Wolfram Mathematica

2:14 AM

@rcollyer pictureshack.us/images/6774essai.gif and pictureshack.us/images/7970anthracene.gif

I remember telling you about my project, here's what it looks like now. The mechanics is coded, I now have to implement multiple electrons (slater determinants). Fun.

rcollyer

2:42 AM

@CHM Those are awesome pictures. Thanks.

CHM

:)

I've yet to figure out how to find the period of my complex exponentials

rcollyer

Check out this code.

Clear[GenerateSymmetries]
GenerateSymmetries[mat_?SquareMatrixQ] :=

 Module[{symmat, vars, s, p, l = Length@mat, res},
  symmat =
   Table[If[i <= j, Identity, Reverse[#] &] @ s[i, j] +
     I Which[i < j, Identity, i > j, -Reverse[#] &, True, 0 &] @
       p[i, j], {i, l}, {j, l}];
  vars = Union@Cases[Flatten@#, (s | p)[_, _], Infinity] &;
  res = ReplaceAll[symmat, #] & @
    Quiet@Solve[ Commutator[ mat, symmat] == 0, vars@symmat ][[1]];
  ReplaceAll[res, Thread[vars[res] -> #]] & /@
   Permutations[{1}~Join~ConstantArray[0, Length@vars@res - 1]]

@CHM Good question.

GenerateSymmetries@ {{a, b + I c}, {b - I c, d}}
(*
{{{0, I + b/c}, {-I + b/c, -((a - d)/c)}}, {{1, 0}, {0, 1}}}
*)

CHM

Commutator is not defined

rcollyer

Sorry forgot two functions:
SquareMatrixQ[_]:= False
SquareMatrixQ[a_List?MatrixQ]:= Equal@@Dimensions@a;

And,
Commutator[ a_?SquareMatrixQ, b_?SquareMatrixQ] /;
Length[a] == Length[b] := a.b - b.a

CHM

Whoa

=)

It doesn't work with Sparse matrices, but that's why there's Normal[].

rcollyer

2:49 AM

I'll fix that at some point.

I'm not sure it is efficient, but automatically generated symmetries!

CHM

yeah, that's definitely going to be useful.

thanks

rcollyer

I've been thinking about it for a while, working on tight-binding and all, and I figured it should be straightforward.

As the symmetries are by definition Hermitian, that implies they should be generators for a Lie Algebra.

CHM

(I've not encountered Lie algebras before.)

But I'm sure they'll come in handy when I want to cut down on computation time, especially since I'll be working with molecules that have a certain degree of symmetry

edges2mat[edges_] :=
Module[{nodes, tmat, n,
smat},(*custom flatten to allow edges be lists*)
Clear[nodeMap];
nodes = Sequence @@@ edges // Union // Sort;

nodeMap = (# -> (Position[nodes, #] // Flatten // First)) & /@
nodes;

n = Length[nodes];

tmat = (({#1, #2} -> 1) & @@@ (edges /. nodeMap)) //
SparseArray[#, {n, n}] & /. nodeMap;
smat = tmat + Transpose[tmat]];

mmol = ImportString["C1=CC=CC=C1", {"SMILES", "EdgeRules", 1}];

hamiltMol = edges2mat[mmol];

GraphPlot[hamiltMol]

rcollyer

You have, but they weren't called that. The angular momentum matrices form a Lie Algebra, and are the generators of rotation which is really what I was looking for.

CHM

That's how I get from a SMILES string to a Graph. edges2mat is a slightly modified version of an answer give on SO by Yaroslav Bulatov

@rcollyer oh.

I've used three different methods for generating my molecules, and settled with that one. It's easier, cleaner, and requires a lot less steps. Why reinvent the wheel?

Back in 5-10. GF call.

rcollyer

2:59 AM

I can appreciate that.

rcollyer

3:09 AM

GenerateSymmetries isn't perfect. One, the above version won't include the identity if it is not an explicit result from the calc. Two, it only generates 10 (11 counting the identity) symmetries for the benzene ring from your code; it should be 12.

CHM

Hmm

rcollyer

Not sure what went wrong with it.

Need to look over the symmetries it generates and see what it missed.

Any way. Have a good night, and if you figure it out, let me know.

CHM

That's what I'm trying to figure out with webqc.org/symmetrypointgroup-d6h.html. Even if I get the forgotten one, I'm probably not going to be able to fix the code :P

Night

rcollyer

3:37 AM

@CHM For the symmetry count, I was looking at D6, not D6h. But, since $[H, A] = [H, B] = 0$, $[H, AB] = 0$, so the algorithm is likely only finding part of the group, the rest of which can be generated by creating the multiplication table.

Now, I'm off to bed.

CHM

5:57 AM

@Szabolcs I've read your comment. Those languages look very interesting. Will take a look.

