Wolfram Mathematica

12:26 AM

@KārlisBaumanis I do not know why it does not work for you. Make sure you clear global context first . I am using V9

@NasserM.Abbasi thanks! I worked after I restarted my PC, but there is another problem - results are incorrect

Here is example where results should be {130,140}

x1 = 100
y1 = 100
r1 = 50.00

x2 = 160
y2 = 120
r2 = 36.06

x3 = 70
y3 = 150
r3 = 60.83

eq1 = (x - x1)^2 + (y - y1)^2 == r1^2;
eq2 = (x - x2)^2 + (y - y2)^2 == r2^2;
eq3 = (x - x3)^2 + (y - y3)^2 == r3^2;

c = CoefficientArrays[{eq1, eq2, eq3}, {x, y}];
LeastSquares[c[[2]], c[[1]]] // N
Graphics[{
{Blue, Circle[{x1, y1}, r1]},
{Red, Circle[{x2, y2}, r2]},
Circle[{x3, y3}, r3]},
Axes -> True]

these simultaneous equations should be calculating point where all circles intersect

rm -rf

12:41 AM

Hey @NasserM.Abbasi are you around?

Nasser M. Abbasi

@rm-rf yes i am here

@KārlisBaumanis i'll try to look to see what is going on. I am not very good in linear algebra though

rm -rf

x = randn(3,4,5) + 1i*randn(3,4,5);
y = squeeze(x(2,:,:));
X = squeeze(x(2,:,:))*squeeze(x(2,:,:))';
Y = y*y';
isequal(X,Y)

Aren't the operations in computing X and Y exactly the same? Why does it lead to FP issues? (I get 0 for the equality test)

Nasser M. Abbasi

@rm-rf I get 1, please see

using Matlab 2012a on windwos 7, 64 bit OS

rm -rf

Hmm... You're on windows, right?

Strange...

acl

@rm-rf I also get 0 on OS X (also on octave)

strange

rm -rf

12:48 AM

@NasserM.Abbasi Could you also try this longer example:

clear all
s1   = RandStream('mt19937ar','Seed',1);
s2   = RandStream('mt19937ar','Seed',2);
dims = [40,30,20,10];
data = randn(s1,dims,'single')+1i*randn(s2,dims,'single');
x    = data(:,:,:,2);
y    = squeeze(x(2,:,:));
X    = squeeze(x(2,:,:))*squeeze(x(2,:,:))';
Y    = y*y';
spy(X == Y)
isequal(X,X')
isequal(Y,Y')

I get 0 and 1 respectively

Nasser M. Abbasi

@rm-rf yes, me too

rm -rf

Ok, so something is off... I wrote to MW support and the guy was like "it's all FP, dude". The fact that X is not Hermitian is f'ing driving me nuts

Not that I can't use temporary variables as in Y, but it caught me completely by surprise... not fun having complex eigenvalues when you don't expect it.

Rojo

Helloooo

Kārlis Baumanis

@NasserM.Abbasi if it helps, in the end of this page everything2.com/index.pl?node=triangulate is described how to calculate it on paper

acl

@rojo back up?

Rojo

12:56 AM

Yeah, not a single issue

acl

I haven't yet played much with the stochastic process functionality but it looks fantastic

Rojo

I'm liking it :)

Nice design

Nasser M. Abbasi

@acl you would need to buy a separate toolbox for that if you were using Matlab :)

Rojo

Haven't had much time yet however

acl

@NasserM.Abbasi I didn't know they had one (I have access to all of them at work but don't really use matlab much)

Rojo

1:00 AM

When I use Matlab, I use it through a Matlab style cell

Nasser M. Abbasi

@acl, I did not either, but I meant in a funny way, that one needs to buy a toolbox in Matlab to do someone a little more advanced

Rojo

and now the RInput style has joined the club,

acl

@NasserM.Abbasi No I got that, I just mean, it might have been useful to know :)

rm -rf

@NasserM.Abbasi That business model seems to work pretty well for them :)

Rojo

@rm-rf Schhht

Just in case... We don't want Stephen to hear that

By the way, anyone understands Leonid's answer?mathematica.stackexchange.com/a/15343/109

acl

1:02 AM

From talking to people around here, those who do use matlab for heavyish numerical things (many more than mathematica) prefer it because it's more predictable. I know a few have tried in the past and got discouraged by apparently random slowdowns and bottlenecks.

Rojo

I may be sleepy but I don't find it obvious why he does the unitConvert redefinitions and Quantity blocking

acl

In a couple of the cases I got them to send me their code and found the slowdowns to be related to unpacking. Well, how would they know it since it didn't appear anywhere in the docs?

