Wolfram Mathematica

Chris K

00:42

@b3m2a1 I'd be curious to hear about it!

@VitaliyKaurov Huh, pretty weird seeing something from our lab's experimental focus here

Jason B.

02:06

@Szabolcs - I'm aware there are a lot of graph bugs, bugs that tend to not get fixed, but I'm not aware of the new-in-11.2 breakages you mentioned in your post. Is there a post here referencing them?

b3m2a1

02:18

@ChrisK I'll try to find an hour or two to write a post sometime in the next few days or this weekend, then.

Sascha

07:41

Sanity check: Subgraph should not yield the complete graph g, shouldn't it?
g = CompleteGraph[10]
Subgraph[g, 1 <-> _]

halirutan

08:50

@Kuba Would it be OK if I add my blog-feed to the feeds of this room? Most of it is connected to Mathematica and I only published 3 or 4 articles. So it won't be too much annoyance.

It seems to have the nice side-effect that users correct me, if I haven't included something (@MichaelE2).

Szabolcs

09:13

@JasonB. I don't think so. I reported it and it was acknowledged. IsomorphicGraphQ simply does not evaluate in M11.2 on Windows. Several other related functions also fail.

@JasonB. Read the chat transcript from here: chat.stackexchange.com/transcript/message/41440702#41440702

Incidentally, these also don't work properly on the Raspberry Pi. My uneducated guess is that some of the isomorphism code is architecture-sensitive and something went wrong when they switched to a newer compiler on Windows for M11.2.

Szabolcs

09:46

@JasonB. Unfortunately, IGraph/M is not a full workaround either. While some of the isomorphism algorithms it includes are algorithmically superior to Mathematica's (and much faster for hard problems), converting the graph to a compatible format is slow, and has a large overhead. This means that on small graphs, it is much slower than Mathematica.

Where this is particularly apparent is eliminating isomorphic duplicates by pairwise checking (but then you shouldn't do pairwise checking anyway—try to always prefer DeleteDuplicatesBy with a canonical form).

What is shocking to me is no one here has yet complained about the isomorphism functionality being completely broken on the most popular OS, in the latest version.

What does that mean? That almost no one uses graph stuff in Mathematica? That wouldn't be surprising given that the functionality area is basically abandoned by WRI. Critical bugs are no longer fixed, unless they're trivial.

Sadly it's a chicken and egg problem. I tend to think that WRI doesn't want to develop it because no one uses it, and no one uses it because it's not being developed and it's full of bugs.

Kuba

11:41

@halirutan I think it will be welcomed :)

Szabolcs

@Sascha What is the output you get? I need a sanity check for myself too. I could swear the first few times I tried this, it just spit out g as it was, i.e. a complete graph. But now it gives me an empty graph, even after kernel restart!

However, if I use \[UndirectedEdge] then it does give me the complete graph again.

I tried using -> as well (since it's commonly used for edge notation), and I discovered that Subgraph will use it as a replacement rule. Not clear if this is a bug (e.g. passing the 2nd argument to Cases without checking) or intentional.

halirutan

11:56

@Kuba Thanks. When I add the feed, I guess all existing posts will arrive at the same time once.

Kuba

@halirutan I don't know but this should not be a major problem

Feeds

halirutan has made a change to the feeds posted into this room

Feeds

12:11

halirutan: Simple scheme for a line-pattern

posted on September 25, 2016 by Patrick Scheibe

This shows an extremely easy line-pattern that can be worked out with paper and pencil, as well as implemented using *Mathematica*.

halirutan: How Key Promoter became Key Promoter X

posted on August 10, 2017 by Patrick Scheibe

This is a story about a small, but very popular plugin for IntelliJ IDEA

halirutan: The Definite ANTLR 4 Referece: A rant about today’s book quality

posted on November 06, 2017 by Patrick Scheibe

This is a short analysis of the poor printing quality of a book from "The Pragmatic Bookshelf".

halirutan: Mathematica: Checking your function for wrong options

posted on December 30, 2017 by Patrick Scheibe

When you define a function in Mathematica, it usually takes also some options. If you provide a wrong option to a built-in function, it has a unique way giving an error message that contains the wrong option and the original call, returning your code unevaluated. This short tutorial shows how to mimic this behavior for your own functions.

Szabolcs

13:21

@Kuba @halirutan It would be nice to have a Mathematica-blog aggregator.

Some blog platforms have separate feeds for tags, so it can be filtered for Mathematica-related posts.

@halirutan Your blog doesn't accept comments, right?

@halirutan We used to add them as ticker feeds, then after the flood change it to message feeds.

@Sascha Have you reported it already?

halirutan

@Szabolcs I have to check the Jekyll docs to see if and how this is possible.

@Szabolcs It used to have this turned on but I see now that it is no longer available. I have to check when I'm home.

