« first day (209 days earlier)      last day (3273 days later) » 

12:10 AM
Hi @Yu-SungChang
12:23 AM
@Verbeia How I hate reading RFPs!
@Verde they are no fun to comment on, either. All sorts of legal nonsense. Part 3 and Part 4 are the main technical content.
@Verbeia I mean, each institution has its own mindset for writing those little art pieces. Some include a reference price and term, others enforce one, others lets you on your own. But they all coincide in a few nuisances. For example, you have to read 140 pages to understand if there is a predefined Go Live date or not
@Verde There isn't a predefined Go Live date in this case.
@Verbeia yeah :D I've read 100 pages to know that :D:D
@Verbeia My company writes RFPs for others, and we always insist about including an informal flyer with the relevant info for the proponents to understand quickly what is ging on inside the monstrous bulk of pages that follows
@Verde The main risk will be the "oh wait, can you also add this?" scope creep. They would rather take the time to do it right than set an arbitrary date. I really like the flyer idea.
12:35 AM
@Verbeia We specialized on core banking RFPs along these years. It's a funny business
I've a flyer "template" with the info we usually include (not the format, just the contents). If you do a lot of RFPing, I could send you a copy
Hi. A new user asked here
A: Plotting contours plot for f(x,y,z)=c

halirutanOk, lets give this a try. @Mr.Wizard already showed you, how you can use Interpolate to make a function from your discrete data and since you didn't provide some test-data, I just assume we are speaking of an isosurface of a function $f(x,y,z)=c$ which is defined in some box in $\mathbb{R}^3$. ...

whether it's possible to upload a large data file. Has someone an idea whether this is possible?
And what happened to the chat? I can not longer press enter to send a message and have to take the mouse??
test pressing enter
<enter> works
who? who is in the second base?
@Verde hehe ;)
@VitaliyKaurov Repeat with me: "I shouldn't delete my questions"
Then this seems to be a plugin for Chrome on Linux.
12:46 AM
@halirutan An enter-nuking plugin? It deserves to be a plugout
@halirutan He could upload the text file to ideone.com, for example
Hmm, the chat in Chrome seems to be broken. In Firefox it looks completely different.
or pastebin
@Verde, but I remember that someone wrote anywhere that it would be good if uploaded files were available permanently.
@Verde thanks for the offer but I hope never to be involved with one again
@halirutan "permanently" is too much a word. that data is not that relevant, (it makes the question Too Localized. Get him to upload it somewhere, and that is it
@Verbeia Good for you!
12:51 AM
@Verde Ok.
Chat's busy tonight.
@halirutan I started using ge.tt, mainly because I saw Szabolcs using it. Seems pretty good.
@OleksandrR. I'll have a look and will post a comment to the guy asking it.
Btw, do you use Chrome on Linux?
Me? Opera on Windows.
Lucky for you.
My chat looks like this now:
No sidebar, no upload button and I have to click send to send a message.
Wowsers trousers! What a mess...
Could switch to Opera... ;)
1:01 AM
I mean.. I have to take my hand off my keyboard, move it to the left, find my mouse, point on the send button and manage to click on the left button..
I used Opera a very long time before I switched to Chrome.
Newer versions of Opera look just like Chrome by default. Personally I hate the modern fashion for how browsers look and behave, so I changed it, but I think if you were to install Opera you'd not have any difficulty in adjusting.
I think I try to get them fix this bug so I can stay with Chrome ;-)
@halirutan Anyone knows how is the Firefox business model? How do they get funded?
@Verde I have no idea
toading danger
1:07 AM
@Verde I looked it up once (because I'm naturally suspicious of people whose motives I can't discern...) but I forgot now; companies donate because they want a decent browser for their corporate users or something. Anyway what I remember is that they are doing quite well for themselves.
Do ye all know the story of the princess and the toad?
@OleksandrR. Same reason for me to ask ... :D
@Verde Well, last annual report they posted on their website was 2010. So, maybe they're completely evil now...
@OleksandrR. as evil as the rest of us, I guess
@VitaliyKaurov why use bitly for a site link?
Damn ... What is "to wrtch"? my English is really bad
And "OX X" ? is a new acronym?
@R.M Because if not, the chat gnome transforms it to a big image
1:16 AM
@Verde Maybe he meant using an XXX Lion whatever dirty thing this should be..
@halirutan I have the stackmma github account. I added you as a user to the attachments repo. If you find something you need to attach, please do so. Szabolcs is also on it and we've been using it for a few questions
@Verde you just need to write something else on the same line... it expands only if it's the only thing on the line
@R.M thanks but it was a new user who wanted to upload something since my solution didn't work on his special problem..
@R.M not sure. I think it is "only if it is the last thing on the line
@Verde nope... it has to be the only thing and not on a multiline
@Verde Ted posts from an iPhone half the time.
1:19 AM
So I think that explains the typos and the terseness.
@Verbeia I am joking, I respect Ted a lot.
So has someone an idea what wrtch means?
@Verbeia "His" site was the first thing I've read about Mma
@halirutan SWITCH
fixed it
1:22 AM
my pork tenderloin with apricot chutney is waiting. Sorry for you.
1:41 AM
Q: Chat behaves/looks strange in Chrome on a Linux box

