed by a comma did not use to parse. Glad to hear that now it does.

• CommentRowNumber209.
• CommentAuthorAndrew Stacey
• CommentTimeNov 2nd 2011

I don’t have any suggestions on good names, though! I think that the volume and issue should be in there, that alone makes it unique. And I would make it a simple name with no “funny” characters (including spaces) to make it as easy to use as possible. Something like v1n1_Leinster.

The other thing to say is: go to the page and take a look at it. Then click somewhere in the text (that isn’t a link). Now click in the border (the whitespace either side, or at the top). Do you like what you see?

And it is all done by CSS! No javascript involved.

• CommentRowNumber210.
• CommentAuthorUrs
• CommentTimeNov 2nd 2011

Then maybe we should just have

 v1n1


and nothing else. Because before long we will have submissions from authors with lots of weird characters in their name, or long lists of co-authors, or the like.

• CommentRowNumber211.
• CommentAuthorUrs
• CommentTimeNov 2nd 2011

Then click somewhere in the text (that isn’t a link). Now click in the border (the whitespace either side, or at the top).

Is this about the page Leinster2011 (journal)? On my system nothing happens when I click anywhere that is not a link.

 #Contents#


which is necessary right before the

* table of contents


and the

{:toc}


to make it appear (the text as such does not matter, only its existence).

• CommentRowNumber212.
• CommentAuthorUrs
• CommentTimeNov 2nd 2011

Ah, no. Now that the nLab has come back to reflect my edits, I see that the toc is still not displaying. How can we fix this?

• CommentRowNumber213.
• CommentAuthorTom Leinster
• CommentTimeNov 2nd 2011

It's a day full of classes and seminars, so quick comments for now and I'll catch up later.

• By any measure it's 2011, not 2010, as I submitted it on 7 Jan 2011. Funny I never noticed that before.

• Re Urs's green - yes, I find the new version much better (and rather pleasant, actually). I'm puzzled that you call it "paler", because it looks to me as if you've mixed the previous colour with black. My feeling is that in this context, links are best in a darker colour (i.e. one close to black).

• I see references above to "Journal of the nLab", but the name decided on (and submitted on the ISSN form) was "Publications of the nLab". (So the short, unofficial form would be "nPublications" or "nPubs", not "nJournal".)

More in a few hours, I hope.

• CommentRowNumber214.
• CommentAuthorUrs
• CommentTimeNov 2nd 2011

but the name decided on (and submitted on the ISSN form) was “Publications of the nLab”.

Oh. I got that mixed up. Will change it now.

• CommentRowNumber215.
• CommentAuthorUrs
• CommentTimeNov 2nd 2011

I have renamed the web. The article should now be at Leinster2011 (publications).

But currently the $n$Lab does not react…

• CommentRowNumber216.
• CommentAuthorMike Shulman
• CommentTimeNov 2nd 2011

The article page itself still says “Journal” at the top.

• CommentRowNumber217.
• CommentAuthorUrs
• CommentTimeNov 2nd 2011

Thanks, fixed.

• CommentRowNumber218.
• CommentAuthorAndrew Stacey
• CommentTimeNov 2nd 2011

Is this about the page Leinster2011 (journal)? On my system nothing happens when I click anywhere that is not a link.

Really? What browser (and version) are you using?

Please tell me that someone can see my fancy CSS! I spent ages on that.

• CommentRowNumber219.
• CommentAuthorAndrew Stacey
• CommentTimeNov 2nd 2011

Oh … bother. That appears to have been me. That’s my CSS hack for making links appear. I’ll have to reconsider how to do it.

• CommentRowNumber220.
• CommentAuthorFinnLawler
• CommentTimeNov 2nd 2011

Please tell me that someone can see my fancy CSS! I spent ages on that.

