Wolfram Mathematica

WReach

01:13

@Mr.Wizard I've been playing with Dataset/Association for about a week now -- I was on holiday when V10 came out. I'm making up for it now :)

Mr.Wizard

01:27

@WReach Impressive. I would have guessed considerably longer. (I was wondering if you had a pre-release copy.)

WReach

01:43

@Mr.Wizard Why, thanks grin. No prerelease copy for me, but I'm guessing that Dataset was strongly influenced by SQL / XPath / XQuery and I've used those before.

RunnyKine

@WReach. What do you think about the capabilities/flexibility of Dataset vs. those ones you mentioned?

WReach

@RunnyKine They are hard to compare directly as each is making concessions to their typical data sources. Having said that, Query feels like it wants to attack the same problems as XQuery. XQuery has quite a bit more functionality than Query, but it is still early days for Query and XQuery is on the endangered species list as XML goes out of vogue.

I found the explicit treatment of ascending and descending operators in Query an interesting spin.

RunnyKine

@WReach. Thanks for the comparison.

WReach

@RunnyKine What is your take?

RunnyKine

02:01

@WReach I have no experience with any of those. I'm new to this kind of stuff :). I was just curious if our beloved Language is up to the competition, being the new kid on the block and if I should put in the effort to get good at it.

@WReach. I guess I'll be studying your answers carefully to learn from them :)

WReach

@RunnyKine I see Query as a welcome addition to the language as it brings yet another programming paradigm into the ~~Mathematica~~ Wolfram Language fold. Now we can program in query-oriented style, joining the good old procedural-, functional-, language- and transformation-oriented paradigms all without leaving our comfy little notebooks. What's next? O-O? Logic? Dataflow? Actors? :)

RunnyKine

@WReach. I agree, it's just so convenient working in Mathematica, I think now that the language has a name, they will definitely try to make it feel like one. I bet O-O is next on the list and I think (hope) @LeonidShifrin is secretly working on that :-) .

halirutan

02:17

@rm-rf Why don't we do some front end tweaking?

Michael Hale

@halirutan Very cool. I've wanted "surround selection with ..." refactorings a lot since I use triple-click and Ctrl+. so much.

halirutan

@MichaelHale Then is here a first version. Maybe we can make this really usable with the help of the community:

createSurroundWith[opener_String, closer_String, shortKey_String,
  moveForward_Integer: 0] := FrontEndExecute[
  FrontEnd`AddMenuCommands[
   "InsertSplitBreak", {MenuItem[opener <> "expr" <> closer,
     FrontEndExecute[Block[{nb = SelectedNotebook[], expr},
       expr = NotebookRead[SelectedNotebook[]];
       If[expr === {}, expr = ""];
       NotebookWrite[nb, RowBox[{opener, expr, closer}]];
       SelectionMove[nb, Previous, Character,
        StringLength[closer] - moveForward]
       ]], MenuKey[shortKey, Modifiers -> {"Control", "Command"}],

createSurroundWith @@@ {{"(", ")", "p"}, {"{", "}", "b"}, {"<|", "|>",
    "a"}, {"(", ")&", "f", 2}};

@MichaelHale It's always Ctrl+Alt+... where ... is p, b, a or f

Ah, yes. And the Module surrounder:

FrontEndExecute[
 FrontEnd`AddMenuCommands[
  "InsertSplitBreak", {MenuItem["Module[{var},..]",
    FrontEndExecute[Block[{nb = SelectedNotebook[], expr},
      expr = NotebookRead[SelectedNotebook[]];
      If[expr === {}, expr = ""];
      NotebookWrite[nb, RowBox[{"Module[{var},", expr, "]"}]];
      SelectionMove[nb, Previous, Expression];
      SelectionMove[nb, Next, Characters, StringLength["Module[{var"]];
      SelectionMove[nb, All, Word];
      ]], MenuKey["m", Modifiers -> {"Control", "Command"}],

Michael Hale

Fun. I'll mess with it some and let you know if I add some scenarios. I can't think of all of them off the top of my head, but since you have this started I'll take note of them for sure next time I encounter them.

halirutan

@MichaelHale In IDEA we have currently the following Live Templates: github.com/halirutan/Mathematica-IntelliJ-Plugin/blob/develop/…

@MichaelHale I consider one of the most useful things to be able to surround an expression with f[...] where the f is selected so that I can instantly start typing.

Michael Hale

Yes

That is the one I was just thinking.

halirutan

02:30

@MichaelHale Just look at my Module impl. The rest is easy.

Michael Hale

Will do.

Will be very useful when I need, for example, to sort an expression that is an argument to a function already, and the expression is using an infix operator so I can't just put Sort@ in front, or if I can't use a prefix form because I need to add an argument.

Another one I've thought about is extending that "roll up" command to work with arbitrary cell selections that use % or %n.

