Wolfram Mathematica

1:13 AM

Stephen_Wolfram's Twitch video RSS: Exploring Version 12: Live with Stephen Wolfram Part 2 - LIVE

posted on April 24, 2019

1:54 AM

90% done with the website part of this tutorial/book plugin (currently working on the CSS for the theme) and then I just have to get the export settings for the PDF right and it’ll be possible to make super distributable tutorials with a minimum of extra effort.

Or books if you want to write Mathematica book with this

Michael E2

3:25 AM

The table of contents of this tutorial used to be expanded: reference.wolfram.com/language/tutorial/NIntegrateOverview.html Now it's not nearly so useful. The idiots don't know how to present information.

Szabolcs

6:41 AM

@MichaelE2 Maybe it was incorrectly closed by the website generation script (and no one has ever looked at it). Did you use the Give Feedback link?

user21

@MichaelE2 I very much sympathize with the part in your comment about the limited usefulness of the table of contents and requested that information will be restored.

Szabolcs

Does anyone here has experience generating usable documentation webpages with Workbench?

The default output (if I just check the box to generate HTML) is not directly usable. I guess it needs to be arranged in the correct directory tree so all resources can be found.

@b3m2a1 This reminds me that I still miss the old, v5-style doc browser.

Carl Lange

7:06 AM

@b3m2a1 As soon as I get the chance, I will, sorry about uninformative messages :)

Szabolcs

7:35 AM

0

Q: Caveats of LinearProgramming on Integers domain

When using LinearProgramming with the Integers domain, Mathematica issues the following warning: In[3]:= LinearProgramming[{1, 1}, {{5, 2}}, {3}, Automatic, Integers] During evaluation of In[3]:= LinearProgramming::lpip: Warning: integer linear programming will use a machine-precision approxima...

mathematical-optimization

My practical interest is solving computationally hard graph-theoretical problems by encoding them into ILP. Unfortunately, I have very little practical experience with solving ILP problems.

I realize that this is more of an ILP question than a Mathematica question ...

Regarding this incident with a semi-new poster: It seems that for some people it is very difficult to understand why their question is not good (even if it has nothing do to with SE rules and question would simply not be intelligible to anyone).

I might be useful to write a meta post which explains in detail how to ask a good question. Then we can simply link to that. The "site tour" is not good enough. A short essay that explains how to ask would save us time and mental effort, and would generally make the site more welcoming. It is very easy to lose patience when lots of such questions are posted.

KraZug

@JasonB., thanks. I'll talk to my colleague at some point next week, but if you are interested in feedback/discussion then I could probably set that up.

Szabolcs

We already have such a meta post for short, self-contained examples.

25

Q: What do we mean by a self-contained (minimal) working example?

Szabolcs linked to this page titled "Short, Self Contained, Correct (Compilable), Example" recently, and it really is well written. But not everything there applies to Stack Exchange, and even more so to the Mathematica SE. Can we write a set of guidelines more tailored to this site, and since t...

discussion questions

I welcome suggestions on how to improve the tone of the answer I have written there. I think the content is fine. What may be improved is the tone and how it reads in general. This would need to be done without making it longer though (making it shorter would be a benefit!)

Szabolcs

8:08 AM

@HenrikSchumacher Yes, LinearProgramming does use the interior point method for integers (setting any other method manually leads to an error). I am not familiar with this method at all though, and it's unclear to me how integer values would be enforced when using continuous optimization.

Michael E2

10:34 AM

@Szabolcs I didn't think about giving feedback -- forgot the link was there. Note that I was led to the website because doc. ctr. in Mathematica had the same limited TOC.

@user21 Thanks. I routinely used the expanded contents to navigate to whatever tutorial I was seeking.

At the NDSolve "Advanced..." TOC is still expanded. :)

Chris K

12:27 PM

@MichaelE2 Looks like it's now a "WOLFRAM MONOGRAPH" not a Tutorial.

At least a few other Monographs have the same format: reference.wolfram.com/language/tutorial/DSolveOverview.html

reference.wolfram.com/language/tutorial/…

Jason B.

2:45 PM

@Szabolcs For canonicalization, it depends on the atom-ranking function, which is laid out here: github.com/rdkit/rdkit/blob/master/Code/GraphMol/…

Szabolcs

3:03 PM

@JasonB. I'm a bit confused because canonical labelling of graphs is supposed to be pretty complicated, and the function you are linking to looks quite simple. Somehow it must be using the fact that molecules can't just be any kind of graph.

But then I do not know how canonical labelling algorithms like nauty actually work.

Michael E2

@ChrisK It was published as a book once upon a time, I believe.

