Wolfram Mathematica

@MonireJalili I'm afraid I'm no expert in that. On the other hand, there is a wealth of questions dealing with how to do list operations fast here. Here's one I asked:

12

Q: Most efficient numerical selectBetween

Often when I'm working with numerical data I want to select those points within a given range. To do this I generally do something like: Select[samp, min < # < max &] or if I'm feeling lazy: Select[samp, Between[ {min, max} ]] But today I was thinking about how inefficient this is and deci...

performance-tuning

Here's one that links from that:

30

Q: Finding all elements within a certain range in a sorted list

Suppose we have a sorted list of values. Let's use list = Sort@RandomReal[1, 1000000]; for this example. I need a fast function window[list, {xmin, xmax}] which will return all list elements $x$ for which $x_\textrm{min} \le x \le x_\textrm{max}$. How can this be implemented in Mathematica? I...

list-manipulation performance-tuning programming algorithm

Here's one off that:

92

Q: What best practices or performance considerations are there for choosing between Cases, Position, Pick and Select?

Cases, Select,Pick and Position each have different syntaxes and purposes, but there are times when you can express the same calculation equivalently using either of them. So with this input: test = RandomInteger[{-25, 25}, {20, 2}] {{-15, 13}, {-8, 16}, {-8, -19}, {7, 6}, {-21, 9}, {-3, -25}, ...

functions performance-tuning pattern-matching filtering function-comparison

And building off those you should have a good sense of what operations are efficient/not

Also read the obligatory note about packed arrays:

108

Q: What is a Mathematica packed array?

A simple sounding question with a few sub questions: What is the difference between unpacked vs packed array? Are packed arrays more space efficent, how much so? Are packed arrays more time efficient for certain types of access over the unpacked form? Bonus: Is it ever undesirable to make u...

list-manipulation performance-tuning programming data-structures packed-arrays

Monire Jalili

thanks. I'll look into this

One quick question

What is the function that returns 1 for when the difference of two values are zero?

b3m2a1

2:37 AM

@MonireJalili not sure if there's any built-in one but here's an option that'll be fast:

UnitStep[-Unitize[table1 - table2]]

Monire Jalili

Great! Thanks

b3m2a1

Only the equivalent positions will be 1 everything else will be 0

b3m2a1

8:50 AM

I've been trying to convince myself to write documentation all day. Instead I created a palette to work with SimpleDocs...

On the plus side I now have a partial palette for working with SimpleDocs

I'm gonna do a video showing how it works, I think, and then write a stub blog post about it

Chris K

9:19 AM

@b3m2a1 looking forward to seeing it

Szabolcs

10:41 AM

@MonireJalili Check out BoolEval bitbucket.org/szhorvat/booleval/src/default

BoolEval[a==b] returns an array that contains 1 where the elements of the arryas a and b are the same.

Carl Lange

11:16 AM

@Szabolcs SW would say that's a good candidate for the function repository :)

Kuba

12:15 PM

CreateDocument[
 Graphics@Disk[],
 WindowFloating -> True
 ]

Can't evaluate enything in the original notebook when the created document is viisble

Can you reproduce it?

Szabolcs

@CarlLange I don't like the idea of the function repository. This kind of functionality should go in a package, not a single function. The BoolEval package has multiple functions.

3

All that effort that went into the function repository would have better been spent on a package repository.

They just ended up having to solve multiple problems that are created by rejecting the package concept. Such as dealing with namespaces.

Those solutions then create additional problems. Why do we have to wrap everything in ResourceFunction instead of having a proper symbol? Then things like messages don't work as they should anymore. Then additional workarounds have to be implemented, which no doubt will come with their own compromises and drawbacks.

Any non-trivial functionality will involve multiple functions, not just a single one.

Even when the functionality can be packaged into a single function, it would typically benefit from being broken down into pieces, or at least exposing some of the sub-functions it uses internally.

Kuba

12:38 PM

@Szabolcs I am surprised you gave it a try. I consider it a slap in a face of everyone who was asking for more attention for serious developers.

I don't claim it is useless though, probably fine for interactive use/quick prototyping etc. Otoh maybe not, have not tried it at all.

b3m2a1

7:03 PM

@Szabolcs I've been brining that up to WRI people whenever they talk about the function repository, but based on the paclets live stream I think many parts of the company (okay just SW) didn't realize to what degree the current package stuff is in use.

2

@Kuba Can't reproduce on my end (macOS)

Carl Lange

8:45 PM

What slightly concerns me about the function repo is the idea that it will be a warehouse for functions that people just don't know about. I worry that it'll just be where functions SW doesn't think are important will go to die, even if they could be used by everybody and might as well be a part of the core language.

But then there are a few things I've written where I think, OK, this WKT parser is literally a single function and is useful to more than just me, maybe it should be published...

But then again, I also think that a WKT parser should be an Import format or similar, there's no reason for it to not be a part of the core language (or at worst, a GIS` function)

b3m2a1

9:09 PM

@CarlLange you can write a paclet that supplies that WKT parser as a direct Import format

That's really the way it should go. You have a useful function/set of functions and so you expose them through the standard paclet mechanism, not the WRI ad-hoc FunctionResource framework.