Like I often write some code to import some data. And I refer to it as %n in the next cell so I don't keep re-evaluating the import. They have the roll up feature for when you click the suggestion bar, but not for when you use % on your own.

Mr.Wizard

11:58

@WReach It's seem I'm behind the times as usual; in which applications is XML out of vogue and what is replacing it?

halirutan

13:06

Is someone here who as V10 and Adobe Illustrator (on OS X)? When I open this in Illustrator I see only boxes instead of ticks:

Export["tmp/test.pdf",  Plot[x, {x, 0, 1}, LabelStyle -> {FontFamily -> "Times"}]]

although Adobe Acrobat tells me that all fonts are included in the pdf.

WReach

13:23

@Mr.Wizard I'm referring to how the "x" in AJAX and in XHR stands for JSON :D

A few years ago, it seemed like XML was poised to take over the web, with REST and RPC endpoints serving up XML and XSLT generating HTML, SVG, etc. My impression is that JSON + JQuery (and rivals) have taken over that role.

When XML had that huge mindshare, development of technologies like XSLT (a beautiful functional/transformational language hidden behind an ugly facade) and XQuery (a beautiful functional/query language, full stop) proceeded rapidly. A lot of that mindshare is now looking at HTML5 / Javascript, and I note that the development of those techonologies is much, much slower.

rm -rf

@halirutan I don't have Illustrator now to try it, but IIRC, it only works with fonts available to it from the system.

You will have to add Mathematica Sans (or anything else) to ~/Library/Fonts, I think.

WReach

@Mr.Wizard I still harbour the secret hope that something like XQuery will take over for SQL some day. I still use Oxygen XML and BaseX.

rm -rf

@WReach Do you find Dataset useful? (i.e. is there a good, non-trivial use case)

halirutan

@rm-rf It doesn't work with "Times" but it works with "Times New Roman" although Acrobat tells me a different font name which is something like Times-Roman in the first case.

rm -rf

@halirutan There's a "Find Font" option in Illustrator, which you should be able to click and select the right match (I realize that ideally, you'd like to avoid this since all the info is in the PDF)

WReach

13:28

@rm-rf I do find it useful in that it allows us to express algorithms in query-oriented style. At the moment it appears to be mainly syntactic sugar as opposed to providing some highly optimized access patterns for hierarchical data. The syntax allows for the latter to implemented some day, and for access to external data sources as well. Keep watching this space :)

halirutan

@rm-rf My use case is that I often use a different font when exporting graphics for publication. What if I want to adjust the position of legends afterwards or do some other fine tuning? Hmm, ..

(Btw, if I export to eps, Illustrator CS6 crashes if I try to open it :-) )

rm -rf

@WReach Thanks :) I was hoping that it would end up being a nice way to do SQL-like stuff in the notebook, but the overhead from all the key duplication makes it a memory hog

WReach

@rm-rf Yes, one has to be very, very careful with hash tries, or shared references can gobble up all of your memory. Of course, "memory hungry" is nothing new to WL :D Still hoping for that "optimization some day" potential of Dataset.

rm -rf

Hehe

Artes

@RunnyKine @michaele2 @bobthechemist Could you explain how this post is not a duplicate of this one Contracting with Levi-Civita (totally antisymmetric) tensor? I've asked you in the comments to that post but got no response. Have I overlooked anything or just don't understand a "deeper level"?

rm -rf

13:48

@Artes RunnyKine's and bobthechemist's names don't appear anywhere on the post... am I missing something?

WReach

@rm-rf As belisarius said:

"If you optimize on V1, you're spoiling the next contract" (one of my first managers, eons ago) — belisarius Dec 17 '12 at 4:26

off for my morning commute, see ya...

rm -rf

bye

bobthechemist

14:10

@Artes Thought I skipped that one, since the phrase Levi-Civita Tensor scared me.

Unless I'm missing something, curated data functions that return new objects no longer work for v9. This fails for me: ElementData["Radon", "DiscoveryYear"]

chuy

What does this return? PacletInformation["ElementData"]

bobthechemist

{"Name" -> "ElementData", "Version" -> "9.0.34", "BuildNumber" -> "",
"Qualifier" -> "", "MathematicaVersion" -> "9.0+", "SystemID" -> All,
"Description" -> "", "Category" -> "", "Creator" -> "",
"Publisher" -> "", "Support" -> "", "Internal" -> False,
"Location" ->
"C:\\Users\\BoB\\AppData\\Roaming\\Mathematica\\Paclets\\Repository\
\\ElementData-9.0.34", "Context" -> {}, "Enabled" -> True,
"Loading" -> Manual}

@chuy you thinking that I need to reload the paclet data?

chuy

I am not sure, this is using version 9.0.1 correct?

bobthechemist

Yup, have both v9 and v10 on the same computer, works in v10 (returning a DateObject). Examples in the v9 documentation page for ElementData all fail now as well.