Szabolcs

@JasonB. I should read this: drdobbs.com/graph-canonicalization/184405341

Jason B.

my c++ is not as good as it could be, so I have a bit of trouble reading functions like rankWithFunctor in the linked github page, but the basics there seem to be ring perception and then a basic atom comparator function

that link does look very interesting, I will read that

This is the basic substructure-matching code: github.com/rdkit/rdkit/blob/master/Code/GraphMol/Substruct/…

The author on that Dr. Dobbs link is one of the active developers for the library I linked to, though the article predates the library

Carl Lange

3:45 PM

@ChrisK Some of this document doesn't really make sense... 'When the second argument to DSolve is specified as Null instead of Null, the solution is returned as a pure function.'

Szabolcs

3:57 PM

@JasonB. I couldn't find the figures in the article (the links don't seem to work). If you managed to get them, let me know.

@JasonB. collaboration.cmc.ec.gc.ca/science/rpn/biblio/ddj/Website/…

Feeds

4:29 PM

Stephen_Wolfram's Twitch video RSS: Exploring Version 12 Live with Stephen Wolfram (Part 2)

posted on April 24, 2019

b3m2a1

4:50 PM

I wrote a Markdown to Notebook converter if anyone needs that

I tested it on some answers here:

Arnoud Buzing

Would it be possible to add "& Wolfram Language" to the header image?

A second question: Stephen Wolfram is interested in doing a live streaming event based on bugs (including these: mathematica.stackexchange.com/questions/tagged/bugs)

11

the topic would include feedback on the utility of an 'open bugs database' (hosted by Wolfram) and discussions about the types of bugs/suggestions that matter most to people (to help with development priorities)

He is curious is there is a) interest in this and b) if anyone has additional thoughts on this

b3m2a1

@ArnoudBuzing Open bugs database is an A+++++++++++++++ idea

Very much so interested on my end

Arnoud Buzing

Great!

For the first question (add Wolfram Language to the header image" who would currently be able to answer it? (I am not sure who runs the admin at the moment?)

b3m2a1

5:06 PM

@ArnoudBuzing probably need to get in touch with SE...?

Like we tell them we want to update the theme and how, etc. I think

But @halirutan would know best

Arnoud Buzing

ok, thanks

b3m2a1

It'd probably also be good to solicit input about the proposed change on meta

Arnoud Buzing

(to be clear: not talking about changing the site name/url, just mentioning ' Wolfram Language' in the header image)

b3m2a1

@ArnoudBuzing other input on an open bug database: chat.stackexchange.com/transcript/message/43479791#43479791

There is I think a lot of desire for that in the community and has been for a while

Arnoud Buzing

I think so too ... trying to make it happen

b3m2a1

5:10 PM

It'll also make managing the bug reports we get here much easier

Arnoud Buzing

The problem with this link: mathematica.stackexchange.com/questions/tagged/bugs

is that it is a pain to track which one of those 1,500 issues have been fixed

b3m2a1

And many things tagged [bugs] aren't bugs at all (or at least have not been confirmed as bugs)

If there were a proper bug tracker we wouldn't have to act as an ersatz one

Arnoud Buzing

oh, I did not know that ... I thought the tagging had been somewhat careful here

b3m2a1

It is as careful as we can make it

But we can't police every single thing tagged bugs since there just aren't enough people on at all times watching for stuff like that

The best we can do is go through after the fact and handle that

One the other hand if people are using the standard header that means that's a legit confirmed bug

But it's hard to keep up to date with which of those have been resolved

Chris K

5:25 PM

@CarlLange those look better in the documentation center (finally got my activation key sorted) - want to report that?

Szabolcs

5:50 PM

@ArnoudBuzing It would need to be a community decision. @Kuba @halirutan @MrWizard can say more, but I imagine that a prerequisite is that the community agrees. Thus a meta post would need to be made (IMO preferably not by you as you are WRI-affiliated), and the responses would need to demonstrate community support (or at least lack of resistance).

@ArnoudBuzing The idea is to keep it so, and I regularly edit posts to try to keep it tidy. But it's an uphill battle.

That's why we have the rule "Don't add the bugs tag when you post a question. Someone else has to add it."

@b3m2a1 @ArnoudBuzing Personally, I think that the state of the bugs tag is not bad. There are a few false ones, but it's more common that bugs are not tagged as such. The bigger problem is that it's so difficult to get people to use the standard header, or to adhere to the standard when adding a header. Without the header, the issues are not easily searchable.

See the message I pinned on the right soon after the v12.0 release.

There are also some disagreements about how the tag should be used.

I believe the original idea (and the version I support) was that the tag can be added by someone else than the original poster once it 1) either becomes clear that the community agrees or 2) WRI support confirmed it.