Rojo

@acl They should come here

rm -rf

@Rojo You can enter 1 m/s, etc. only if you're using W|A and that too, using the cludgy = box. Leonid's solution is a way to make it work without using the W|A interface, but you'll have to define stuff yourself

acl

@Rojo a) I think the point is that to get mma fast for numerics sometimes requires way too much black magic compared to other languages, b) here didn't exist when they were doing this :)

@rm-rf some would argue that the WA thing is beautiful as opposed to a kludge

I disagree, but...

Nasser M. Abbasi

1:05 AM

@acl this packing and unpacking is really confusing. In Matlab there is no such thing. I think a programmer should not have to worry about details like this while writing an algorithm

Rojo

@rm-rf Yeah, but what's the point in blocking Quantity and defining an upvalue to change UnitConvert[arg_, Quantity[, unit] into UnitConvert[arg, unit]

acl

@NasserM.Abbasi I agree, but mathematica is so high level that this seems hard to avoid

Rojo

?

@acl That's true

acl

actually this was in a sense Leonid's point about Julia, that it doesn't have this sort of thing

python, for instance, does

Nasser M. Abbasi

@acl Julia does not even have named parameters (for function calls). Very bad omission from a language that is supposed to have fixed flows in Matlab.

acl

1:10 AM

@NasserM.Abbasi I'm not advertising Julia, much as I like her ;) merely mentioning one positive aspect

Nasser M. Abbasi

Matlab is very simple and powerful when one is working only with matrices/vectors. Anything beyond that, it starts to fall apart and gets messy actually and not well designed. But if I stick to just numbers/vectors/matrices, then it is very nice.

acl

@NasserM.Abbasi that's also the impression I have, but I'm not really that knowledgeable about it

but then mathematica has some of the same characteristics, except that the Compile thing sometimes lets you get away with just mathematica instead of mathematica and C

in my case at least

introducing that in v8 let me more or less stop using C

Nasser M. Abbasi

In MMA, the list is a ragged array, hence same List is used for matrices and for anyting else. In Matlab, one uses normal matrices for non-ragged arrays, and cells for ragged arrays. Then conversion and access between the 2 gets confusing. [] vs. () vs. {} and many functions to work around these things. I still get confused to this day when I use cells in Matlab :) struct arrays are also bit confusing. But all these were added on later to Matlab.

acl

@NasserM.Abbasi I see, I never ran into this (probably because I write mathematica in matlab :) )

soandos

What is the easiest way to turn {{x->12345}} into 12345?

acl

1:22 AM

oh no I did, you're right

Nasser M. Abbasi

@soandos may be
expr = {{x -> 12345}}
x /. First@expr

or First@(x /. expr)

Rojo

I don't feel comfortable with matlab cells either

Furthermore, the indexing in matrices is different, not to say weird

You can index a matrix by a single index, and it doesn't mean the ith row but the ith element couting as if reading a book order

Nasser M. Abbasi

logical indexing is very powerfull in Matlab.

Rojo

When you need 3D matrices, you have to build them in stages

Nasser M. Abbasi

@rojo, Matlab can use linear indexing, or subscript indexing.

Rojo

1:26 AM

because with "," and ";" they ran out of punctuation

Hehe

rm -rf

@Rojo I like that, actually.

Rojo

@rm-rf Really?

soandos

@NasserM.Abbasi is there a faster way to do Level[{{x->12345}},-1][[2]] or thats pretty much it (syntactically)

Rojo

Why do you like having to use a for to build a 2x2x2 matrix?

rm -rf

@Rojo Yes, it makes for pretty nifty coding

@Rojo Who uses for for that when you have cat?

acl

1:28 AM

@rm-rf you'll love C

Rojo

People that don't know much Matlab like me

:)

rm -rf

:)

Rojo

I'm still expecting your blog post on that

soandos

@acl least you don't have to manually do all the indexing

rm -rf

@Rojo Oh... ermm. :D

(this is why you shouldn't commit to things you won't have time for :P)

Nasser M. Abbasi

1:30 AM

@soandos I do not understand. You should use x/.rule type thing to obtain x (do not use Level[] command myself. Keep it simple :)

Rojo

@rm-rf Don't worry, I have my share too

like the undo thing

or the TLetris game that noone cares for but I want to do some dayh

acl

@Rojo you are not alone in that; I know an entire company that doesn't have time to implement undo

4

Rojo

Starrr

@acl Hehe

acl

2 stars in 19 seconds??

Rojo

You got more stars in 19 seconds than votes I got on the Legended styling question in 2 hours

acl

1:33 AM

@Rojo not any more

Rojo

@acl I wasn't begging

but it's true that I was wondering why it was unliked

acl

@Rojo I had neglected to upvote it (I'd read the answer earlier and didn't understand, then worked it out and forgot to upvote)

@Rojo it's cryptic

Rojo

@acl That's good input

How could I improve that

acl

@Rojo more detail? I guess I needed to go to mathematica and play with the options inspector for a few minutes, and couldn't work out how to get it to do what you said in the beginning.