Also gives you much better control over what goes on. If you want an example of where this is used, take a look at this: github.com/c3m-labs/ImportMesh/blob/…

(excuse the formatting on that, tab-based as opposed to space based)

But basically you write a paclet, then you register the format inside the paclet

Carl Lange

9:41 PM

Well, it's a bit weird for that specific example because you don't get WKT as raw files, you get it as strings inside other data (like the result of a database query or in a CSV or so on), so maybe an Import format doesn't necessarily make sense. I dunno. I suppose I'm just bemused that it's not already supported within the language

b3m2a1

@CarlLange ImportString :)

Carl Lange

In any case, yeah. I dunno. I think there are some cases where the function repo makes sense - I often see functions on the SE that are pretty much exactly right for that type of thing.

@b3m2a1 I suppose so :)

eg perhaps something like mathematica.stackexchange.com/a/98161/57593

b3m2a1

@CarlLange sure, but those could also just be cooked into a proper paclet (something like SETools). In fact, one could write an SE scraper that finds highly-upvoted functions and automatically packages them up (minor cleaning would be required) and provides the links to the answers in the usage message.

Carl Lange

@b3m2a1 Ha, that's a nice idea

b3m2a1

It's one of those things I've had on my mind for a while but never got around to

Carl Lange

9:47 PM

Is the little orange widget thing new for the function repo? resources.wolframcloud.com/FunctionRepository/resources/FileQ

Is that what it'll look like in 12?

b3m2a1

The little circle? Yep

You can see it already live in the cloud

Carl Lange

Huh

I'm not completely sure if that's great, since it's not typeable, right?

well, maybe it's semi-typeable, like Quantitys

b3m2a1

It definitely is not. But Richard Henninagan (?) devised a typeable template box for it that he demoed here a month or so ago.

Sometime after the Function Repository live stream I think

Carl Lange

Oh, it's in 11.3 as well apparently.

Hmm, would be nice.

b3m2a1

Oh?

How do you get it?

Carl Lange

9:50 PM

b3m2a1

Oh by updating the "ResourceSystemClient"

Carl Lange

And I definitively do not have 12 :)

Hm, InputForm of it is

b3m2a1

This is definitely another nice feature of using paclets as a distribution mechanism. WRI can back-patch stuff.

Carl Lange

ResourceFunction[ResourceObject[<|"Name" -> "BirdSay",
   "UUID" -> "a24858c8-45c5-4020-a8c2-c23bd5d26f04", "ResourceType" -> "Function",
   "Version" -> "1.0.0", "Description" -> "Have a bird say an expression",
   "RepositoryLocation" ->
    URL["https://www.wolframcloud.com/objects/resourcesystem/api/1.0"],
   "WolframLanguageVersionRequired" -> "10.0", "SymbolName" ->
    "FunctionRepository`$0945b71887c34426994f0d8340e95064`BirdSay",
   "FunctionLocation" -> CloudObject[
     "https://www.wolframcloud.com/objects/5e407f09-81ca-4549-a06c-2c4c982bb88d"],

That's pretty cool.

Yeah, I like that stuff gets quietly updated behind the scenes.

(Provided nothing breaks, but in fairness I've never seen that happen)

(Except in the livestreams, but none of that is released anyways :) )

b3m2a1

@CarlLange there was a disaster once where every Mathematica crashed on startup because of a bad paclet update

Carl Lange

9:54 PM

Haha

Oh dear

b3m2a1

But that's the kind of thing WRI should just catch

Like the fact that they did not catch it makes you wonder what QA was doing

Carl Lange

It's mildly surprising that they don't do a tiered rollout of these things

Entertaining function: resources.wolframcloud.com/FunctionRepository/resources/…

🤷‍♀️I think I'm basically fine with the function repo. It's nearly as useful as the neural net repo, but I suppose it's pretty easy access to some goofy functions resources.wolframcloud.com/FunctionRepository/resources/…

b3m2a1

@CarlLange I like to read the definitions of these things

You occasionally find useful stuff that you'd only know about if you spelunked a lot or worked at WRI

Carl Lange

@b3m2a1 "Download definition notebook", at the bottom of the page

b3m2a1

@CarlLange that's exactly where I find the defs

Carl Lange

10:07 PM

Ah, sorry, I thought you meant you would like to

Yes, seems very useful

Hmm, interesting stuff. I hope they've upped their timeline for the paclet repo though.

b3m2a1

11:07 PM

@CarlLange considering that 12 is like 50/50 never going to come out it's hard to be optimistic on that

Carl Lange

continue waiting

b3m2a1

Continue waiting never works

That dialog is the symbol of doom

Carl Lange

Not as bad as the "Formatting Notebook Contents" one.

b3m2a1

It's basically saying "abandon hope all ye who see me here"

Carl Lange

Anyway, what I'm trying to say is WRI are conditioning us to be patient :D

b3m2a1

11:09 PM

L o l

Or you get impatient and just implement it yourself ¯\_(ツ)_/¯

Lol SE thinks the \ in the shrugging emoji is an escape character so you have to format it as code to get it to show

Carl Lange

ah, I never thought to format it, I just decided the arm was implied

b3m2a1

11:41 PM

Ah the beauty of the incomprehensible FE crash. I love how finnicky it is! :) So quirky! :) So fun! :) I love having to restart what I've been working on for the past few hours because I pressed a button and the FE freaked out :)

Transcript for