I can -- it looks great, but seems to make my somewhat elderly computer struggle a bit. Could it be done for each paragraph independently, maybe on mouse-over? (The trouble with having to click on the text to see the links is that you don't know what's a link and what isn't, so you can be suddenly whisked off to topos theory without warning if you click in the wrong place without watching the mouse pointer. Not a huge deal, but a little annoying.)

By the way, for anyone leaving links to the article, please remember that not everyone has write access to the Publications web -- when I click on the link above I'm asked for a password and have to go to the published version of the HomePage and navigate to the article from there.

Overall, though, the whole thing looks fantastic. I don't have the time or the expertise to make any significant contributions to the n-Publications, but it's a really exciting project to watch.

• CommentRowNumber221.
• CommentAuthorAndrew Stacey
• CommentTimeNov 2nd 2011

I can – it looks great, but seems to make my somewhat elderly computer struggle a bit.

It shouldn’t. CSS is much simpler than javascript. But I don’t really know anything about it …

Could it be done for each paragraph independently, maybe on mouse-over?

That’s actually a really, really good idea.

This is a “bug” with the nForum software. It doesn’t know which webs are write-protected so just sends everything to show instead of published. That could rapidly get annoying. I’ll see what I can do to fix it.

• CommentRowNumber222.
• CommentAuthorAndrew Stacey
• CommentTimeNov 2nd 2011

Done, and done. Links from the forum of the form [[publications:page name]] now go to the published version of the page (this “back dates”).

Hovering over a paragraph now means that the wikilinks will get revealed. There’s also a “click to reveal” box which ought to stay up in the top right corner of the page.

Some of this uses stuff in CSS3 which is supported in Firefox 7. Please let me know if it looks awful in a different browser. I have a suspicion that I’m trying to be too clever and that it might look just horrendous in old/broken browsers. If it does, I’ll have to think of a different way of achieving these ends. But if it looks okay in other browsers, then that’s great.

• CommentRowNumber223.
• CommentAuthorFinnLawler
• CommentTimeNov 2nd 2011

That seems to work perfectly, thanks! (I'm using Firefox 5 on Debian.) The mouse-over effect is smooth, so I suspect it was the sheer size of the page that caused my computer's wheezing fits when switching links on and off.

A couple of very minor points: on a few wiki-links the anchor text shows up in my browser only partly underlined, but the effect is rare and scattered, so it may well be just a browser bug. Also, there are a couple of unresolved wiki-links inside MathML, such as in Example 8.1, and then in the section Embeddings and Grothendieck toposes, in the first display after the picture of the covering family. Those were the only two I could find.

That's just nitpicking, though -- the article and the journal are looking better and better.

• CommentRowNumber224.
• CommentAuthorUrs
• CommentTimeNov 2nd 2011

Really?

Yes. I wish I could help it.

What browser (and version) are you using?

Firefox 7.0.1 on Windows 7.

• CommentRowNumber225.
• CommentAuthorMike Shulman
• CommentTimeNov 2nd 2011

Looks fine to me in both Firefox and Chrome, although the page is really slow to finish loading in both. I really like the displaying of links on hovering over a paragraph.

• CommentRowNumber226.
• CommentAuthorEric
• CommentTimeNov 3rd 2011
• (edited Nov 3rd 2011)

Where is the actual paper? It does not seem reachable from the home page.

Clicking the link that would seem to take me to the paper sends me here: http://ncatlab.org/journal/published/Leinster2011

But that page displays “Unknown web ’journal’ “

Edit: I found the nLab version here

Edit^2: I see. The home page I should have started from is http://ncatlab.org/publications/published/HomePage

Very nice work everyone. I haven’t read the article yet, but the formatting and CSS is very pleasant to look at. I hope more people begin publishing there.

• CommentRowNumber227.
• CommentAuthorTom Leinster
• CommentTimeNov 3rd 2011
• (edited Nov 3rd 2011)

This is looking really nice!

