Maybe of interest here:

phys.org/news/…

@CarlLange. Thanks.

@channel I'm working on more toolbars to add functionality to my IDE (plus a way to more easily access these toolbars). Anyone have something they'd be interested in seeing implemented? I currently am thinking about writing one for creating tutorial-type websites like this one that can also be condensed down into a multi-section PDF / min-book.

Once you have the stuff to manage the directories/paging etc. all this stuff actually becomes pretty easy...

Something odd I'm running into. This won't update:

PopupMenu[
 Dynamic[EasyIDE`Notebooks`FormattingElements`Private`state$99501], \
{"BugTracker" -> "BugTracker", "CodePackage" -> "CodePackage",
      "Docs" -> "Docs", "Markdown" -> "Markdown",
  "Package" -> "Package"},
    BoxID -> "Toolbar"
 ]

But this will:

PopupMenu[
 Dynamic[EasyIDE`Notebooks`FormattingElements`Private`state], \
{"BugTracker" -> "BugTracker", "CodePackage" -> "CodePackage",
      "Docs" -> "Docs", "Markdown" -> "Markdown",
  "Package" -> "Package"},
    BoxID -> "Toolbar"
 ]

The only difference is I made the first one with Unique

@JasonB. @b3m2a1 I don't know anything at all about computational chemistry, but I was wondering is the SMILES string is unique. Wikipedia suggests that it is not. MoleculeEquivalentQ just compares SMILES strings. Is it the case then that in Mathematica all SMILES strings are canonicalized?

(Personal motivation: MoleculeEquivalentQ looks like vertex-coloured multigraph isomorphism.)

@Szabolcs as best I remember SMILES strings are not necessarily equivalent, esp. with respect to stereochemistry. On the other hand, there are extensions to the basic SMILES format that handle that too.

In terms of raw graph structure they should be though

Like the bond graph of a molecule doesn't know about stereochemistry so SMILES should capture that

It seems a bit tedious to enter Molecules using graph notation. All the Atoms and Bonds.

Yeah usually in chemistry we just import from file

Or use simpler input formats like what's known as z-matrix

Basically an iterative spherical coordinates spec

@JasonB. Sorry, I think I managed to break it.

From the docs,

> Two molecules are by default considered equivalent if there is a reordering of atom and bond lists that makes them the same.

bonds = Join[
   Partition[Range[6], 2, 1, {1, 1}],
   Partition[Range[7, 12], 2, 1, {1, 1}],
   {{6, 7}, {11, 13}, {2, 14}}
   ];

SeedRandom[123]
mol1 = Molecule[ConstantArray["C", Max[bonds]], bonds]
mol1 = Molecule[ConstantArray["C", Max[bonds]], RandomSample[bonds]]

MoleculeValue[{mol1, mol2}, "SMILES"]

MoleculeEquivalentQ[mol1, mol2]

@JasonB. The only difference between these two "molecules" is the order of the bond list, thus by the definition in the docs, they should be equivalent.

@Szabolcs, that code snippet shows mol1 and mol1, instead of mol2.

If I fix it then it works

@KraZug Thanks!

@JasonB. Sorry, I was wrong (but I can't delete the message above anymore).

np. @JasonB., what is the long-term plan for computational chemistry in Mathematica? Is it going to be a field where you are trying to compete with the specialised software?

@KraZug I don't know that there's much hope to compete against much of the specialized stuff. In my experience, computational chemistry breaks down into either problems that cannot be solved on a standard laptop simply due to memory/time/crunching constraints or problems that require a lot of specialized knowledge of a specific system.

But perhaps there's some space in the middle to explore

@b3m2a1 I will try to address your comments next week I am mostly busy/afk these days.

@b3m2a1, yeah, I showed the new features to a computational chemist I work with (with the caveat that I knew that it wouldn't be anything novel), and she was interested that they were going in that direction.

This is why I am frustrated with a certain (unfortunately very common) type of new poster and more and more often I just refrain from touching first-time posts: mathematica.stackexchange.com/q/195825

He literally posted 500 lines of code with a single cryptic comment: "How can I make the below Mathematica code small using Block by considering all possible way?"

I have great difficulty in seeing how anyone (perhaps other than the poster) could fail to notice that this is not a good question. This is not about SE rules, just common sense.

I tried to point this out by asking him to think about what the question would look like to a reader. It is in his interest to write an answerable question.

Now apparently I am "harrassing" him.

The easy way would have been to just vote to close without any comment. That wouldn't give him a chance to react early fix it.

@Szabolcs The substructure matching stuff does use vertex- and edge-colored isomorphism. I had originally implemented MoleculeEquivalentQ by checking that the atom count is the same and that one is a "substructure" of the other. But the SMILES canonicalization routine is good enough that for all the types of molecules we support, the same canonical SMILES string is found for the same molecule

I prefer to think of different bond types as edge colors rather than multi-edges. Even though we use the language of single, double, triple, it breaks down if you want to represent other types of bonds (aromatic, dative, hydrogen)

@b3m2a1 As to doing electronic structure on a single machine, there's always a tradeoff between the system size/basis set size/method, but there are many applications where you can get results on your PC. In grad school our group bought a single license of Gaussian and I ran it locally to get a potential surface

I don't think we have any plans to compete directly with software like molpro, GAMESS, or Gaussian. Being able to import the output from those files into a convenient and computable form is a more immediate goal

@KraZug I am interested in what direction users would like us to expand. I am interested in having a representation of a chemical reaction, and to what extent the graph and network functions can be used for retro synthesis

@JasonB., @b3m2a1, my colleague does use Cresset's software on her PC, as well as things on the cluster. I'm new to the area of chemistry and drug discovery, so I don't really know. Jason, is this your area that you are developing?

@Szabolcs I certainly think you did nothing wrong there

@b3m2a1 Unfortunately that also didn't work - there's some problem where the plugins button isn't clickable...

@KraZug Yes, I'm the developer for all of the Molecule functions in V12

@JasonB. Do you know what algorithm are being used for 1) substructure matching 2) canonicalization?

Hi all, please allow me to ask a quick question here, I might frame it as a full question posted later: If I have a definition of a function f[a][x_] := f[a][x] = x^2 and f[b][x_] := f[b][x] = x^3, is it possible to access all it's values (up or down, I am not sure) somehow like UpValues[f]? The case I am working with is more complex, but this is a minimal example that fails. I'd expect UpValues[f] above to yield a non-empty list, but it gives an empty one instead.

Or maybe you could suggest a material to read about that?

Are you supposed to get an email receipt when submitting a question to wolfram? I'm still trying to get my activation key working for 12

Sent support a message via the user portal. Web page says "We have received your message and will respond shortly." but no case number by email

@mikeonly It's called SubValues (not sure where it's documented).

@Szabolcs Thanks, that's it!

Hey guys...sorry for the late notice, but Stephen is going to stream a development meeting in about 10-15 minutes. We also have another livestream at 4:00 central on the Wolfram Research twitch channel that will be a live coding event. Lastly, I'm sure you already know, but Stephen will be continuing his V12 demo tonight at 7:00pm central.

How to do Transfer learning using VGG -16 or so on data that is just (1x80x80) ? I tried on pretrained Lenet-5 but I didn't succeed. This is because it was trained on 1x28x28

need to get new key from our local site manager .. crosses fingers

@b3m2a1 About the obsession: I don't get it either.

I made another extension for EasyIDE. Now for making little "book" notebooks and PDFs:

That notebook actually has no dependencies on EasyIDE anymore, so you could easily send it off to someone and they'd see exactly that.

If you specify an author, too, it'll show up on the first page