@Szabolcs Hmm, we still have half of them as messages

I post usually not more than once every 2 months max. I guess this amount of spam is acceptable.

Szabolcs

13:39

@halirutan Sure, you should make it a message feed. What I meant was that it is possible to avoid the initial flood of messages by adding it as a ticker feed, then changing it to a message feed an hour later.

@halirutan If it's a static site, the only user-friendly solution is to use an external service like Disqus ... Some people put in an email form and then add the comments semi-manually. Or set up a server script that parses the email, adds it to the static site, and re-generates the site.

@halirutan I am not very happy with error handling in Mathematica and never know what it the best way to do things ... I would have tried to leave option checking to OptionValue (it already does this). Something like this: dropbox.com/s/sro5z5bwjc8ew1f/ErrCheck.nb?dl=0 But I'm not sure.

Yours may be more convenient though.

One risk with your approach is that sometimes the function pattern needs to be such that it prevents the OptionsPattern[] matching too many things. mathematica.stackexchange.com/a/124208/12

There are tricky things :-) Consider that MatchQ[{}, OptionsPattern[]] is True

halirutan

@Szabolcs I looked into the implementation of Plot and this is how they do it there.

About the feeds: There is an open PR for Jekyll that adds category feeds which would be exactly what I need.

About the comments: I had Google+ comments turned on but I merged my Jekyll framework to the latest master and that obviously went missing.

I'm also not really happy about the comments in static sites but that's life.

@Szabolcs OK, now I understood it.

Szabolcs

I'm just trying hard to break your function :-) But then in Mathematica, there will be a special case that break just about anything.

I haven't yet managed to find a very reasonable example where it doesn't work correctly. (I found a few less reasonable ones.)

halirutan

@Szabolcs Hehe :) I think @Jason should fiddle with the details.

Szabolcs

13:58

It wouldn't work in Replace, which has a Rule-second-argument, and it also takes options. But that is a very special situation.

That function won't even take more than one option: Replace[{2}, x_ -> x^2, {1}, Heads -> True, Heads -> False] shows an error.

Is there any reason why args__ is not args___?

halirutan

14:19

@Szabolcs Absolutely not.

Szabolcs

Error checking is such a pain. I would really like to see a general, but similarly compact (or at least easy to set up) solution for it. Also check number of arguments, validity of argument, and make sure that all checks work together.

2

Feeds

15:08

Wolfram Blog: Goodbye, 2017! It Was a Marvelous Year for Wolfram Research

posted on January 04, 2018 by Michael Gammon

Whew! So much has happened in a year. Consider this number: we added 230 new functions to the Wolfram Language in 2017! The Wolfram Blog traces the path of our company’s technological advancement, so let’s take a look back at 2017 for the blog’s year in review. Announcing New Products and Features The year 2017 [...]

halirutan

15:29

Btw, I'm still very proud that I have found such a fitting quote for the Ad of our chat room. It still seems to hold true that we are a very friendly bunch and I'm not sure we any flags in this room so far

Just saw it on the main page.

Edmund

0

Q: How to sort alphanumeric codes by their alpha and numeric tokens?

I have a large list of codes of the following form: codes = {"Fee.C21HJ.S4", "Fee.TOxC12.S1", "Fee.TOxC9.2 BB", "Fee.ROxC9.4", "Lag.C39"}; I would like to sort codes of this form by their alpha and numeric tokens. What I mean by "alpha and numeric tokens" are any runs of DigitCharacter or Let...

Vitaliy Kaurov

17:13

@halirutan love it ;-)

Folks this challenge took off: http://community.wolfram.com/groups/-/m/t/1257547
Please see some nice submissions in the comments

halirutan

@VitaliyKaurov I already had a look at this. I somehow expected a big or complicated data-set, but this is not the case. With such a simple table it's hard to come up with something that is outstanding.

C. E.

18:46

@Szabolcs Alas, there aren't any Mathematica blogs that write regularly AFAIK...

Vitaliy Kaurov

19:22

@halirutan i do not think it is about "outstanding", just creativity in different types of dataviz. you can see on that thread 4 current submissions are quite different. Nothing mindblowing, just fun ;-)

but fun of course is very subjective...

Chris K

@halirutan want to have a go at our compilation of the same kind of data? preston.kbs.msu.edu/reprints/files/… has > 2900 values :)

halirutan

@ChrisK Wow, and everything in an Excel-sheet :)

Feels almost like home.

Vitaliy Kaurov

BTW @halirutan you said: "If there are any Ruby enthusiasts here that want to take a look at the code, I would be delighted." Did you host the code anywhere? I am forwarding to some internal folks in case they wanna take a look.

halirutan

19:52

@VitaliyKaurov you can direct them to this pull request or my fork here. It is the feat_Mathematica branch.

Vitaliy Kaurov

23:39

@halirutan Thank you! As this is public, could you add this links to your community post? It will be much more efficient.

Transcript for