Rojo

:)

Ok, thanks, I'll add more info

acl

1:40 AM

well actually what you say is enough; don't know what else you could add

how did you work that out?

Rojo

Well, they are pretty simple in design

acl

@Rojo no I mean, how did you know Legended?

Rojo

There's a Legended[Graphics[...], legend]

that doesn't evaluate

That's the output of plots with legends

And that formats as some box structure

rm -rf

@Rojo Ok, I see your point. UnitConvert works with a Quantity in the second argument as well. He must've thought it only accepts the unit as the argument

Rojo

that includes a TagBox[graphics, "Legended"]

rm -rf

1:42 AM

However, his code doesn't work the second time it is evaluated... I get

UnitConvert[Quantity[3600,CalculateUnits`UnitCommonSymbols`UnitClique[Calculate‌Units`UnitCommonSymbols`Meters]], Quantity[1, "Kilometers"]]

Rojo

@acl TagBoxes with a second argument string always style as that string

Nasser M. Abbasi

I just download the update V9 from portal area. It still says 9.0.0, I would have expected it to say 9.0.1 or something. I check the .exe size, little different so I assume it is the updated one. It would have been better if WRI have named the file 9.0.1 , please see screen shot

acl

@Rojo OK I see your point. I don't understand frontend stuff well enough to do this

Rojo

@acl Try this

RawBoxes[TagBox["2", "acl"]]

Then go to the stylesheet editor, add "acl" style and increase the font size

rm -rf

@NasserM.Abbasi Probably not a good idea to name it 9.0.1 when the problem is localized to Windows only. It'll leave the mac users wonderign what's changed..

acl

1:44 AM

@Rojo I see!

@NasserM.Abbasi also apparently they changed the installation procedure, not mathematica itself

@Rojo maybe I'll play with this stuff over christmas to detox from physics :)

Rojo

:)

Nasser M. Abbasi

@rm-rf, ok, but the file name should have been named something little different, since it is _WIN in the name allready. So calling it 9.0.1_WIN would be ok

Rojo

@rm-rf Then I'll leave him a comment to see if he had other reasons for removing all the Quantity local definitions in his dynamic environment

Nasser M. Abbasi

I do not like to see same file name, when the file content is different. confusing.

acl

@NasserM.Abbasi better not look into my hard disk then!

rm -rf

1:46 AM

@Rojo The second tag is to handle the case of 0 quantity

acl

@NasserM.Abbasi I have tens of files named analyze.nb

Nasser M. Abbasi

@acl I name all my notebooks A_v1.nb, A_v2.nb, A_v3.nb, etc.. my version control system :)

acl

and notes.tex

Rojo

and Untitled-1.nb

:)

and init.m

acl

@NasserM.Abbasi and disorder_24_correct_2_fast.nb

Nasser M. Abbasi

1:48 AM

@acl but they are in different folders? so ok. Not when the same file is meant to replace an old one in the same place.

rm -rf

@NasserM.Abbasi That is precisely (from their point of view) when you should use the same filename

acl

@NasserM.Abbasi no right, but I think they meant for this to be the definitive file. no point in renaming it

rm -rf

Why would they want you to have the BSOD version lying around?

acl

@rm-rf to prove a font file can kill windows?

hello?

Rojo

Hello acl

How's it going

acl

1:49 AM

(always wanted to say that)

hey!

nice to see you around!

where have you been hiding!

Rojo

Oh, I was right here, but my avatar doesn't draw enough attention

acl

@Rojo not loud enough, yes. make it pink

rm -rf

@Rojo Maybe it needs more red.

Rojo

@rm-rf ok, I just used cat

acl

hm I need 7 votes for 10K. why doesn't someone ask anything I can answer!

(don't vote for my previous answers; let's make this worthwhile!)

Nasser M. Abbasi

1:54 AM

@acl i'll go now upvote you for all your questions. give me few minutes. I myelf do not like reputation count. If there is a way I can give you all credit points I have, you can have them.

Rojo

@acl What's your strong area?

acl

@NasserM.Abbasi ah no, the fun is in seeing people get something out of your answers, don't upvote me for that

@Rojo drinking beer

and yours?

Rojo

I'll ask something about drinking beer with MMA

I don't believe it's too localized, we all drink it

acl

@Rojo go ahead!

rm -rf

Someone's upvoting acl rapidly... probably a good idea to stop, lest you trigger all kinds of scripts and kick him well below 10k

acl

1:59 AM

@rm-rf ha, that'll teach me!

