Wolfram Mathematica

Mike Honeychurch

00:25

hmmm ...however I don't think I have EVER got an email about a new version release

Oleksandr R.

@LLlAMnYP it's a way to write a memoization

Not sure if it has a specific name

kirma

05:02

@MikeHoneychurch I have gotten an email occassionally. I'm not certain if there's logic behind it, or not.

Guess who it is.

06:36

chat.stackexchange.com/rooms/27413

8

Finally went for it. *crosses fingers*

LLlAMnYP

07:58

@OleksandrR. so after some staring at FullForm it seems, e there is simply shorthand for happyQ[i], however I was surprised to find that a SetDelayed to Pattern[e, happyQ[i_]] gives DownValues to happyQ and not to Pattern. Am I missing something here?

Patrick Stevens

09:48

@LLlAMnYP consider the output of `r_mod := ReleaseHold[5 (Hold[r] /. mod -> List)]`
then `mod[3]`

LLlAMnYP

10:18

@PatrickStevens yes, I kind of get it, that is : is useful for writing out more complicated patterns and such, I'm rather looking for reliable reading material on how and when DownValues are set. If I write out your example in FullForm, i.e. Pattern[r,mod]:=definition, mod will get a DownValue. If I write pattern[r,mod]:=definition, pattern will get a DownValue. Obviously, there are some special rules for the head Pattern, but I can't, at first glance, find a good reference.

LLlAMnYP

10:28

Perhaps it's worth creating a question for this purpose?

Of course I meant to say Blank[mod] rather than just mod

Patrick Stevens

Yeah, make a question - I'm not totally sure how it works

LLlAMnYP

11:37

@PatrickStevens done

Szabolcs

14:00

Downsample is unusable slow with large high-dimensional arrays. I can do the very same thing manually with Part and Span, and it's fast on a packed array. What is then the point of Downsample?

Guess who it is.

14:18

I am thinking at this point that it might be useful to list down built-in functions that are easily beaten by user-made ones…

3

Searke

14:54

@Szabolcs Downsample is not for you

Or really for anyone working with high-dimensional arrays

The point of downsample is that downsampling is a common task that people want to do, but don't know how to do at all.

Downsample and many other functions are more or less syntatic sugar which help people find something corresponding to what they are looking for in the documentation.

Anyone with reasonable programming experience wouldn't search the documentation for "downsample" if they wanted to downsample a set of data.

But a sufficient number of inexperience programmers do this to warrant a function which does the basics of that.

That said, there's no reason for it to be inefficient and I'm sure the developer would be glad to see what example you got.

Guess who it is.

15:22

In the case of Permanent, I was specifically asked (told) to implement it but not make a huge project out of it. The general issue of whether/when to put in a new function with possibly suboptimal performance does get raised at times. Suffice it to say that there are considerations, such as getting new functionality up and running in finite time, that sometimes lead to such judgement calls. — Daniel Lichtblau Aug 17 at 6:53

I've been mulling over Daniel's comment for quite a while. I'm not sure how to react.

At the very least, there indeed are functions that were implemented just because it was felt they had to be there, suboptimality be damned.

Meta

16:04

1

Q: What is mathematica?

I thought it is a site for Mathematics, but just now came through another site. What is the difference by the way? I am new here. Thanks

discussion

Szabolcs

@Searke I thought the point of adding so many builtins was to provide shortcuts and to provide more efficient (or general) implementations. I didn't search for Downsample, I just noticed it somewhere a while ago, possibly in the list of new functions. Recently I tried to use new builtins more, assuming that they provide benefits.

Patrick Stevens

I have used Downsample myself extensively on a recent project. Nice piece of syntactic sugar.

Makes it very clear to read.

Szabolcs

Yes.

Guess who it is.

Exactly; the code is easier to parse if the "syntactic sugar" is well-designed.

Szabolcs

But isn't Downsample[arr, {k1, k2, ...}] the same as Part[arr, ;; ;; k1, ;; ;; k2, ...]? Am I missing something?

Patrick Stevens

16:14

It is, but Part is such a general thing that it's not easy to tell immediately at a glance what it's doing

Downsample is instantly obvious

Guess who it is.

But if what's going under the hood is not too good, then what's the point? The user is forced to go back to his stone age tools.

Szabolcs

I just looked at the Downsample implementation to try to figure out why it's so slow and it seems to be quite complicated, relying on a 1D downsampling function Signal`FilteringDump`DownsampleIPP, which only seems to work for machine reals. Why is this necessary instead of implementing it in terms of Part?

This is unusably slow: arr = ConstantArray[0., {100, 100, 100, 3}]; Downsample[arr, {3, 3, 3, 1}];.

arr[[;; ;; 3, ;; ;; 3, ;; ;; 3]]; is instantaneous.

Patrick Stevens

bizarre

but the principle of "use specifically applicable tools rather than generally-applicable ones whenever possible" is a good one for code readability

and, for that matter, for mathematical readability

Guess who it is.

…and yet, if you're reading the code a few months later, you'd need more thought to read the Part[]/Span[] version.

Szabolcs

Yes, Downsample is easier to read and write.

Guess who it is.

16:19

But due to these inefficiencies, we go against one reason why Mathematica is fun to write in: it makes for clean readable code.

Szabolcs

Also, Downsample unpacks. This has to be a bug, I see no other explanation. @Searke

Guess who it is.

"Downsample unpacks" - definitely an oversight by the concerned developer…

Szabolcs

Maybe unpacking is done by some of the code that checks the arguments.

Patrick Stevens

It didn't unpack for me on this one: Downsample[ConstantArray[1, {5}], 2]

how are you checking for unpacking?