chuy

so it returns $Failed in v9.0.1?

bobthechemist

14:24

yup

chuy

hmm I don't have that issue, I get DateObject[{1900}] in v10 and 1900 in v9

I would try PacletCheckUpdate["ElementData"] maybe

bobthechemist

Thanks for checking - it was working for me, and I started to worry that something on the Wolfram end had changed; must be a local problem.

chuy

you could also just remove the directory indicated by

'"Location" and that should force a new paclet download

Michael E2

@Artes I didn't get the comment ping - or missed it. IMO, if the other two answers are what the OP is after, it is a duplicate. My answer adds SymbolicTensors`ArrayContract, which is not considered in the potential duplicate. I have doubts whether my interpretation is correct, but twice the OP seem to indicate that nu is a free variable (in Q and in a comment).

Artes

@rm-rf I mean I asked them here [Sum with Levi-Civita ](mathematica.stackexchange.com/questions/56063/…). They voted to leave that post open but as far as I understand it was an exact duplicate of another one.

chuy

14:28

btw bob I really enjoy your raspi projects

bobthechemist

Thanks Chuy, I've been having fun this summer. Sadly, it's time to get back to lecture-writing, so I don't know how much raspi work I'll be able to do.

RebuildPacletData[] did the trick; the location pointed to in "Location" /. PacletInformation["ElementData"] didn't exist.

Michael E2

@Artes Well, assuming the OP (seen 1hr ago) has seen my answer and still has accepted QuantumDot's, I would consider casting a close vote, unless you think my interpretation is sufficiently distinct from the dupe.

Artes

14:43

@MichaelE2 Thanks for your response. I think that duplicates to earlier posts make this site disordered. Of course there is some kind of freedom of deciding whether a post is a duplicate or not however in this case it seems to me quite clear it's a duplicate. Your method seems to be an overkill but I haven't known it, so +1.

Michael E2

15:08

@Artes Thanks. I got it from the docs, so I'm not sure about overkill. I mean sum/table is sufficient for dealing with symbolic tensors without TensorContract, but contraction is a useful operator. And applying it with List instead of Plus might be convenient for a symbolic tensor. I don't understand why TensorContract itself does not have an argument for a function, instead of relegating it to ArrayContract.

@Artes I had already upvoted your answer to the original, but I just upvoted the question, too. It bothered me that the new dupe had more votes.

rm -rf

@Artes ahh, you were referring to their votes on the review... Looks like the first flag was "unclear what you're asking", so the review question would have been "Should this post be closed as unclear" in which case the votes were fine. In any case, you cannot ping someone who has not commented on a post, which is why you didn't get a response :)

acl

@rm-rf @Artes I have to admit, this looks like a "read the manual" question

Artes

15:27

@rm-rf Sometimes I used to ping someone this way like here...

Michael E2

@acl Personally I feel questions should be closed if there is a simple and clear answer in the docs. We would still argue about what's clear enough, but "easily found" is not a characteristic of the docs that people readily acknowledge. I think "found in the docs" + a link is good enough for closure. Plus the closed question can still be guide to new users.

acl

@MichaelE2 Yes, I do agree. But I would say that posting a question with incorrect syntax (LeviCivita instead of LeviCivitaTensor), missing brackets and not realising that if the output is expected to be a tensor of rank $n$ then we should expect lists to depth $n$ hints at not having spent any time thinking about the question

which isn't really a problem here, but the number of questions by people who have not bothered to even try to solve their problem before asking them is increasing

or rather

Michael E2

15:52

@acl I don't get it -- from a site Q? If we're just sharing strange behaviors, in V10 Beta, the output formatting of foo generates errors: foo = Quiet@NDSolve[{y'[t] == -y[t], y[0] == 1}, y, {t, 0, Infinity}]; foo.

acl

@MichaelE2 No sorry, I was just venting. Your example throws up errors in V10 final too

rm -rf

16:34

@Artes Ah, at the time, I was the last editor... you can also ping the last editor and also a moderator that closed the question (if the question is still closed) :) Lots of complicated rules... I'm slightly embarrassed that I remember them. There are much better things to remember instead...

rm -rf

16:44

belated response: You're welcome :)

PlaysDice

16:57

Is there a way to feed back the result of a Reduce into the original equation? (toy example...) whereas Solve[x^2==16] gives rules { x->4, x->-4 }(easy to feed back into the formula), Reduce[x^2==16] gives x=4||x=-4 - I cannot see how to feed this back.

chuy

{ToRules@Reduce[x^2 == 16]}

PlaysDice

@chuy thanks! I also like the use of outside brackets to turn Sequence into List.

Gabriel

22:33

is there a trick to getting the c compiler to work in windows 8? I have usually just installed the windows sdk 7, but this doesn't seem to work in 8

Transcript for