Nasser M. Abbasi

@rm-rf I upvoted acl for few questions., To get him the count he wanted., He is good any way

acl

well 9992 is not a bad place to stop :)

rm -rf

@NasserM.Abbasi Try telling an automated script that he's good =)

Nasser M. Abbasi

@rm-rf he asked for it :)

acl

@NasserM.Abbasi thanks, but don't bother; it's all for fun anyway :)

halirutan

2:00 AM

Quick question: When I announce some a conference and I say who is talking.. is it: "Speaker" or "Speakers"?

acl

ok guys off to bed here. see you

rm -rf

@halirutan If you're referring to one person, then "speaker", otherwise "speakers".

Rojo

@acl See you. Since you go I'm going to self answer after I post :P

halirutan

@rm-rf No, to many. But speaker sounded like information to me.. always plural.

thx

rm -rf

@acl Congrats on 10k :)

Nasser M. Abbasi

2:03 AM

it is much easier to answer Matlab questions than Mathematica questions. Most Matlab questions are easy. Mathematia questions are much harder

people in the Matlab group ask questions on things like how to find a number in a matrix :)

rm -rf

@halirutan "Speaker" is enumerable — 1 speaker, 2 speakers, etc., so it's a "countable noun". On the other hand, "information" is non-enumerable. You can't say 1 information, so it is an "uncountable noun". It doesn't have a singular/plural distinction (although, there are some cases where this rule doesn't quite apply rigidly).

halirutan

But at least it's a rule.. German has information and informations

rm -rf

@halirutan I'll take the German language's rules any day over English's :)

(although, the gender rules can be totally messed up... Never understood why it's die Sonne and der Mond, and millions of such silliness)

halirutan

Because the sun is bright, shiny and beautiful like a girl. The moon, hmm

rm -rf

@halirutan Heh, and surely das Mädchen is lovely and beautiful like a... oh, wait.

halirutan

2:18 AM

Maybe it is too young for you and you rather like to find eine Frau ;-) ;-)

rm -rf

lol

Rojo

2:32 AM

I am definately not good at doing pretty graphics

halirutan

2:47 AM

@Leonid @RolfMertig I made a community promotion ad for the Mathematica day in Berlin next year.

0

A: Community Promotion Ads - 2012

xzczd

5:09 AM

Hello? Er…does anyone know the exact number of built-in functions in mma?

Gabriel

5:20 AM

@xzczd 12000.org/my_notes/compare_mathematica/V9/main.html

xzczd

6:06 AM

@Gabriel Oh, Thank you!

acl

9:55 AM

ha thanks

halirutan

10:18 AM

Because of the comments here mathematica.stackexchange.com/a/15550/187 I tried yesterday to find any sign of David B. Wagner. I dug up a lot of old stuff from internet-archieves but the essence is: He vanished completely somewhere before 2000.

His book is really good (maybe my favorite) and since it is long out of print maybe we can arrange to get this stuff. Some of it is still available in the Wolfram Library.

acl

@halirutan I also tried to find it a few times, but it only turned up a couple of times on amazon and cost a few hundred euro!

halirutan

@acl I have it, but I think it would be worth a try to make it available.

But for this we need to to find David Wagner first.

acl

@halirutan yes...

halirutan

@JohnFultz Do you have any idea what happened to David B. Wagner? He thanked you in his book and it sounds like you helped a lot in digging in the depths of Mathematica. He was quite active on MathGroup and wrote a few times for the Mathematica Journal and then he disappeared..

acl

11:09 AM

well something funky is going on with the licensing; after upgrading my home license, it showed a 1 month month $LicenseExpirationDate and a professional license in the $LicenseType. After I produced the password myself (from the online portal) and re-activated, it now is a Professional license with infinite duration...

I guess I should let them know when they bother answering my enquiry as to the 1 month duration of the first attempt (once they recover from the launch, probably)

Rolf Mertig

11:22 AM

System`Debug is missing. Debug is really nice ... Mathematica 9.0 for Microsoft Windows (64-bit)
Copyright 1988-2012 Wolfram Research, Inc.

In[1]:= ?Debug
System`Debug

Attributes[Debug] = {HoldFirst, Protected}

In[2]:= Debug[a=1;b=2;a+b]
a = 1
action: ?
i: inspect current state
s: step through execution
n: continue to next expression (default)
c: continue to end of block
f: finish computation
a: abort computation

a = 1
action: n
b = 2
action: i

>> >> a

1

>> >> Quit[]

acl

11:43 AM

@RolfMertig good point; here (OS X) it shows "unknown messages", takes 100% of a single core and I have to kill mathematica

Rolf Mertig

@acl: too bad. I think WRI should document it, make it work in the FrontEnd and do something about comments like reddit.com/r/programming/comments/13xxtn/… I have been arguing with a WRI director two years ago that WWB should be integrated into Mathematica. He said plain no ..

acl

@RolfMertig yes, I think integrating with mathematica would help in countering such arguments (on the other hand, mathematica does have a debugger and the profiler works without the wwb, even though it is not documented). but indeed for writing a lot of code, the WB is a much more convenient environment.