Haha, funny you've just logged in.

yoda

6

Q: Superposing music tracks in Mathematica & MIDI pedal events

All of the following refers to music/MIDI (SoundNote objects), not sampled sounds. Unfortunately Mathematica doesn't seem to be able to import MIDI. I'm trying to make a simple MIDI importer based on MIDI <-> CSV for single-instrument files (focused on piano). What is the simplest way to su...

Did you end up writing your MIDI representation @Szabolcs?

Szabolcs

6:19 AM

Not yet ... not time now. But I haven't given up on it. I have some MIDI files to play with in Mma :-)

Why is the background of some posts blue? --> i.stack.imgur.com/VnUzP.png

CHM

I've wondered too.

Maybe it has to do with the number of views per x time.

Mine has blue background, 7 views in less than an hour. :/

CHM

6:35 AM

@Szabolcs hmm. Pd is not what I expected. I thought it would be programming.

Szabolcs

@CHM I haven't used it much ...

@CHM It's a special kind of programming :-)

CHM

It actually looks difficult to use! Lol.

Yeah... well it's click-gramming

Szabolcs

Well, I looked at it several times, but I never did anything interesting with it. The only similar "language" that I have actually done something worthwhile with is Grasshopper.

@CHM One real language that you might be interested in that has an interface to Pd is Pure. I don't know much about this language and I can't program in it. But I know it's based on term rewriting just like Mma. It seems to be the only practically usable general purpose language based on term rewriting.

CHM

I'm not familiar with "term rewriting", but I'll give it a try. Processing is more promising, currently doing the tutorial.

What actually interests me is to program sounds.

create* them programmatically.

Szabolcs

@CHM chuck.cs.princeton.edu

It seems NSolve can return Deferred results! (???)

Try this:

f[x_?NumericQ] := Sin[x]; NSolve[f[x] == 0, x]

and reevaluate the output

CHM

6:53 AM

hmm

I'm not sure what you're seeing.

Szabolcs

I'm seeing f^(-1)[0] which should auto-evaluate to 0, but it doesn't until I press shift-enter again

CHM

(i.stack.imgur.com/uN7JC.png)

Szabolcs

@CHM Yes. Now go into the output cell and press shift-enter.

CHM

Done.

I see.

Why have you found that?

Szabolcs

I was looking for a function that will at least attempt to find all roots in an interval. I was testing if NSolve can do this when it doesn't know what's inside the function (in old versions it couldn't, but lately it does some amazing things)

I saw the output and I was surprised because I knew that InvserFunction will usually try to evaluate the inverse numerically when floating point numbers are passed to it

So why didn't it evaluate?

I re-typed it and then it did ...

I copied the output, and that evaluated as well

I went into the cell and pressed shift-enter ... that worked as well.

It behaves like Defer

Back to the original question about finding all roots in an interval, see this: mathematica.stackexchange.com/a/3022/12 (All the comments on Heike's answer, for the full story)

I wanted to come up with a simple way to find the root on the part of the surface which is visible (closer to the viewer)

What happened to Arnoud? He hasn't been around for a long time.

CHM

7:05 AM

Interesting.

(Beyond my competences)

I'm wondering if Processing could generate smoother animations than Mathematica.

I've never used Java, but the tutorial is pretty easy.

Szabolcs

9:06 AM

For real-time animations, definitely go with Processing. It was made for this. For pre-generating and exporting animations, everything that's not buggy will be equally smooth

@RebeccaChernoff If there's anything more we can do to help with the syntax highlighter, or if you have any concerns, please do let us know!

Szabolcs

12:34 PM

@yoda What do you suppose has happened here ? Just curious.

Heike

1:09 PM

@Szabolcs Same thing happened to me.

And to belisarius and Mr.Wizard as well by the looks of it

Brett Champion

@Heike According to the graph on SO, my rep "dropped" by 85, but the total is still where I expect it to be. I don't see anything in /reputation.

Heike

@BrettChampion My rep did drop, but only by 40 points or so.

Brett Champion

Although the list by date shows a series of unupvotes about an hour ago.

@Szabolcs @CHM The questions with blue backgrounds are the ones that are tagged with one of your favorite tags.

Szabolcs

1:42 PM

@Heike I think someone has deleted their account. I wonder why.

yoda

1:54 PM

@Szabolcs I think this is related: meta.stackoverflow.com/questions/125740/…

I lost 1.7k, and judging by the questions voted on, I think it was an old colleague of mine

Eli Lansey

ouch!

yoda

leonid dropped 500ish

gaaah, now I'm below 15k. Should work up that 2.5k in spent bounties back to push me over 15

but I hate going back to answering MATLAB :)

@Szabolcs congrats on the 10k (almost double 10k)! :)

Szabolcs

@yoda What happened? We just got back the votes. That's why it went back above 10k

At least some of us. You didn't get back all of them yet.

yoda

Hmm.. looks like only you got back

Szabolcs

Heike also got them back.

yoda

2:11 PM

@Szabolcs Looks like someone upvoted you all for that... see the timestamps on the upvotes

it's only a few mins ago, and don't all have the same time stamps — i.e., sequential, whereas the unvotes are all the same time

Heike

@yoda Lets hope you and the others will get their rep back as well (and that we don't lose it again due to suspected serial upvoting)

yoda

oh dear, not that again

Szabolcs

3:56 PM

@yoda If that is the case, I wish they didn't. Rep is not that important. And it's easy to trigger the vote fraud detection.

yoda

4:09 PM

@Szabolcs @CHM The system automatically detects your "favorite" tags (if you haven't set any) based on the questions you answer and the tags you frequent. These are highlighted in blue

rcollyer

6:57 PM

@Szabolcs @yoda @Heike I was hit also, but only for 90. While not a major blow, annoying. And, no I haven't gotten them back.

