Wolfram Mathematica

Pickett

00:01

As to what I'm doing, I'm using the API function in the cloud.

Nasser

@Pickett ok, thanks for the information, very useful.

Nasser

09:55

"If one sees a mathematical or algorithmic 3D graphic in a technical book or article today, there is an overwhelming likelihood that it was produced with Mathematica." From wolfram.com/products/mathematica/analysis/content/… hum... I love M graphics, I think they are amazing, but I wonder if the above statement can be justified by facts and statistics?

halirutan

10:51

@Nasser Hehe.. overwhelming likelihood. I should have written that in the statistics exam: "The probability that A has cancer is overwhelmingly high".

brama

11:07

@halirutan I know Clip is not a compilable function. Any similarly efficient compilable function available? I want to do something like this

a = 1; Vf = 100; cap = 2500; p1 = RandomReal[50, {10}]; p2 = RandomReal[2000, {10}]; demC = Compile[{{a, _Real, 0}, {p1, _Real, 1}, {p2, _Real, 1}}, Clip[(Clip[p1 Vf, {0, a cap}] - p2), {0, a cap}], Parallelization -> True, RuntimeAttributes -> {Listable}, CompilationOptions -> {"InlineExternalDefinitions" -> True}]; demC[a,p1,p2]

Nasser

any one knows when `Plot` will return `GraphicsComplex[]` instead of the normal `Line[]` ? I found it strange it returns `GraphicComplex` here
f[x_] := ((x - 50000)^4.51846) (E^(50000 - x)) (4.6924*^-7);
p = Plot[f[x], {x, 0, 5}];
InputForm@p

help says it will return GraphicsComplex when using Filling option. But there is no Filling option here. Even though the data itself is complex, it should still return Line[] ?

brama

@Nasser No idea :(

Kuba

@Nasser you can always use Normal.

Nasser

11:23

@Kuba use it where?

Kuba

@Nasser on that plot so it won't produce GraphicsComplex :)

Nasser

@Kuba humm... it still does?
f[x_] := ((x - 50000)^4.51846) (E^(50000 - x)) (4.6924*^-7);
p = Normal@Plot[f[x], {x, 0, 5}];
InputForm@p
Graphics[GraphicsComplex[{{1.0204081632653061*^-7, 0.},

Kuba

@Nasser well... that's unexpected...

Nasser

@Kuba should I post this on main forum? I thought I am overlooking something simple

Kuba

@Nasser I'm more interested in why Normal fails than the result of Plot.

Nasser

11:34

@Kuba I did not know even that one can use Normal on plot to do what you suggested it should do. Mathematica can be confusing sometimes.

Kuba

@Nasser I'm really sad I can't trust her :P

ok, have to go now, cya

Nasser

11:51

bye, thanks for the help.

Nasser

14:16

WRI should really fix the large PDF export size. Just exported small demo of 100 lines, but it had 3D plot, the final size of the pdf was 75 MB !!

Searke

The 3DPlot is probably highly detailed and a complete 3D plot in EPS

That is, its an actual 3D plot, which could in theory be rotated and everything if that were possible in PDF etc.

The best thing you can do to reduce the size of the image is to first rasterize it

If there are parts of the the 3D plot that aren't visible (like some part is hidden behind another part), then you may want to modify the plot so that it just doesn't exist in the output.

tl;dr You want to rasterize your 3D graphics before exporting them in a PDF unless there's a good reason to leave them as an expensive 3D vector graphic

Nasser

@Searke I actually end up just using website called compress.smallpdf.com which does amazing job at shrinking pdf file size with no noticeable loss of quality. Drop the pdf on its web page and then download the small pdf.

Some other times, I delete all the Manipulate output, then save as pdf. Makes the pdf much smaller also.

Searke

Yeah. Mathematica PDFs tend to be large either because they include 3D vector graphics or because they include fonts to help ensure correct rendering. These are the two major things I've usually had to work around when making smaller PDFs.

Nasser

14:33

Just wrote this demo this morning, to learn about spinning wheel precession effect. Mathematica graphics makes doing these so easy.

halirutan

15:24

@brama Yours is a good example why I wouldn't use "InlineExternalDefinitions". You define p1 and p2 outside, as external symbols and then you use them as local variables to compile. This is quite disturbing for the reader.