Nasser M. Abbasi

@RolfMertig I been saying this for long time about an easy to use build-in debugger. I spend more time debugging my Mathematica code than actually I spend time writing it. In Matlab, I never spend more than a minute to find where the bug is. (debugger, please stop on error and show me the line number) that is all.

Rolf Mertig

@acl Do you actually use and like the WB debugger? I tried, and I do not find it convenient. I can debug faster otherwise.

acl

@RolfMertig sometimes, but it's not as useful as a debugger for eg python or C.

Rolf Mertig

11:55 AM

@NasserM.Abbasi I once had a tough consulting project where, in a 100k lines .m file I had to put hacks like f[x]; DB[1]; g[y]; DB[2]; ... debugging functions riffled programmatically in the code which then helped in finding exactly where the code did get hung up (so DB was some kind of function which logs time, memory usage to a log file ). So I do not want only line-numbers, but kind of pointers to where exactly something goes wrong. Check is not enough.

@NasserM.Abbasi I think the problem is that with the exception of the Wolfram Alpha crowd there are not a lot of people who write longer Mathematica packages at WRI (not sure of course about that). So hard-core programmers are a minority.

Nasser M. Abbasi

@RolfMertig WOW, 100k lines of Matlab file! I never even knew such a thing can exist. My Matlab files are small. Just for school HW's. I do not think I had one file more than 100-200 lines. But the point is, for me, it is much easier to debug Matlab code than Mathematica.

But I read that Alpha is written in Mathematica and it is millions of lines of code.

Rolf Mertig

@NasserM.Abbasi Mathematica. Though there were a lot of redundant Plot's etc.

@NasserM.Abbasi Sure it is possible to write good code in Mathematica. And WWB is also useful. Definitely. It is just much more cumbersome than it should be and if more people would know about Debug and there maybe would be some kind of better IDE ...

Arnoud Buzing

2:22 PM

@acl I am checking on this. Thanks for mentioning this.

John

2:51 PM

Good morning form Mexico. I have a problem with applying several patterns to StringSplit. Lets say I have StringSplit[data,patterns]with pattern=Not[#=="a"]&&Not[#=="b"]&&...&&Not[#=="blabla"]]. How can I define pattern` without writing &&Not[#=="char"] for every single pattern?

Rojo

3:08 PM

@John You mean Except["a"|"b"|"blabla"]?

John Fultz

@John It seems a bit odd to StringSplit on the absence of a pattern. If you're not careful, this could split in places you don't intend. But I would use Except. E.g., StringSplit["abcd", Except["b" | "d"]].

acl

@ArnoudBuzing I've sent a report (which was asking for help because at the time I could only get it to be a 1 month long license; haven't sent any other since). feel free to contact me if you need more info

John Fultz

@halirutan I haven't seen or heard from David Wagner in years. Last I heard, he got a permanent consultant gig, and then he just completely dropped off the map. He hasn't been to any conferences, and I'm not aware of any exchanges with WRI personnel. I just checked, and his last bug report/TS exchange was in the 90s. His old consulting website has been snarfed up by a domain name troll.

P. Fonseca

4:46 PM

Does someone know if M9 added something new that could help me in this case: stackoverflow.com/questions/7351519/…

Eli Lansey

Is anyone else find M9 on Windows really laggy and slow on the Front End? Simple things like searching the documentation center kill it. Simple calculations sometimes also lag hugely.

Rojo

@EliLansey Not so far, working smoothly I think

P. Fonseca

@EliLansey I have problems

Eli Lansey

@Rojo hm, ok. very frustrating. and i'm running a pretty powerful machine

P. Fonseca

Which OS?

Eli Lansey

4:49 PM

@P.Fonseca ok, glad it's not just me. i'm almost tempted to go back to V8, only because it's stable

windows 7

P. Fonseca

XP SP3 here!

Rojo

@EliLansey After the font fix, I didn't have any kind of crashes, or slowness. In fact my v8 was slower I believe

Gabriel

@P.Fonseca keeping it real

@Rojo I have had the same speedup since the fix ... very snappy. Sucks @EliLansey hate than kind of issue

Eli Lansey

x64, with 8GB RAM

@Gabriel I'm running the newer download -- i never got to the buggy font one

Gabriel

@EliLansey do you have discrete graphics? Not sure how this works ... but people say it is auto turned on? I have a gaming gpu in both my machines, not sure if this effects front end or not

P. Fonseca

4:52 PM

Mine crashes if I open and close chapters on the first page of the documentation center. @EliLansey can you try to open and close them really fast for one minute (always without leaving the documentation center first page).