All I have to say is that I noticed a slight oddity. Go to the page http://ncatlab.org/publications/published/. This is entitled "Home Page". But just below the title "Home Page", there is a link called "Home Page" (in smaller type). Click it, and it takes you to a slightly different page, namely http://ncatlab.org/publications/published/HomePage. This second page is more recent.

The first URL, http://ncatlab.org/publications/published/, is the one on the ISSN form, so this should remain functional for a while at least.

• CommentRowNumber228.
• CommentAuthorDavidRoberts
• CommentTimeNov 3rd 2011

I’m in Firefox 3.6 at work and the mouseover to make links appear works for me, as does the click-box, but that is very slow.

• CommentRowNumber229.
• CommentAuthorTobyBartels
• CommentTimeNov 3rd 2011

Andrew, the CSS in the article is great! (Firefox 7 on Ubuntu.)

• CommentRowNumber230.
• CommentAuthorTim_Porter
• CommentTimeNov 3rd 2011
• (edited Nov 3rd 2011)

I am seeing nothing except: Unknown web ’journal’ ! This is on a MacBook with Firerox 3.6.23

Going in via : http://ncatlab.org/publications/published/Leinster2011

• CommentRowNumber231.
• CommentAuthorAndrew Stacey
• CommentTimeNov 3rd 2011

A couple of very minor points: on a few wiki-links the anchor text shows up in my browser only partly underlined, but the effect is rare and scattered, so it may well be just a browser bug.

If you look very closely, then you’ll see (I think) that when this occurs then there is a WikiLink below and the bit that is not underlined corresponds to the overlap. I suspect that this is due to the interaction of borders in the CSS model. I don’t know if it is fixable.

Eric and Tim, this is because Urs renamed the web from “journal” to “publications” so any link that used “journal” is dead. From here, you can tell since the web name is in parentheses after the page name. On the nLab (and other webs) you can tell by the pop-up. I’m not inclined to fix links here (though if there’s a real need I could) but links on the nLab should, of course, be corrected.

We still need a proper (short, concise, no funny characters) URL for the article. Urs suggested above v1n1 for “Volume 1, number 1”. Seems fine to me.

I didn’t know if wikilinks worked inside iTeX or not, seems that they don’t so I’ll take those out (I only see two as well). Thanks for spotting those, Finn.

The two different versions of the Home Page seems to have been a cache thing, probably to do with when I turned off CamelCase wikiwords. I’ve deleted the cached pages and it seems to work now (please check).

I’ve encountered two browsers where the CSS show-links thing doesn’t work: Mobile Safari and Opera Mini (on an iPad). The problem is due to the limitations of mouse events with those so the pseudo-classes don’t work in the same way as on an ordinary machine. On the other hand, since neither is a MathML-enabled browser, I have to use MathJaX to see the page and that takes an absolute age to render: on Mobile Safari then it was several minutes before the page had finished rendering, on Opera Mini then I gave up waiting.

On the speed issue, I don’t know if javascript would be any faster. With the hovering, maybe it’s not necessary to provide the ability to show the links in the whole document at one time (particularly since much of the page isn’t actually visible). Perhaps a better way to do this would be to offer different styles via the Stylish firefox extension, and similar for other browsers. That way, people can customise their view as they like and we can offer a few basic styles to get them started.

There isn’t yet, as far as I know, a decent MathML browser on the iPad. However, there may be eBook readers that can handle MathML so it might be possible to release an eBook version of the journal (indeed, of the whole nLab!). It’s something to ponder at any rate.

2. I’m on Firefox 3.6.23 on Mac OS 10.4.11: everything is perfect. more than perfect, awesome! great job, Andrew! thanks!

• CommentRowNumber233.
• CommentAuthorAndrew Stacey
• CommentTimeNov 3rd 2011

Okay, is there anyone using something other than Firefox?

• CommentRowNumber234.
• CommentAuthorTom Leinster
• CommentTimeNov 3rd 2011

Bilmey Andrew! That's beautiful!

Really: wow.

