Wolfram Mathematica

rm -rf

12:10 AM

bah... the madness won't stop

acl

@rm-rf what is this asking? how does one "protect the solution for finding the correct construction of regular enneagon"?

(whuber did deal with that question rather comprehensively!)

but, what's with the influx of this sort of question? do I just notice them more?

rm -rf

One of the few rare cases of answers slipping through post closure:

0

A: Construction of regular nonagon

I'm the first! p[n_] := Graphics@Line@Table[{Cos[2 Pi k/n], Sin[2 Pi k/n]}, {k, 0, n}] Sorry, I've forgotten to bring my ruler today!

acl

I'll reward him with multiple downvotes!

popery getting to his head, it seems

belisarius

@rm-rf hehe

@acl C'mon

What kind of behavior is that!

acl

12:26 AM

@belisarius that's what I was thinking yes

rm -rf

@belisarius You should start wearing the pope hat

I even made it for you!

halirutan

Good morning/evening.

acl

@halirutan morning

s0rce

good evening

halirutan

@rm-rf One thing I wanted to ask you for quite some time: Where is JM? I do miss him really much.

@acl @s0rce Hi.

acl

12:29 AM

random aside: I was giving a talk a couple of days ago and realized, just now, that one of the people speaking before me is a regular contributor here... I just now connected the name to the handle he uses here! didn't get a chance to talk to him

s0rce

I'm going to be at the american physical society meeting next monday giving a talk in baltimore if anyone else is there!

rm -rf

Jan 16 at 17:27, by J. M.

@belisarius It's only partly work that kept me away from computer use the past few weeks. More of "sticky personal affairs", to put it delicately... :)

Your guess is as good as mine :)

acl

@s0rce I know a couple of people will be (I'm not going)

halirutan

@rm-rf Personal affairs is better than something worse like illness. I'm just glad he is OK.

acl

@s0rce what is it on? your work I mean

halirutan

12:32 AM

@acl Someone from Germany? Europe?

acl

@halirutan yes, dpg-physik.de/veranstaltungen/tagungen/index.html

s0rce

The formation of disordered iron oxides in mollusk teeth, I'm not really a physicist; my supervisor knows the chair of the session and asked me to go.

acl

@s0rce wow sounds complicated...

s0rce

mostly presenting X-ray spectroscopy results

halirutan

@acl Oh. Then maybe I know him too.

s0rce

12:33 AM

meetings.aps.org/Meeting/MAR13/Event/183763

thats me

but that abstract doesn't really describe very well what I'm talking about since the results were a bit unexpected

acl

@halirutan possibly, he's in regensburg

belisarius

@rm-rf Be My religion forbids me to take any part in any religion

acl

@s0rce hm, I don't know half the words, so even if it was accurate...

s0rce

perfect :)

belisarius

acl

12:37 AM

@belisarius the iniochos of delphi? what does he have to do with anything?

or did I misidentify it?

belisarius

@s0rce Why the chitons want magnetite in their teeth? Do they eat iron?

@acl That costume is a chiton

acl

@belisarius ohhhhh!

belisarius

@s0rce Understanding the biological stabilization ... 4:18 PM–4:30 PM you better don't blink while talking or you'll consume all your time

acl

@belisarius it's fun when people hurl the contents of their entire PhD thesis on you in 12 min, then ask "any questions?"

rm -rf

"What's your favourite flavour of ice cream?"

belisarius

12:45 AM

@rm-rf Is there a wrong answer?

acl

@rm-rf best I heard in a talk (by a distinguished senior guy, who gave a long talk to about 50 others): "why is your fly open?"

guess it wouldn't work that well in a 500 person auditorium though (at least, not for the speaker)

in the event he laughed it off

rm -rf

@belisarius Yes. "Ketchup and Wasabi"

belisarius

@acl He! I saw it! And then someone asks a completely unrelated question just to get the usual "I don't know" answer