@Gabriel not my choice...

Eli Lansey

@Gabriel not sure. how do I check?

P. Fonseca

@EliLansey I don't have discrete graphics, and it's buggy...

Gabriel

@EliLansey I don't know :-( just noticed in chat before people saying that their discrete card was being turned on automatically in v9 but not in v8 ... I really don't know what it all means sorry :-(

Eli Lansey

@P.Fonseca which pages? it takes forever to load up the documentation center in the first place

Rojo

@P.Fonseca I just did that for half a minute full speed clicks and nothing hapepned

Eli Lansey

4:54 PM

do you mean just keep clicking on "core language"?

P. Fonseca

@EliLansey first page. Just menu->help->documentation center

Alternate between them

Eli Lansey

@P.Fonseca "them" = ?

@P.Fonseca "menu->help->documentation center" lather, rinse, repeat?

P. Fonseca

Mine crashes in 20 to 30 seconds of clicking on the different chapters

Just click quickly to open and close the different chapters on the home page of the documentation center

Eli Lansey

@P.Fonseca as in, have the expanded TOC expand and collapse?

P. Fonseca

Yes.

Eli Lansey

4:57 PM

@P.Fonseca no problems

i've found where it breaks

if I search for a single word

P. Fonseca

Mine crashes systematically with the clicking :-( I'm waiting for a reply)

But I can search for words. Most of the time it doesn't crash.

Eli Lansey

if i search for a single word, it seems to work fine. if i search for two words -- dead

like "plot" entered into the search box is fast

"plot things" -- dead

also, hitting F1 from a nb also kills it

Gabriel

Sort of a similar question. This is a first time major number upgrade (v8 -> v9) and I find a lot of my code doesn't work the same way (ie I get new error messages etc). Is this normal for Mathematica? did anyone have similar issues with going from v7 -> v8? Some of these issues are probably worth fixing (on my side) ... but it is a strange feeling coming from the matlab/R world where I have never had this issue

Rojo

@Gabriel Got an example?

P. Fonseca

@EliLansey Plot things didn't kill it, but it toke 5 seconds to unfreeze, and do nothing. After clicking it again, it responded. I think it might be related, but if it crashes or not, depends on the time it takes to respond. How does it react to the inline Wolfram|Alpha?

Gabriel

5:03 PM

@Rojo I have posted a couple of issues ... NSolve gives new error messages for equations that worked fine in v8 ... NDSolve has strange error checking on variable names that are built-ins. And some numeric code such as optimization etc gives different results. I have been slowly trying to find out the exact issues, but I have a lot of code to check!

Eli Lansey

@P.Fonseca yeah, maybe kill is overstating it. but it takes ~10s to unfreeze, the search doesn't work. i get something like this:

Gabriel

just seems like Wolfram changes the underlying algorithms a lot between releases, which is nice, but makes it hard to predict what will happen between versions

P. Fonseca

Mine returns to a correct HOME after unfreezing (but doesn't search)

Eli Lansey

maybe after a system restart things will be better?

i'll need to do that later, though

P. Fonseca

I've been complaining for some time, but everybody was more worried with the BSOD

Eli Lansey

5:05 PM

have some calcs running

Gabriel

@Rojo I guess my big issue is that if I just do an "EvaluateNotebook" in v8 vs v9 I don't get the same output!

Eli Lansey

well, BSOD is kinda a big deal... :)

Rojo

@Gabriel can you link me to one of your questions?

P. Fonseca

Restarting didn't help me

Eli Lansey

@P.Fonseca bummer

unrelated -- does the "option templates" autofill work for anyone?

Rojo

5:17 PM

@EliLansey What do you mean by "work"?

If I write Plot[x, {x, 0, 1}, Pl

I get the popup with the options, and if I place enter it autocompletes with the added :>

Eli Lansey

@Rojo here: reference.wolfram.com/mathematica/tutorial/… (search for Option Templates) they imply just putting a comma after the arguments. this does not work for me

if i start filling something in, like ", Pl" then autocomplete works

but only for that entry

Rojo

@EliLansey Yes, I get the same behaviour, but it works for Im and other options too

Eli Lansey

@Rojo so, not the "comma ... here's a list of options" thing

Rojo

Right

But it works the same way for custom functions

@EliLansey but pressing Ctrl+k opens the lists of options

Eli Lansey

@Rojo oh, let me try that

yup, that works. ok, that's prlly sufficient for now. i hate running to the documentation center every time I forget what the options are.

P. Fonseca

5:22 PM

the untis are very anoying. I enter 2.0 feets + 3.0 meters, and it returns 1504/127 ft. If the two numbers were machine, why does it return an exact?