I know you want to know about people not using Firefox, but just for the record: Firefox 3.6.23 on CentOS Linux.

• CommentRowNumber235.
• CommentAuthorUrs
• CommentTimeNov 3rd 2011
• (edited Nov 3rd 2011)

I am now trying Leinster2011 (publications) with IE on my system.

Here the file starts with visibly displaying the line

<style type='text/css'>.newWikiWord { background-color: white; font-style: italic; }</style>


before displaying the blue page title. Then in the following all links are blue and underlined, and all graphics are broken. Clicking anywhere away from a link has no effect.

I am wondering what’s going on here.

• CommentRowNumber236.
• CommentAuthorMike Shulman
• CommentTimeNov 3rd 2011

Andrew 233: I mentioned Chrome.

3. Also works fine with TenFourFox 5.0 and with Safari 4.1.13, both on Mac OS 10.4.11

• CommentRowNumber238.
• CommentAuthorTobyBartels
• CommentTimeNov 4th 2011

Andrew wrote:

We still need a proper (short, concise, no funny characters) URL for the article. Urs suggested above v1n1 for “Volume 1, number 1”. Seems fine to me.

I like it. So the URL would be http://ncatlab.org/publications/published/v1n1.

I didn’t know if wikilinks worked inside iTeX or not, seems that they don’t so I’ll take those out (I only see two as well). Thanks for spotting those, Finn.

They don’t, and we sometimes work around this by putting things that should be in iTeX out of it:

• n-category’ (only I seem to bother with ‘$n$-category’, which is more trouble and still not perfect),
• ‘for $C =$ Set’ (one of Urs’s favourite tricks).

I’ve encountered two browsers where the CSS show-links thing doesn’t work: Mobile Safari and Opera Mini (on an iPad).

Neither Opera Mini nor the default browser work on my Android, but I didn’t bother to check this until now, because nothing works right on my Android.

By the way, Google informs me that Leinster 2011 won the championship, so that’s good!

• CommentRowNumber239.
• CommentAuthorEric
• CommentTimeNov 4th 2011

I agree. The CSS is really great. It makes a huge difference.

On an editorial note, one thing about the nLab that really bugs me is seeing some words partially linked, e.g. sets. I spent hours and hours going through the nLab correcting these, but gave up because they were being produced faster than I could clean them. It looks amateurish and very displeasing to my eye.

PLEASE make it an editorial policy for publications to have links fully underlined and NOT partially underlined. I do not see any partially underlined words in Tom’s article (which is a relief!), but just noting this for the record.

• CommentRowNumber240.
• CommentAuthorTobyBartels
• CommentTimeNov 4th 2011

If I understand the automatic system correctly, Tom’s article should have ‘sets’ instead of ‘sets’ because the former exists (as a redirect). If the redirect does not exist, then the latter form would be used.

• CommentRowNumber241.
• CommentAuthorTom Leinster
• CommentTimeNov 4th 2011

I don't disagree with Eric's objection in 239. However, I'll note that when viewing an nPubs article with the links invisible by default (i.e. only visible when you move the mouse appropriately), it would be much less of an issue anyway.

• CommentRowNumber242.
• CommentAuthorUrs
• CommentTimeNov 4th 2011

(only I seem to bother with ‘n-category’,

One reason that I am not bothering with it is that to me it looks less good than n-category. It’s true that it has an italicized “$n$”, but the trouble is that it looks like it will take you to category when clicking on it.

• CommentRowNumber243.
• CommentAuthorTobyBartels
• CommentTimeNov 4th 2011

Right, I disagree as to which looks better, but clearly neither is what we really want.

• CommentRowNumber244.
• CommentAuthorzskoda
• CommentTimeNov 4th 2011

Unlike Eric, I prefer sets to sets. First, the latter form, that is an alias will not backlink (when used within $n$Lab). Second I prefer to know the original form of the title and not the alias (I use the wanted pages mainly by typing their name, and if I use source code or print version it will not react to alias, if I use the main view then it will start with ugly warning that I was redirected).