I guess you only want to inline cap and Vf, right?

acl

@Nasser I remember the lecturer for my first year mechanics course demonstrating this with an actual spinning wheel. I was impressed. You can also feel it by holding a spinning hard drive and rotating it about different axes

halirutan

@brama I would suggest something along these lines

clipC = Compile[{{x, _Real, 1}, {bounds, _Real, 1}},
   	Table[Max[Min[i, bounds[[2]]], bounds[[1]]], {i, x}]
   ];

demC = With[{clip = clipC, Vf = 100, cap = 2500},
  Compile[{{a, _Real, 0}, {p1, _Real, 1}, {p2, _Real, 1}},
   clip[(clip[p1 Vf, {0, a cap}] - p2), {0, a cap}],
   Parallelization -> True, RuntimeAttributes -> {Listable},
   CompilationOptions -> {"InlineCompiledFunctions" -> True}]
  ]

Nasser

@acl, yes. I am now writing one on spinning top and gyroscopic motion, which is a little more general case and little more complicated that this one. It is always fun to simulate these things on a computer screen.

halirutan

@brama But note that inlining of compiled, listable functions is not possible. I had to write a clip-function that works on lists.

brama

15:39

@halirutan yes..I mistakenly used the same variable names for p1, p2 and a

brama

15:52

Hmm....custom clip function is a good idea. I was looking for a function corresponding to `Clip` that is compilable. I guess there not one available and writing a custom is the way to go. Interestingly, I found that if I just use the `Clip`in a function definition as in `d[x1_, x2_, x3_] :=
Clip[(Clip[x2 Vf, {0, x1 cap}] - x3), {0, x1 cap}];`, it seems to run about 5 times faster than the compiled listable function formulation as you suggested. I am kind of curious under what circumstances I will reap the benefits of your formulation.

halirutan

@brama What you try in your example is not a situation where I would consider Compile.

Here, I guess the performance of pure Mathematica code is already very fast and you have to perform a deeper analysis to make it any faster.

brama

@halirutan ok...

halirutan

@brama Additionally, think about what a simple and readable code your highlevel d[..] definition gives compared to two separate Compile calls which you have to inject into another.

@brama Can you try the performance of copy&pasting the Min[Max[..]..] definition into demC itself?

@brama As I expected it. If you do the following:

demCHal =
 With[{Vf = 100, cap = 2500},
  Compile[{{a, _Real, 0}, {p1, _Real, 0}, {p2, _Real, 0}},
   Min[Max[(Min[Max[p1 Vf, 0], a cap] - p2), 0], a cap],
   Parallelization -> True, RuntimeAttributes -> {Listable}]
  ]

it is the fastest on my machine.

(@brama note that I changed the signature of the compiled function! It really uses now the listable and the parallelization)

PatoCriollo

16:14

@bobthechemist saw the article in MagPi on the DIY Spectrophotometer. magpi.techjeeper.com/The-MagPi-issue-24-en.pdf . Nice!

brama

@halirutan give me few minutes...I am chewing on it.

@halirutan wow...that's amazing. I am also getting the best performance for demCHal. I wonder why it works if you make the variables as scalars {p1,_Real,0} instead of the arrays {p1,_Real,1} ?

halirutan

16:30

@brama Because this is how Listable is supposed to work! You write a definition which takes a number and the Listable makes that if you put a list in, the function is threaded over the list.

brama

@halirutan cool...Awesome...you made my day!!

halirutan

@brama Glad I could be of help.

brama

@halirutan Big day at the worldcup today. I guess you guys will be through!!

halirutan

@brama I'm not a football guy but I heard Algeria is and was always a very hard enemy. And the worst situation is when you underestimate your enemy. Let's see what the evening brings.

brama

@halirutan I am not that big into it too..but keep track of the scores. what sport do you like better?

halirutan

16:44

@brama Everything I can actively play. I hate watching while I have to sit still. Soccer is quite nice, Volleyball is also good, and I like Squash and Tennis. Those are my favorite team sports.

brama

@halirutan Tennis is my game too.

@halirutan Anyway need to go now...I will have the window open. As always, thank you a ton for making me better at MMA. Have a great day!!

halirutan

@brama You too. Bye

rm -rf

17:29

@brama btw, you don't have to star every chat reply you get :) (please don't)

Öskå

18:22

@rm-rf Are you around?

This edit is uber spam, I thought it was worth being noted. Don't know if you could see/block the editor after rejecting so I post it now.

And since there is no flagging process on edits.. :)

Rojo

18:44

Still no v10?

bobthechemist

18:58

@PatoCriollo Thanks - I think I'm almost up to 6 readers!

brama

19:11

@rm-rf Sorry for the confusion, I only star the chat messages which I think will be useful for me to refer to in future. I thought that that way I can go back and get them by looking through "starred by me". Sorry if this is not the appropriate way to tag messages for my use. I do not do this to appease others, but to help myself. Any suggestions on alternative ways is welcome.

halirutan

@brama Usually, we star only funny chat-messages or really informative stuff. You, for yourself, can get the permanent-link of a specific messages by clicking the very left of a message.