Rojo

@P.Fonseca That doesn't happen for me :S

Quantity[2., "Meters"] + Quantity[3., "Yards"] gives Quantity[5.18723, "Yards"]

P. Fonseca

It happens if you pass by Wolfram|Alpha, since it simplifies your number to an integer...

Rojo

Quantity[2., "Feet"] + Quantity[3., "Meters"] gives Quantity[11.8425, "Feet"]

Right

P. Fonseca

It's what happens when we want to use a simplification philosophy to help us on the input of an exact machine...

Still not convinced...

Rojo

@P.Fonseca It's an alternative

I would really get mad if they stopped supporting one for the other

P. Fonseca

5:33 PM

I agree, but wolfram|alpha was already there, and it can be always a complement.

What I want is an input that doesn't become too cryptic for the reader

Arnoud Buzing

@P.Fonseca You could try to select Quantity[...] and convert it to StandardForm?

P. Fonseca

something like esc units esc, that would bring their new list, where we could choose the units we would like to enter (by name, category, etc), that could be complex stuff, but immediately transformed into their pretty alias on the input cell. I can think of many different ways to avoid the current cryptic scenario

@ArnoudBuzing Hi. I know I've been complaining a lot lately, but this happens when we are used to receive a lot.

Arnoud Buzing

That will turn Quantity[...] into a TemplateBox which displays it in a more recognizable formaty.

P. Fonseca

let me give it a try...

Arnoud Buzing

*format

Gabriel

5:37 PM

@Rojo I posted this in chat before and it has been filed as a bug by @ArnoudBuzing, but just an example of the type of issues I am running into:

eq = {0.` + 1.` (1 - 0.5` R) R - (1.4058533129758728` C R)/(
1 + 0.05623413251903491` R) ==
0, -0.618575457709384` C - (0.7905694150420949` P R)/(
1.` + 0.0316227766016838` C) + (1.4058533129758728` C R)/(
1 + 0.05623413251903491` R) ==
0, -0.3478505426185218` P + (0.` + 0.7905694150420949` C P)/(
1.` + 0.0316227766016838` C) == 0};
NSolve[eq, {R, C, P}]

Rojo

@Gabriel Oh, I think I remember now

Gabriel

also here is an issue with using C in NDSolve that is new
http://mathematica.stackexchange.com/questions/15672/ndsolve-and-c-as-a-variable-name

Rojo

@P.Fonseca It's also quite simple to create a palette that does that for all Quantity's in a notebook

Gabriel

I have many more similar issues ... but they come from large notebooks with harder models to pair down to a clean example. I just wonder if this type of output/answer changes is normal for large Mathematica upgrades ... Wonder if I should be hard coding in the Method options etc to heop mitigate this, or is it just a case of v9 being such a major release?

Guess I am used to a world where "bugs" are kept in code for backwards compatibility ... not that I like that solution ... just new to what the upgrade path of Mathematica looks like

P. Fonseca

@ArnoudBuzing It works fine, but I need to memorize them all, or call wolfram|alpha. Since Wolfram|Alpha call is freezing, I'm still not convinced. Also, during a meetings I don't have internet access (no wireless here), which means that a fast list of entry would be nice. I need to think on my esc units esc implementation of a list

@ArnoudBuzing have you tried reproducing my crash test? I can crash it in 20 seconds, on 90% of the sessions, just by clicking fast in the TOC expand and collapse. It crashes for other reasons, but I somehow think they are all related (crash, meaning, Mathematica closes)

Arnoud Buzing

5:47 PM

@P.Fonseca Yes, I did try this, but was not able to reproduce it. Would you be willing to show it to me?

P. Fonseca

Sure. Tell me how. I can also send you what windows regiters in its bug report

Arnoud Buzing

If you send me email (arnoudb@wolfram.com), I can send you details.

P. Fonseca

Just did.

Just to make sure that it still crashes, I just opened mathematica, went straight to the documentation center, and opened the first theme. It crashed on the first click

I have the windows message opened. Should I do something with it?

Eli Lansey

mathematica.stackexchange.com/questions/15709/… Ditto

Oleksandr R.

6:32 PM

@EliLansey I get pretty much the same thing with the Documentation Center. Anything you search for that isn't a direct page lookup (i.e. that returns the search results page) freezes it for about 10 seconds. Weirdest thing about it is that I installed it using an administrative account but of course I actually run it under my normal restricted user account (I'm on Windows 2003, so no UAC). In the admin account it works perfectly. It's just in the other that it messes up.

Also, it works perfectly fine on Windows 7 too. (Installed using UAC.)

Eli Lansey

@OleksandrR. i have admin privilgles and installed it under my username. dont have access to admin account.

@OleksandrR. im running windows 7 and have the problem

Oleksandr R.

@EliLansey I dunno. It's weird.

Eli Lansey

@OleksandrR. and very frustrating. finding the new version basically impossible to use

Oleksandr R.