Szabolcs

On["Packing"]

It doesn't unpack on that one.

But it does on 2D arrays.

Try {5,5}.

Searke

16:25

@Szabolcs Undesirable at very least

But yes the mindset for some of these new things is to match customer behavior

They can always be sped up later

but at least having them out helps those people who wouldn't know how to do it themselves

That said, unless someone writes [email protected] noting that this, the functions probably won't improve

While there's a quorum of people here

has anyone use GrammarApply, GrammarRules for anything?

Patrick Stevens

the unpacking is in Signaling`Resample`oDownsample

it Maps downsample1d over the input data

I shudder slightly to read it:

Do[xt = Map[
Signal`Resampling`downsample1d[#1, f[[rank - i + 1]],
p[[rank - i + 1]], realQ, prec] &, xt, {rank - 1}];
xt = Developer`ToPackedArray[xt];

not that i know anything about working with packed arrays in practice

but it really seems to defeat the point if you unpack and then re-pack :P

Guess who it is.

If a user suddenly notes an increase in memory consumption while using Downsample[], then that'd be it, yes.

barznjy

17:00

HI

which function is faster; If or Do ?

Patrick Stevens

@barznjy It's not a meaningful distinction, really

You may ask Do to do hundreds of thousands of things

If is restricted to doing two

Like asking "which tool is better at making houses: hammer or chisel?"

Guess who it is.

"doing two" - or three

Why again are you comparing a conditional to a looping construct?

barznjy

@PatrickStevens let me rephrase my question; for the same set of calculation, which one is more efficient ?

Guess who it is.

What exactly are you doing that has you asking such a question?

barznjy

17:17

@Guesswhoitis. I am now using "If" in my code, but it is very slow, it take long time to give me a numerical result. I wonder if "Do" function can give better performance

Guess who it is.

It is very unlikely that the conditional is your bottleneck. What are you doing inside the If[]?

barznjy

@Guesswhoitis. I use it as a loop

Guess who it is.

(That you are even asking this tells me you are confused with how loops and conditionals work.)

If[] does not work as a loop.

barznjy

@Guesswhoitis. I think I need some rest. I am really confused ..

Guess who it is.

If[condition, p, q, other] will return either of p, q, or other depending on what condition evaluates to. That's all there is to it.

Wolfram Blog

18:18

Wolfram Blog: Releasing the OPCClassic Library

posted on August 26, 2015 by Patrik Ekenberg

Wouldn’t it be great if you could easily connect your simulation models to your existing infrastructure? Whether you are working in industries such as oil and gas, industrial energy, or life sciences, connecting to your processes in order to monitor and control them is vital. The OPC (Object Linking and Embedding for Process Control) standard has [...]

halirutan

21:10

Someone around who would like to test this

Manipulate[
 BoxWhiskerChart[{0.5, max, 1.0, 1.5, 2.5}, {{"Outliers"}}],
 {{max, 9.5}, 0, 10}
 ]

and tell my why the whisker goes to the inside when the outlier is drawn it?

Michael Hale

21:43

@halirutan The "box" part is calculated including the outliers. The "whiskers" ignore the outliers. So with a small number of data points and an outlier, it is easy for the 3rd quartile to be between the outlier and the rest of the data set.

m_goldberg

22:27

@Guesswhoitis. or anyone, please help me reopen this question -- I have an answer I want to post.

belisarius

22:41

@m_goldberg $10

@Guesswhoitis. Perhaps you could help here

1

Q: Range of summation in simple Plot seems off

I was trying to reproduce a picture in a book by Havil of the sum, $$s = \sum_{r=1}^{\infty}\frac{\mu(r)}{r}\left(Li(x^{\rho_k})+Li(x^{\rho_k*})\right) $$ using s = Sum[(MoebiusMu[i]/i)*(LogIntegral[x^(ZetaZero[1]/i)] + LogIntegral[x^(Conjugate[ZetaZero[1]]/i)]), {i, 1, 30}]; The gra...

plotting special-functions summation

m_goldberg

@belisarius. You take MasterCard?

belisarius

@m_goldberg Yup. Just mail me your card

m_goldberg

@belisarius. I thought you were in Argentina, not Nigeria

belisarius

@m_goldberg As a bona fide showoff, here's mine

ups! It's my old one

@m_goldberg Scam is a global sport now. Ashley Madison, Shirley Temple and Barbra Sterisand are all into it.

m_goldberg

@belisarius I thought Shirley had passed away.

belisarius

22:51

@m_goldberg Perhaps that German doctor forgot my pills

m_goldberg

23:02

@halirutan. Tried your code and some variations on it. Don't know what's going on, but it looks a bug to me. Or maybe the upper whisker is afraid of the outlier.

belisarius

23:17

@halirutan Manipulate[
{q1, q3} = Quantile[{0.5, max, 1.0, 1.5, 2.5}, {0.25, 0.75}];
iqr = q3 - q1;
BoxWhiskerChart[{0.5, max, 1.0, 1.5,
2.5}, {"Outliers", {"Outliers", "\[FilledCircle]"}, {"FarOutliers",
"\[EmptyCircle]"}}, AspectRatio -> 1/10, ImageSize -> 500,
BarOrigin -> Left,
GridLines -> {{{q3 + 1.5 iqr, Dashed}, {q3 + 3 iqr, Dotted}}, None},
FrameTicks -> {{None,
None}, {{0.5, max, 1.0, 1.5,
2.5}, {{q1, "q1"}, {q3, "q3"}, {q3 + 1.5 iqr,
"near"}, {q3 + 3 iqr, "far"}}}}]
, {{max, 9.5}, 0, 10}]

belisarius

23:32

It's a quartile mess

Transcript for