• CommentRowNumber245.
• CommentAuthorMike Shulman
• CommentTimeNov 5th 2011

Well, [[set|sets]] will backlink and also look non-ugly. But the non-backlinking of redirects is really a bug that ought to be fixed, not something we should expect to remain the same in the future.

• CommentRowNumber246.
• CommentAuthorUrs
• CommentTimeNov 10th 2011

What’s the status? Is the publication of the article done and finalized?

If so and when so, I will post an announcement to the $n$Café.

• CommentRowNumber247.
• CommentAuthorAndrew Stacey
• CommentTimeNov 10th 2011

I’ve been meaning to check the behaviour with IE. I’m looking at it now.

• CommentRowNumber248.
• CommentAuthorUrs
• CommentTimeNov 10th 2011

Okay, thanks. I am not meaning to rush you or anyone. I am just checking.

• CommentRowNumber249.
• CommentAuthorAndrew Stacey
• CommentTimeNov 10th 2011

Okay, it looks as though the per-web extra styles don’t get applied for IE, at least for IE8 which is the version I have access to. Still, the rest of the page looked simply awful so I don’t think anyone would get far enough to notice that the links are a bit too full-on. Does it look reasonable to you in IE apart from the hyperlinks?

Nonetheless, that line at the top (style and so forth) shouldn’t be there. It looks like a bug in Instiki (a rare one) so I’ve reported it to Jacques. It also looks like a really simple fix so it shouldn’t take long.

I’m certainly happy with it now. I can think of a few style tweaks, but the link-system isn’t set in stone and might be better to mess with once more people have seen it.

So what’s our system for finalising? I guess we also need Tom to give his final stamp of approval (though maybe he’s already done so). Is there anyone else who should look at it?

• CommentRowNumber250.
• CommentAuthorUrs
• CommentTimeNov 10th 2011
• (edited Nov 10th 2011)

Does it look reasonable to you in IE apart from the hyperlinks?

In my IE the hyperlinks look nice (after reloading the page, for some reason they were blue and underlined before I hit “reload”) but all the graphics are invisible and all the displayed formulas are broken. So the article is effectively not readable in IE for me.

In the top right corner I see a box saying “Click here to show links to nLab pages”, but clicking there shows no effect (and on my system this is the same on Firefox).

• CommentRowNumber251.
• CommentAuthorAndrew Stacey
• CommentTimeNov 10th 2011

What version of Firefox? The “Click here to show links” bit uses some CSS3 stuff and I’m not sure how far back in Firefox’s history that goes. Maybe I should tweak it so that that only shows if the browser can cope with it.

but all the graphics are invisible and all the displayed formulas are broken.

Is that different to when you see the version of the article on the main nLab?

• CommentRowNumber252.
• CommentAuthorUrs
• CommentTimeNov 10th 2011

What version of Firefox?

When you asked me last week it was still the latest, Firefox 7.0.1 (on Windows 7).

Is that different to when you see the version of the article on the main nLab?

It’s the same there. I never noticed it, since I never use IE.

• CommentRowNumber253.
• CommentAuthorAndrew Stacey
• CommentTimeNov 10th 2011

When you asked me last week it was still the latest, Firefox 7.0.1 (on Windows 7).

You expect me to remember stuff from last week?

That’s then doubly odd since I’m using Firefox 7.0.1 here on Linux and it works just fine. I’ll experiment with Firefox on Windows as well. Personally, I’m not bothered about IE since just about everything’s broken with that, but we ought to support the latest (or almost the latest since FF8 was released yesterday) Firefox on all platforms.

• CommentRowNumber254.
• CommentAuthorTom Leinster
• CommentTimeNov 10th 2011

Re 249, yes, I give my stamp of approval. Thanks for asking.

Of course, I'll be happiest if it's viewable on lots of different browser setups, but you guys seem to be taking care of that. (I have no easy way to view it on IE.)