halirutanHere, on my Ubuntu 64bit with Chrome the chat does look very different from what I was used to see for a long time. the right info-bar with the list of online users and the starred comments is completely gone. on the bottom I have only a send button where there should be and upload button too...

9 hours later…
10:52 AM
@Verde We have a toad hanging around here; who gets to be the princess?
1 hour later…
12:10 PM
@J.M. We could hold an election on meta to select a princess for our toad. Just beware of underdogs
can someone work out how to pose this question correctly?
Q: How to ensure that Polygon[ list ] plots a convex polygon?

ndroock1Consider the following code which plots a triangle. p = {{0, 0}, {.2, 0}, {0, .2}}; {Cyan, Polygon[Dynamic[p]]} // Graphics Then adding ( for example ) {.1, -.1} yields a non-convex polygon. AppendTo[p, {.1, -.1}] Question: Given a list of 2D points that are plotted as a polygon by Graphi...

he has clarified (sort of) what he wants in the comments, and it's not what he actually asks, but I don't know how to express it well. it would be nice if this could be fixed rather than randomly upvoted!
I can't tell what he wants. He should probably post his actual problem...
Yeah me neither
@J.M. I think that this is his problem: given a list of points, arrange them so that Polygon does not give a polygon with a constriction
so, arrange them so that polygons which can be cut in two by removing a single point do not get produced
that is what I understood
(from the comments)
12:20 PM
So, apply FindCurvePath[] first?
I don't know if that is always possible
well, I don't even know if that is the question. I asked him if this is what he wants in a comment
geometry problems can't be stated as "I wish something like"
@J.M. Try
`p1 = {-1, 1}; p2 = {1, 1}; p3 = {-1, -1}; p4 = {1, -1}; p5 = {0, 0};
p = {p1, p2, p4, p3, p5};
I can't try at the moment; I have something rendering. :(
FindCurvePath returns {}
probably FindShortestTour is the way to go
but the problem is blurry
12:29 PM
@Verde Hmm, odd. It doesn't know what to do.
and FindShortestTour was already suggested in the comments
I think it is a NARQ
Anyway, Verde... can you be my Guinea pig for the moment?
(You too, @acl, if you can...)
@J.M. that depends!
Okay, execute this first: test = Interpolation[Join[{{0., 1.}}, Delete[First[Cases[Plot[Cos[x]/(1 + Sin[x]^2), {x, 0, 2 Pi}], Line[l_] :> l, Infinity]], {{1}, {-1}}], {{N[2 Pi], 1.}}], PeriodicInterpolation -> True].
then, try NIntegrate[test[x] Cos[3 x], {x, 0, 2 Pi}]
12:36 PM
MapThread::mptd: Object InterpolatingFunction[{{0.,6.28319}},{4.,7.,0.,{313.},{4.},0.,{1.},0.,0.,Automat‌​ic},{{<<1>>}},{Developer`PackedArrayForm,{<<1>>},{1.,0.999994,0.999978,0.99995,0.‌​999911,<<42>>,0.956497,0.948089,0.946993,<<263>>}},{Automatic}][<<1>>] at position {2, 2} in MapThread[NIntegrate`InterpolatingFun
That's what I got too. I wonder why...
without PeriodicInterpolation->True it does not do that
Try NIntegrate[test[x] Sin[x] , {x, 0, 2 Pi}]
same message
@Verde Still rendering something, so maybe later...
@acl So, it's the periodic setting. Damn.
This throws a wrench into my Fourier expansion attempts...
12:45 PM
@J.M. After the error, you get the same result as with Priodic->False
But with Priodic->False it takes MUCH longer
add a Quiet@ before NIntegrate and presto!
I could use Quiet[], but it feels like cheating...
@J.M. results of timing
{{0.031, 0.446494}, {3., 0.446489}}
a hundred times slower
@J.M. so, either you cheat or you wait 100 times longer. sounds like daily life then
@acl Tsk, Scylla and Charybdis...
1:07 PM