@rm-rf Sooo my "wasabish french fries" are ok

s0rce

yes i dont have very long to convince them of all my results

belisarius

Magnolia Ice Cream

Magnolia Dairy Ice Cream is a Philippine brand of ice cream sold domestically in the Philippines and in other markets such as Thailand, the United States, Australia, Malaysia, and Singapore. Magnolia was founded in 1925 and the company's tropical ice cream flavours include Halo-halo (a mix of fruits and beans), Buko Pandán (young coconut with pandan), Maís-Queso (corn and cheese), avocado, Nangkasúy (jackfruit with cashew), Ube and Buko Salad (young coconut and fruit), and Macapuno (coconut sport). The brand also offers mango and other flavours like chocolate, strawberry, and vanilla. The...

rm -rf

12:52 AM

"jackfruit with cashew" that would be a hell of a flavour. I LOVE jackfruit

Now where can I find some Nangkasúy

belisarius

@s0rce You only need to seduce them. Forget about convincing them in 12 minutes

acl

@s0rce just report them, convincing is too hard in 12 minutes. better for people to get a picture of what you're doing (so they can come and find you and ask if interested), than get all the details in

othewise people will get lost

s0rce

@belisarius @acl, yup, going to try and show the exciting stuff

belisarius

@s0rce Set 2000 chitons free in the place and you'll impress them forever

s0rce

especially if they are the creepy foot long kind

rm -rf

12:57 AM

Most conference talks have a standard formula: "Soo... as you know, 1+1 = 2 and 2+3 =5 and... oops, I'm running out of time. To summarize, we get these complex results and fancy curves"

s0rce

I have lots of fancy curves

!Mathematica graphics

belisarius

@rm-rf I attended once to a talk given by a very experienced speaker. Between the two phases he showed up two naked girls. I'll never forget that talk

sorry, I mean two pictures

not the actual girls

s0rce

maybe I'll try that strategy

rm -rf

@s0rce Hey, someone's putting v9's plot legends to good use already!

It only took them 9 versions to get it right

s0rce

yup!

rm -rf

1:05 AM

@belisarius You should go to playboy conferences then, if you want the actual girls (and real Lena topping)

s0rce

I wish the style of the line in the legend followed PlotStyle

acl

@rm-rf I used them in my most recent talk!

best talk I can think of is one where the speaker played Strauss' "Blue Danube" while showing a bunch of swimmers rotating in a vortex-like structure. something went wrong and he couldn't switch the music off at the end, and he used up half his time (6 minutes) fiddling with that...

belisarius

1:25 AM

@acl I went once to a talk about the thermodynamics of rubber-like compounds. Lots of pictures of the experiments .... all of them made using condoms

s0rce

I posted my line legend issue: mathematica.stackexchange.com/questions/21370/…

halirutan

3:11 AM

This title is so funny. Could have been from a horror movie: Problem with the Dead: The Dead are not dead!

0

Q: problem with show : show doesn't show?

I try to compine three list plot with show command. Here is the first one: asteroidCount = BinCounts[ Sort@Cases[(AstronomicalData[#, "SemimajorAxis"]/149597870691) & /@ Join[AstronomicalData["InnerMainBeltAsteroid"], AstronomicalData["MainBeltAsteroid"], Astrono...

plotting

rcollyer

@halirutan Kill It Before Dies

halirutan

@rcollyer Yes, that fits

rcollyer

@halirutan an okay question, I could see where the OP was confused.

warped title, though.

halirutan

@rcollyer I was indeed not rating the question which seems better than most these days.

rcollyer

@halirutan i can work with honest confusion.

belisarius

8:24 AM

anyone around?

belisarius

8:53 AM

Ok. anyway:

I can't understand the second OP's comment in my answer here:

0

A: Troubleshooting an NDSolve::ndfdmc error message

Replace the second part for: Py[ξ_] := D[y[ξ] /. j1[[1]], ξ] Pz[ξ_] := D[z[ξ] /. j1[[1]], ξ] P1[ξ_] := Py[ξ]^2 + Pz[ξ]^2 A0 = 7; δ = 1/15; r = 1; φ = 5; κ = 1/Sqrt[2]; α1 = 0; α2 = 0; Δ = 1; yo = 0; zo = 0; Pyo = 0; Pzo = 0; j2 = NDSolve[{x'[ξ] == P1[ξ], x[0] == 0}, x[ξ], {ξ, 0, 1.6*2 π/δ}] Pl...

he says " I appreciate your help, but could you please tell me what this symbolize, so that for future works, I should be vigilant. – Vikram Sagar 46 mins ago"

I'm just not sure what he wants

Can anyone help him?