• CommentRowNumber255.
• CommentAuthorUrs
• CommentTimeNov 10th 2011

Andrew,

maybe there is something wrong with the installation of my system (I wish I knew what). If nobody else has this problem, don’t worry.

• CommentRowNumber256.
• CommentAuthorTobyBartels
• CommentTimeJan 1st 2012
• (edited Jan 1st 2012)

I removed the (two) links to cohesive from Leinster2010, since these don’t really go anywhere; the article is a pure disambiguation page in category:adjective. (In future, I propose not automatically creating links to articles in this category.) This links might profitably go to a page cohesion (now asked for from the adjective page), but that doesn’t yet exist, so I didn’t put them in.

• CommentRowNumber257.
• CommentAuthorUrs
• CommentTimeJan 1st 2012

This links might profitably go to a page cohesion (now asked for from the adjective page), but that doesn’t yet exist, so I didn’t put them in.

But from the context in Tom’s text the link clearly wants to go to cohesive topos. Why not just make it point there?

• CommentRowNumber258.
• CommentAuthorTobyBartels
• CommentTimeJan 1st 2012

I take it that a cohesive topos is a particular formalisation of the idea of a topos of sets with cohesion, but if you think that it’s general enough, then I certainly have no objections to putting the link there.

• CommentRowNumber259.
• CommentAuthorTom Leinster
• CommentTimeJan 2nd 2012

More importantly, I inserted a sentence after the title saying that this was the changeable version, not the frozen, published version. I probably haven’t phrased it in the best way, but it’ll do for now. Perhaps it should be routine to add a sentence such as this on the editable version of published articles. (The different colours do help — I like that system — but the more clarity the better.)

• CommentRowNumber260.
• CommentAuthorUrs
• CommentTimeJan 2nd 2012
• (edited Jan 4th 2012)

I take it that a cohesive topos is a particular formalisation of the idea of a topos of sets with cohesion, but if you think that it’s general enough

Maybe there are other axioms that you might argue would formalize a notion of “sets with cohesion” (which ones?), but Tom’s text at this point is talking about Lawvere’s such notion, which is the one described at “cohesive topos” (including some variants of the axioms that Lawvere himself considered). So it seems suitable to me.

By the way, in Tom’s text a sentence following the mentioning of cohesion reads

For example, there are toposes of smooth spaces, which are the setting for synthetic differential geometry.

I am wondering if we could just slightly expand on that or modify it to make it more correct. Because (currently at least) the entry smooth space defines the sheaf topos over CartSp, which is cohesive, but is not a model for synthetic differential geometry (because it has no infinitesimal objects).

A cohesive topos that is a model for synthetic differential geometry is the Cahiers topos, being sheaves on $CartSp_{infinitesimally thickened}$.

Here (and in section 2.4) I have proposed additional axioms on a cohesive topos that are supposed to ensure that it comes with a notion of infinitesimals, hence that it behaves like a model of some kind of synthetic differential geometry. The only point that I am aware of where Lawvere tries to formalize this relation is the brief paragraph named “section IV” in his Axiomatic Cohesion.

• CommentRowNumber261.
• CommentAuthorzskoda
• CommentTimeJan 3rd 2012
• (edited Jan 3rd 2012)

What is the status of the convention-proposal (which I like) in 210 ? (I apologize for interrupting very interesting ad hoc discussion on cohesiveness which should continue).

• CommentRowNumber262.
• CommentAuthorTom Leinster
• CommentTimeJan 4th 2012

@Zoran (261): I wondered about that too, and I also like the proposed convention.

• CommentRowNumber263.
• CommentAuthorUrs
• CommentTimeJan 4th 2012
• (edited Jan 4th 2012)

concerning #260:

to account for that I have expanded the entry smooth space a little. Added a section Variants and generalizations and a brief section Properties. (Both of which could easily be expanded further…)