Same symptoms but obviously different cause. When I get time I'll debug it and let you know what I find.

Eli Lansey

you running a 32 or 64 bit OS?

Oleksandr R.

6:34 PM

64-bit, both 2003 and 7.

Eli Lansey

k

also running 64-bit

anyone else here have this issue on 32-bit systems?

Oleksandr R.

I was also very surprised to see that they chose Java 6 for M9. Why not Java 7? It's much faster.

Gabriel

@OleksandrR. yeah can you tell it to use the java 7 runtime or is this some how hard coded?

Oleksandr R.

@Gabriel just rename its built-in Java directories if you want. But the supplied version is somehow special--it contains javac as well as the JRE. Thus if you do this you will not be able to use Leonid's Java Reloader without installing the JDK.

@EliLansey when it freezes do you get the internal self-test error Kernels:something:something?

(Cannot tell you what the somethings are at the moment as I am happily on Win7 for now.)

Eli Lansey

@OleksandrR. sometimes.

Kernels-c-6568

is one

Oleksandr R.

6:48 PM

@EliLansey yep, that's the one (I think)

Eli Lansey

i get this on win7, fwiw

Oleksandr R.

One of the kernels is lagging for some reason. I bet it was a network access or an access denied that timed out. Will have to fire up procmon later on and figure it out, but no time now!

Eli Lansey

@OleksandrR. do you think the V9 documentation center uses W|A or other internet tools when searching?

Oleksandr R.

(Unless you want to do it: technet.microsoft.com/en-gb/sysinternals/bb896645.aspx)

@EliLansey I suppose, like the previous versions, it sends your query to the Wolfram servers to get back the "n matches on all Wolfram sites" text. To be sure I disabled internet access (which disables this), and it didn't change a thing.

Also, I can run M9 in the admin account and the user account simultaneously and the first one works but the second doesn't. It's definitely something to do with the account.

Eli Lansey

7:03 PM

@OleksandrR. strange!

acl

7:19 PM

@EliLansey it connects to search.wolfram.com

(but if you don't let it connect, it doesn't hang or anything)

Eli Lansey

@acl what if there's a delay in connecting? for example, if i open a new browser (ie incognito window in chrome) it takes a while till a request to navigate to a site goes through

acl

@EliLansey I don't know; certainly if I block it, I can't perceive any lag.

Eli Lansey

@acl ok

interesting. blocking internet access resolves this issue

the documentation center is zippy

definitely an internet issue

acl

@EliLansey must be platform-dependent at least; here it seems to make no difference

Eli Lansey

unchecking "automatically check for doc center updates" seems to fix it

Oleksandr R.

7:35 PM

@EliLansey weird! Okay, I will try that on mine too!

Although, since I already had internet access disabled, it may not be the same cause?

Eli Lansey

nope, scratch that. on MMA restart, it's laggy again.

very strange

but with internet completely disabled it's zippy

Arnoud Buzing

@EliLansey Do you have any special firewall or proxy settings?

Eli Lansey

@ArnoudBuzing probably. i'm at work. if i open a new browser (ie incognito window in chrome) it takes a while till a request to navigate to a site goes through (resolving proxy, etc)

but doing an F1 help search freezes the kernel

Oleksandr R.

Got to go. Bye all.

Arnoud Buzing

@Eli

Eli Lansey

7:41 PM

@ArnoudBuzing i have also gotten internal self test error: Kernels|c|6568

Arnoud Buzing

@EliLansey Do you know your proxy setting? Under Edit -> Preferences, then Internet Connectivity, then Proxy Settings, you may be able to explicitly set the proxy.

Eli Lansey

@ArnoudBuzing no, i dont. and, this is a laptop, so the settings will change depending on where i'm located

P. Fonseca

@ArnoudBuzing sorry for the delay. I'm available if you still can...

Arnoud Buzing

@P.Fonseca OK, I will call you

halirutan

8:02 PM

@JohnFultz This is really unfortunate. You were my biggest hope.

Szabolcs

9:18 PM

@Arnoud I remember someone was asking if the Featured Examples can be downloaded as a notebook (it seems they cannot). This is one example where it would really be preferable if a notebook download were available. Copying a (large) 3D image as plain text from a webpage is not too convenient and freezes my browser for a while.

Oh, I love these so much! When I was little, it was not uncommon to see them in my place, but now they're very rare because of all the pollution ...

Hmmm ... the SE Uploader doesn't seem to work with 3D Images. Need to find out what's going on.

Or at least not with BIG ones.

I suspect a preemptive evaluation timeout.

rm -rf

9:47 PM

@Szabolcs Do you know how to set up different licenses for different versions?

v8 and v9 on mine seem to pick up the first entry in my mathpass

Transcript for