Öskå

1:05 PM

Anyone around? I'm having a weird problem

Oleksandr R.

@Öskå hello

Öskå

@OleksandrR. Hey

As I don't know if it worthes being asked on the site I rather like asking on here.

can I drop a sample of code and someone test it?

Oleksandr R.

@Öskå yes, but pasting into the chat is not so easy... you might prefer to use pastebin

Öskå

Wait wait wait, FindRoots is confusing me.

Oh well, I'm confused.., I never get the same results after changing the Range in the FindRoot

Michael E2

1:21 PM

@Öskå Hi Öskå. Usually that's because of the function and the way Newton's method works.

Öskå

Any best way to Calculate roots of a function like f[x_] := 2 x^5 (-1 + Cos[2 *x] Cosh[2 *x]) ?

Michael E2

@Öskå Does something like Last@Reap@NDSolve[{y'[x] == f'[x], y[0] == 0, WhenEvent[y[x] == 0, Sow[x]]}, y[x], {x, 0, 10}] work for you?

Öskå

Nope, I'm on Mma 8

Michael E2

@Öskå Try Last@Reap@NDSolve[{y'[x] == f'[x], y[0] == 0}, y[x], {x, 0, 10}, Method -> {"EventLocator", "Event" -> y[x] == 0, "EventAction" :> Sow[x]}]

Öskå

1:36 PM

It works.., I don't know how it works but it does! Thanks

Michael E2

You're welcome. I suppose it can miss very close roots, but that shouldn't happen for your function, unless you give it an extremely large domain for x.

Öskå

Althought try f[%[[1,3]]]

or worse, f[%[[1,5]]]

Michael E2

Oh boy...

Öskå

Indeed :) That's so confusing

The precision doesn't match but I assume that if you change on digite in the ~10^-6 you'll get a closer result.

Michael E2

Not really: calculate the derivative at those points -- they're immense. The error in y is going to be about f'[x] times the error in x.

Öskå

1:46 PM

Oh well, I assume that they are still "not too bad" roots, enough to calculate eigen frequencies

Michael E2

You can post process each root from NDSolve with FindRoot[f[x], {x, x0}] to improve accuracy. Newton's method will now converge starting at these near-roots, at least for this function.

Öskå

Smart, it works pretty well: -0.00263784

Michael E2

You could even replace Sow[x] with Sow[t /. FindRoot[f[t], {t, x}]] and do it in one command. But sometimes it's nice to separate the steps. Both NDSolve and FindRoot can generate a lot of warnings about convergence, and you might want to know what exactly is causing the trouble.

Öskå

Indeed, working perfectly again

Michael E2

Great. Have a nice day.

Öskå

1:54 PM

Thanks a lot, have a good day too

Szabolcs

2:29 PM

Is there anyone here familiar with FindClusters and clustering methods? Is there a method available that will work on points in N-dimensional space and will always put two points in the same cluster if they're closer than a given (Euclidean) distance?

I would compute the distance matrix and use ConnectedComponents (which works on graphs), but I was hoping there's a setting for FindClusters that does the same thing and is hopefully more efficient

Lou

3:01 PM

Hi, I found that with MM9.0.0 the communitygraphplot renders different (it doesn't take over the ImageSize values of the graph object) then with MM9.0.1. Is that worth sharing as a question? Or is it not needed since most people likely already use 9.0.1

Oleksandr R.