@J.M. What do you think of deleting Shog's default community ad in favor of cormullion's? I don't see a reason for two of the same ad which reduces chances for others
@R.M I prefer cormullion's myself.
But more people voted for Shog's...
3 hours later…
4:01 PM
@R.M ...and it's deleted.
2 hours later…
6:17 PM
Random thought: other graduated sites have renamed their default chatrooms. For ours, how about "The Notebook"?
Also, sites like Programmers have their blog chat room accessible on their main page, also. Any one know how we would do that, also?
6:41 PM
Please vote on this. It's an effort to get some unanswered questions answered (or cleaned up by editing/closing/deleting). Maybe we can also have a chat event to comb through the list and pull the weeds.
@rcollyer it is shown automatically, possibly to a random selection of visitors
@R.M That's not what I meant: on Programmers their blog chatroom is also showing up, not just their normal chatroom.
@R.M That list has a few weeds, but a lot of them are good, and very, very hard. :)
@rcollyer It depends on activity... our blog chatroom has not had a message in 10 days, so it won't be shown.
@R.M ah. Right. I should post something on the blog, but the damned dissertation, death in the family, new job, and just life in general has made it very difficult to get anything done. :P
@rcollyer yes, some are definitely good and hard, which is why more eyes are needed. Not many know about the tab
@R.M Do me a favor: reread that sentence of yours, then go wash your mind out with soap! :)
6:59 PM
That was an excerpt from your thesis
So, will you be relocating or working remotely?
For the moment, remotely. Then, relocating once everything with school is done.
Oh, you started? Nice!
When's your defense?
Not scheduled, yet. Which is polite speak for I'm not done writing, yet.
7:02 PM
@rcollyer thanks for recommending "Hyperion"
@acl You're welcome.
@rcollyer I've read Hyperion, Fall of Hyperion and now am reading Endymion
I take it you enjoyed reading it?
amazon.com thanks you
@acl I haven't had the time to get the rest.
7:02 PM
@rcollyer yes, very well-written
@rcollyer oh congratulations!
what sort of job?
@rcollyer I see. well, you could explain it later though right?
(I mean, had it existed etc etc)
@rcollyer well, then don't say that in here...
@acl absolutely. It will be several months, though.
@R.M To bad we don't know a moderator who can erase it for me?
hi folks
speak of the devil
7:08 PM
need any erasing done?
Hey, Sjeord: the big block several lines up, replying to acl.
rcollyer's thesis
@R.M yeah, off with its head
@rcollyer Hope that was the one
Screen saving ... done!
7:10 PM
@SjoerdC.deVries Yes it was. How about this one, too.
I was slow
@Verde How can I create fake chat transcripts in Mathematica?
Cool, thanks @SjoerdC.deVries.
@R.M why bother on creating fake ones, if the real ones are alreay surreal?
@Verde isn't that always true?
(I tried to resist, but I couldn't)
7:12 PM
@rcollyer Ok. I was going to keep my screen save for me, but now ...
Poof! If all problems could be solved so easily...
@SjoerdC.deVries Isn't that the truth.
@Verde :)
Any thoughts on this?
@Verde you'll email it to?
@R.M About the SphericalPlot3D question...
@rcollyer I'd go for "The Kitchen Sink"
7:13 PM
@rcollyer <<Chat`
@acl KitchenSink[]?
@rcollyer ah, where is @heike when you need her!
@acl slacking, as usual.
@R.M You knew it didn't draw anything. So, "get a faster machine!" is a bit harsh, especially for a response to a first post
7:15 PM
@Verde speaking of surrealism, I randomly ended up reading the following paragraph a few minutes ago:
Fernand Meysonnier is France's last executioner. Over the years, he has helped chop off the heads of 200 people.