Szabolcs

@brama When you star a message in chat, it will be visible in the list on the right to everyone. Usually we start messages that could be of interest to many people

@brama This is different from starring questions on the main site, which is indeed a way for you to keep track of interesting question only for yourself.

brama

@halirutan and @Szabolcs Got it...Thank you for clarifying.

rm -rf

20:02

@Öskå Thanks for letting me know. This is of the same type as these, and they're sent in anonymously (i.e. no user for me to ban)

@brama Aha! I never thought of using stars like that... that's actually quite handy. Unfortunately, as Szabolcs and halirutan have told you, that isn't how stars are being used currently and might throw a few folks off (the "starred by me" filter is somewhat recent). Sorry for the confusion!

brama

@rm-rf No problem...I learnt something new :)

brama

20:44

can some one comment on defining variables inside a compiled function using With Vs. define as global variables and use InlineExternalDefinitions. I need to write several compilable functions that use few global variables. Performance is very critical. What is better for me? Does it matter?

bobthechemist

21:56

@PatoCriollo Updated some Wolfram/RPi stuff on my website: getting Mathematica to output to an RGB LCD display.

brama

22:56

For my simple compiled functions, the performance did not change, but was wondering if any one has any comments!!

Murta

No Wolfram Desktop version of Wolfram Desktop... strange name no??

acl

23:09

@Murta I think the same is mentioned here

A new kind of naming?

(I'm sorry)

What does the great @Belisarius say (if anything)?

bobthechemist

I'm just glad we are discouraged from doing a product-to-feature comparison, since I have no idea how to do that.

acl

23:25

@bobthechemist Yes, any idea what that is supposed to mean?

bobthechemist

@acl Absolutely no clue.

halirutan

I rarely advertise other posts then my own, but this here is really worth a read:

14

A: What is the relationship between Mathematica and Wolfram Desktop?

We will of course have to wait and see, but it seems pretty clear to me what's going on. Wolfram's technology stack is based on the "Wolfram Language" and the "Wolfram Knowledgebase" (curated data), and although both are very powerful they have not been competitive for commercial application depl...

belisarius

@acl They already have a steep learning curve for their products. Why not do the same with the product names?

Murta

@bobthechemist I fell better now, I thought that it was my ignorance to don't know how to performe "product-to-feature" comparition :p.

halirutan

@Murta @acl You kids are just getting old. I guess the young people nowadays will have no problems with the new naming of products...

acl

23:35

@bobthechemist @Murta maybe it has to do with "Mathematica is a Product" vs "Wolfram Desktop is a Platform"?

belisarius

@halirutan I think you'll need a Doctor Who timeline to understand it all

PatoCriollo

@bobthechemist very nice!

halirutan

@belisarius hehe.. and here is another proof:

bobthechemist

@acl Oh don't get me started: "When deciding which Wolfram product is best for you, you’ll want to make a product-to-product comparison"

acl

@bobthechemist Ah. Oh well.

halirutan

23:38

@halirutan But I need my fxxx Wolfram license, from the Wolfram User Portal right now! I don't need to manage my Wolfram ID.

belisarius

@halirutan I'll switch to hibernation mode until v10 settles down

acl

@bobthechemist Sounds like a tautology if it's put this way.

bobthechemist

Shouldn't it be a product to platform comparison? But wait, comparing Mathematica to Wolfram Desktop is apples to oranges, we have to compare Mathematica to the Cloud, which is a product (suite).

belisarius

@bobthechemist Nah, I just need my computations done. Anywhere, every time and mostly cheap.

bobthechemist

@belisarius True enough

belisarius

23:41

@bobthechemist And I don't care if my computations are running on a "fancy named" platform, of course

Murta

@belisarius I enjoyed the hibernation idea. I'm just worried to wake up and find some "Wolfram Cloud Alpha Desktop Suit for DataScience as Service" and other new strange stuffs created...

bobthechemist

@belisarius I'd prefer they spend less time figuring out how to name/promote their products (or are they services, or platforms) and more time optimizing the Raspberry Pi version.

Clearly, since they made no money on me with the RPi version, it should nonetheless be their top priority.

acl

@bobthechemist they could also fix some bugs

halirutan

@bobthechemist Nope, that's not the way the almighty free marked works.

bobthechemist

wishful thinking, I guess.

acl

23:54

@halirutan marked for free? I'm in!

Transcript for