@Szabolcs isn't absolute distance between two points an insufficient criterion for cluster membership?

rm -rf

@Lou If the question is just that it renders "different", then that's probably not worth asking (but depends on "how different"). There were a lot of bugs in 9.0.0 many of which were fixed in 9.0.1, so I don't know if what you're seeing is a result of that

Lou

@rm-rf haha yes different is very exact. When you create a graph with many nodes but want some nice size shown on a print (or in the notebook) I typically use ImageSize-> Scaled[x]. Then I export it to PDF. But when I use CommunitiesGraphPlot the imagesize doesn't work. It keep the plotsize to small. With 9.0.1 the communitiesgraphplot takes the size asked for in the Graph object.

rm -rf

@Lou Ok, I've never used that function and I don't have 9.0.0 to test it. If you think it's a useful distinction or if it is helpful to know that it is fixed in 9.0.1, then by all means, ask and self-answer your own question :)

Lou

@rm-rf Ok thanks I'll post it. Have a nice day.

rm -rf

3:15 PM

You too

Szabolcs

3:48 PM

@OleksandrR. maybe I should call it connected components, not clustering.

Do any of you have experience debugging memory leaks?

In C or C++, not in Mathematica. It's not my code I need to debug, so I'm a little daunted

Szabolcs

5:20 PM

If any of you are working with CUDA on OS X, don't update your OS to 10.8.3 just yet! --> forums.geforce.com/default/topic/534924/…

s0rce

5:33 PM

Anyone have a circular hough transform implementation in MMA?

rm -rf

5:53 PM

@Szabolcs Why not? The page you linked to also mentions an NVIDIA update for 10.8.3 here nvidia.com/object/macosx-cuda-5.0.45-driver.html

Szabolcs

@rm-rf I misunderstood. I thought that was still the old version.

Gustavo Bandeira

6:09 PM

0

Q: Is it adequate to ask for hints about how to build projects on this site?

Sometimes I have a project in mind, I know what it's going to do but I'm not really sure on how to build it yet. Is it adequate to ask for tips about ways for this project to be done?

discussion support

Mathematica StackExchange (meta)

6:26 PM

0

Q: Is it adequate to ask for hints about how to build projects on this site?

Sometimes I have a project in mind, I know what it's going to do but I'm not really sure on how to build it yet. Is it adequate to ask for tips about ways for this project to be done?

discussion support

s0rce

6:52 PM

@Szabolcs you are the best!!! see: mathematica.stackexchange.com/questions/21412/…

clr66

7:56 PM

Does anyone know a good place to go to hire a freelancer to work on a specific Mathematica problem?

Searke

8:12 PM

@clr66 That would probably depend very much on the kind of problem.

There's a department specifically for doing this kind of work at Wolfram Research.

clr66

Just a few hours working on solving a differential equations problem I'm working on... I think the answer may be quite simple, but I just can't figure it out so far.

Searke

0

Q: Applying l'Hopital's Rule to NDSolve with Piecewise Function

Ok, so I've been working on this problem for over a week and it's driving me crazy. I'm trying to solve a system of two differential equations in $x(t)$ and $y(t)$ using a "time-elimination method," such that $\tfrac{d x}{d t}/\tfrac{d y}{d t}=\tfrac{dx}{dy}$ reduces the system to a single diffe...

differential-equations

related to your stiff systems?

clr66

Yes. I tried to phrase the question in a way that could lead to a general answer... but there's also the possibility that the solution I need would be peculiar to my particular problem.

Searke

the code to reproduce the problem isn't there unless I work backwards from some of the definitions

try of course StiffnessSwitching->True

reference.wolfram.com/mathematica/tutorial/…

clr66

8:27 PM

Unfortunately, the differential equations producing the problem are too long to post. I've tried StiffnessSwitching, but no luck.

Searke

9:01 PM

the first thing to know is that it's not always possible to overcome the stiffness of an equation

its sometimes impossible

If you're already familiar with what stiffness is in this context then you should be able to identify why they are stiff

you may need to related your current differential equations symbolically or approximately to another set of differential equations which are not so stiff

none of this is easy and only sometimes am I ever able to get a solution to something NDSolve considers to be stiff

clr66

Yeah, that's what I'm afraid of. Unfortunately, I need an answer to this problem and I've run out of ideas myself. Part of the problem is that equations themselves are so messy, I'm not sure how to go about approximating them in a useful way.