That's an "or" not an "and".

Some people also want to use it for documentation bugs, which IMO is pointless and dilutes the more useful use.

C. E.

@ArnoudBuzing You can use this report to filter out the ones that have been marked as fixed.

Feeds

6:06 PM

Stephen_Wolfram's Twitch video RSS: Live CEOing (238): Language Design in #WolfLang - LIVE

posted on April 24, 2019

C. E.

6:18 PM

FWIW, I don't think many of the posts tagged with the bugs tag aren't bugs. We're pretty good at moderating that tag. I randomly opened posts tagged with bugs and they were all confirmed bugs.

Arnoud Buzing

@C.E. thank you

C. E.

@ArnoudBuzing No problem at all.

Arnoud Buzing

7:04 PM

@Szabolcs ok, that sounds good. The argument for including it is that there are a number of product variants these days (like the Wolfram Cloud, Wolfram Desktop, etc.) which all have 'Wolfram Language' as the common denominator for the underpinning programming language.

C. E.

@ArnoudBuzing One argument against it is that we as a community don't have the expertise to answer questions about products other than Mathematica. Sure, there are a few people who use e.g. Wolfram Cloud but almost all top users here only know Mathematica. I still don't even know what Wolfram Desktop is.

Carl Lange

@C.E. Same thing as Mathematica with a different name and very slightly different functionality, based on license.

C. E.

7:23 PM

@CarlLange ok, then it's what I thought it was.

Carl Lange

In general I think it makes sense for the site to actually be Wolfram Language, not Mathematica, but the trouble is that WL has less mindshare than Mathematica does

Arnoud Buzing

@CarlLange Right, so it makes sense (to me) to at least include 'Wolfram Language' at the end for the sake of clarification

Carl Lange

@ArnoudBuzing I agree. I'm not in a place to make decisions like that though ;)

Szabolcs

8:06 PM

Could a chemist explain why this input is not accepted? Molecule[ConstantArray["C", 5], Subsets[Range[5], {2}]] I understand that such a molecule does not exist. But how can that be inferred in a simple way and how does Molecule infer it? It clearly checks for more than just that every "C" has degree 4.

Jason B.

8:22 PM

@Szabolcs - it's the ring-finding code that probably fails on this. I think I've seen this before, when just running Molecule[ <some random graph>] but haven't seen it come up on a more traditional molecule. The failure mode is bad here though

Szabolcs

@JasonB. You mean this above, or the email I sent with the input that creates ugly error messages?

I assumed that this really wasn't a proper molecule.

Jason B.

the one in the email was worse, but in that case I should have a better check for a self-bond.

Szabolcs

I was actually curious if the molecule canonicalization code would fail on these 4-regular graphs: GraphData[{"Antiprism", 6}] and GraphData["Qt19"]. This is based on a comment in the Brian Kelley article that the code presented there may fail on certain graphs.

Jason B.

for this one, there isn't anything obviously wrong with it, but I think it trips up on the "smallest set of smallest rings" search

I'll try this with all the debugger on (print statements) to see if I can pinpoint the issue

Szabolcs

Oh for this one it will work (4-vertex complete graphs).

Here all degrees are 3 (so we either include double bonds or add hydrogens)

My example was the 5-vertex complete graphs where all degrees are 4.

I assume such a thing can't really exist in reality, right? Perhaps due to the angles between bonds?

I was also messing with trying to construct fullerenes naively and I was amazed that MoleculePlot3D plots them nicely (but it takes very very long)

Here's a messy messy attempt from someone who forgot all his chemistry:

bucky = GraphData[{"Fullerene", {60, 1}}];
mol = Molecule[
  ConstantArray["C", VertexCount[bucky]],
  Bond[#, "Aromatic"] & /@ List @@@ EdgeList[bucky]
  ]

Now MoleculePlot3D[mol]

@JasonB. Is there a built-in way to try to look up if the things I create actually exist? I tried to google for the InChIKey

Jason B.

8:34 PM

for generating the 3D coordinates for that one, if I'm not mistaken it has to punt and ask Graph3D for initial coordinates (then minimizing using a force field) since the normal distance-geometry based method fails. The time is taken in trying the distance geometry first.

googling the inchi key is my go-to for finding it also

Szabolcs

8:51 PM

Check this out: community.wolfram.com/groups/-/m/t/1668066

It probably does not do pairwise comparisons for many values because there are lots of pairs. It might do some sort of normalization such as rounding, the comparing the result exactly

b3m2a1

11:24 PM

@Szabolcs @JasonB. why does the buckyball for that have such a distorted structure?

Transcript for