CHM

7:24 PM

@yoda @BrettChampion thanks

CHM

8:33 PM

@Heike I've replied to your answer.

Heike

8:50 PM

@CHM I've seen your reply. I've tried my code with your timeList and it seems to work fine for me:

imgur.com/zIMKh

CHM

Hmm. There's something I'm doing wrong

I can get it to work but the graph is tiny.

Heike

@CHM What code are you using?

CHM

Mine, but with your modifications added.

ighamiltMol =
Dynamic@AdjacencyGraph[hamiltMol, GraphStyle -> "SpringEmbedding",
VertexShapeFunction -> ({{Opacity[0], Disk[#1, #3]},
Disk[#1, #3 timeCoeff[#2]]} &), VertexSize -> 4/3];

Animator[Dynamic[t], {0, 19}, .5]
ighamiltMol

It might be that the EdgeShapeFunction option is missing....

It was. Weird!

Heike

@CHM I added EdgeShapeFunction because the edges weren't drawn correctly without it. I didn't realise the default setting would screw up other things as well.

CHM

9:07 PM

@Heike yes that's what happened to me, the edges were transparent.

Thanks a lot

Heike

@CHM You're welcome

yoda

@Heike I don't remember if it was you or @acl that was talking about julia —but have you by any chance gotten it to work on a mac?

Heike

@yoda I haven't looked at it yet to be honest.

yoda

alright, no problem :)

Heike

9:29 PM

@yoda do you have problems installing?

yoda

yeah, the version I had was one from 2 weeks ago, so I just checked out the latest one and tried again (in the hopes that it has been fixed) and it's building right now (hope it completes)

Heike

9:43 PM

@yoda I wonder if it will be added to MacPorts

yoda

ld: symbol(s) not found for architecture i386
collect2: ld returned 1 exit status
symbol(s) not found for architecture i386
collect2: ld returned 1 exit status

Heike

@yoda :-(

yoda

I use brew, but it's not there yet

Heike

@yoda Nor in fink

Szabolcs

10:15 PM

@yoda if you have trouble, write them (mailing list). They're really friendly. I corresponded a bit with them about the benchmark (have an almost finished Mma translation, but didn't have time to properly test yet)

acl

@yoda yes compiled fine for me

@yoda does it say which symbols?

(do you have a fortran compiler etc?)

acl

10:40 PM

I just spent 30 minutes trying to set the date of something to 15/03/2012 in applescript and could not figure it out. then i randomly moved something and it worked; I have no clue why. what where these people thinking when they designed this atrocity?

(rant over, I just happened to have this window to hand when I wanted to let off steam)

Heike

I never got the hang of AppleScript. It's too fuzzy to my taste.

acl

yes, fuzzy is the right word

OK, it's meant to be "friendly-looking" or something, but really, I know more than 10 languages (well, "know") and still can't predict what will work in applescript. how is this not badly designed?

Heike

it's the Wolfram|Alpha of script languages

acl

10:56 PM

@Heike ha

yoda

@acl yeah, a bunch. Can't make head or tails of it though: pastie.org/3611467

And yes, I have g++, gcc, gfortran and m4

Verbeia

Looks like the Area 51 stats don't sync with the traffic numbers on the StackExchange site proper. Currently reading 809 and 757 respectively.

acl

@yoda really useful! maybe it's related to "ignoring file /Users/username/Desktop/julia/external/openblas-v0.1alpha2.5/libopenblas.a, file was built for archive which is not the architecture being linked (i386)"?

Heike

Looks like a problem with the LAPACK package

yoda

Yeah, I did look through the dev-list the last time and someone else had mentioned the same. Can't find the thread though

Heike

11:11 PM

Why can't macports just distribute binaries? Every time I update the database I have to wait for half an hour for the installed packages to recompile.

Happy π-day by the way

yoda

π day?

wasn't that a couple of days ago? (that too, only if you write the date the way americans do)

Heike

@yoda Yes you're right. Why did I think it was tomorrow?

There's pi-approximation day on 22 July

acl

i thought it was exact? 3.14, right?

(in the spirit of an exam a friend once was marking, in which there appeared the following now-legendary statement: "1560=1200, therefore...")

yoda

If you write your dates as DD/MM, then 22/07 is pi approximation day (since 14th march doesn't work)