At this point, I'm ready to try to hire someone to work through the specific problem with me. I'm looking at freelancer.com, but I have no experience with any of those kinds of sites...

Searke

9:29 PM

Knowing what is solvable must always inform your choice of how to model something. Are you at a university? what kind of equations are these

Searke

9:47 PM

Sometime I've had luck making subsitutions into my equations like y(x) = Exp[f[x]]

and this makes for a more tractable problem even if the actual equations get messier

rm -rf

@Searke While I have you here, do you know why the following doesn't return an integer for the second case Head /@ Rationalize@{5.4 10^10, 5.4 10^11} and returns the same number?

The docs say this:

but it seems like the condition is satisfied for p = 54*10^10 and q=1

Searke

I honestly don't know.

I've never questioned rationalize's capricious decisions

rm -rf

Heh

Searke

hold on a sec and let me look at it

rm -rf

Ok, thanks

Xerxes

9:55 PM

@rm-rf isn't it because the binary approximation to 5.4 when multiplied by 10^11 is not an integer?

notice it works for 5.4*^11

Searke

So you could always choose a p and q to get as arbitrarily close to a number as you want

rm -rf

@Xerxes Umm... it doesn't. That's just a short hand for 5.4 * 10^11

Searke

so Abs[p/q -x] < c/q^2 confuses me since p and q could always be chosen to satisfy the relationship

except when floating point precision subtraction is taken into account

Xerxes

ah, sorry, I had Rationalize[#,0]&

Searke

Abs[54000000000 - 5.4000000000000006`*^11] < 10^(-4) returns false

chuy

9:59 PM

I would guess a precision issue : Rationalize[5.4`17 10^11]

Searke

oh wait I did that wrong sry.

rm -rf

@Searke you're missing a 0 :)

Searke

yeah and it seems to work when I use another 0

I'm not sure how the function could guarantee to find a result in this case

What one of my colleagues believes is that the documentation is saying "if Rationalize gives a result, then it satisfies this criterion"

rm -rf

No, later on the docs are more explicit (see properties & relations):

Searke

Not "Rationalize will give a result if this criterion is satisfiable". The criterion appears to be always satisfiable in normal arithmetic

rm -rf

10:04 PM

> When Rationalize[x] returns x unchanged, there is no rational number satisfying this

That is a pretty strong claim

Searke

maybe it means in machine precision

this is legit confusing

this is why my code has Rationalize[#,0]& so much

I'd send in an email to [email protected]

rm -rf

Ok, will do today

Searke

the worst that would happen is that we'd file a report to try getting the documentation clearer and hopefully get a concrete answer about why this is happening in this case.

rm -rf

(or you could file it right now for me :D)

Searke

Personally, I've always treated rationalize like Simplify. You get what you get.

It's unfortunately after 5 here and I'd need your email address anyway. I gtg.

rm -rf

10:09 PM

@Searke That's how I've always used it too. I happened to notice it accidentally when writing this answer

I simply couldn't explain why I wasn't getting an integer, but didn't want to waste time over it, so just inserted the integer by hand

rm -rf

10:39 PM

Something else to not try:

@DanielLichtblau I naïvely thought one could do

Block[{$ModuleNumber = 1},
 Unique["x"]
 ]

for the addendum, but... what happened was not pretty:

Mathematica has detected an internal error:
  vMessage ENULL

(v9). However, it doesn't crash in v7 and v8 (displays a warning). Any thoughts on why that happened? My guess would be that $ModuleNumber is critical to several internal routines and this messed with it and it cannot be Locked either, since it needs to be incremented. But somehow I don't think that's right (and I'm not convinced by my explanation) — rm -rf 13 mins ago

acl

11:41 PM

@rm-rf hm, mine does this about once every 10 minutes anyway. I thought it was normal :)

Transcript for