In a small village in Provence an old man packs up a selection of terrifying torture instruments. This is M. Meysonnier, chief executioner until the guillotine was axed by the government in 1977. He performed his first execution at 14 and never looked back.
I like TheForbiddenPentagon[]
@SjoerdC.deVries Actually, I fixed the code to make it run and then commented
@R.M Per the prior conversation, today I spent trying to debug my install of LevelScheme` . The default layout is fubared, and I was trying to fix it. Guess what amount of work I got done today? :P
@R.M It doesn't run at all.Just a blank box
Ok, r could've been something more complicated, but it turned out it wasn't
7:17 PM
I thought that the OP forgot to include the function to be drawn. One that could have been quite complicated; hence, his laments about speed
At the moment, there's still no function at all
@SjoerdC.deVries actually, it wasn't meant as a harsh comment or sarcasm; I meant it genuinely... I also told him it ran smoothly on mine (didn't mention I fixed it) and that he should use PerformanceGoal -> "Quality" in case the slowdown he was seeing was due to plotting twice
I've got to run. Hopefully, I'll talk with you all again soon. I don't expect it to be this week, though.
@rcollyer see you
@SjoerdC.deVries From his update, it looks like my interpretation and Simon's was what he wanted
7:21 PM
@rcollyer Good luck!
He was trying to approximate a solid by drawing closely spaced shells... My guess is that he used a very fine spacing
@r.m. OK. Apologies for misinterpreting you
no worries :)
@R.M which update?
oh I see, he added a paragraph
7:56 PM
It would be nice to know how many years it took the experts here to become experts at Mathematica
I've been using it for more than 2-3 years now, and I feel I'll never be good at it :)
I mean not as good as the experts here.
I think one is born a Mathematica expert or not. It can't be just a learned skill.
@NasserM.Abbasi I'm not an expert, but I think this statement ("one is born an X expert, it's not a learned skill") generally doesn't apply to anything but the furthest extremes of performance
like Olympic athletes; no matter how hard I train for the 10km race, I'm not going to the olympics; but I can get much, much better at it than pretty much anybody who doesn't train
it's the same for less physical pursuits
it's just a question of motivation and organization (and of course being "smart" helps!)
8:20 PM
@Verde bit anticlimactic, this
@acl hehe "Born to Map[]" is not bad
@acl I always loved all those. They are so realistic!
now I have to try really hard to not make one and print it on our A0 poster printer at work
@Verde yes, that's why they're so funny!
8:26 PM
@acl Map like the eagles or loop like a groundhog
the last few days/weeks, I am doing some exact calculations for quantum dynamics. that means you calculate some gigantic matrix (the hamiltonian), exponentiate it, and multiply the state vector by it. that's generically quantum time evolution. as you can imagine, doing this sort of thing for days loosens one's grip on reality.
one early morning, just after finishing work for the "day", it occured to me that, to the best of my knowledge, that's what the universe is really doing: multiplying two huge matrices! all our hopes, fears and dreams are basically artefacts of matrix multiplication!
well, it seemed funnier at the time, but I still think it's hilarious
(and funny for the same reason as those posters; it's true, actually)
@acl Nahhh The Universe is just drawing Feynman Diagrams. True Art.
@Verde no but seriously it really is just multiplying a matrix and a vector (well, if we ignore relativity)
in fact I tried making this point to my mother a few months ago when I was visiting
didn't work
and I hate diagrammatic perturbation theory with a vengeance
this is one question I kept meaning to ask but never got around to
@acl Please! Oh Please! Put your feet on the ground again! Those are just models ... Reality is a a philosophical entity, not a physical model
@Verde an excellent point, there
8:36 PM
many nights calculating bra-kets here
renormalizing, trying vacuum perturbations
and getting a divergence again and again
yes I know the drill. now I stick to lattices
@acl What kind of problems are you tackling?
@Verde right now? a lattice system which is driven with some periodically modulated force (I am trying to make some more general point and use this as an example)
basically trying to come up with something analogous to the Gibbs ensemble for non-equilibrium systems (ie, just unitary time evolution)
for closed systems
well, this is the main think I myself am doing the others in my group are doing other things which I just help with (in principle)
@acl closed but driven .. ok
@Verde in fact, even closed, not driven but not in an eigenstate is interesting
if you have a two state system it's very easy to see how the time evolution goes; but take now some huge system, then it's not easy to see what will happen.
8:43 PM
non-equilibrium physics is always interesting
and "generically" what does happen is that each possible state i is visited with "probability" exp(-E_i/T) with T some lagrange multiplier (recall this is all unitary time evolution)
@Verde yes, actually, one question I always had since I was a teenager is, how come I can look at a glass of water and describe it so well with so few parameters?
and what the spectrum looks like?
I guess I am now trying to answer that
@Verde it depends on the system, but the point is precisely that for "generic" spectra, you should go to a time-independent state at long times
... or chaos
"generic" means something analogous to "chaotic" for classical systems
I am being loose and use "chaos", ergodic etc interchangeably
8:46 PM
ohh ... sorry
@acl and how you could prove that a generic driven system achieves a time independent state?
you need some mechanism for entropy dissipation
well the basic insight is very simple, if the initial state is $|\psi(t)>$, and the hamiltonian has eigenstates $|u_n> with evals $E_n$, expand $|psi(0)>=\sum_n c_n |psi(0)> with the $c_n$ constants, then $|\psi(t)>=\exp(-iHt)|\psi(0)>
@Verde no hang on
@acl, if you are not an expert in Mathematica, then I do not know who is :)
and $\exp(-iHt)|u_n>=\exp(-iE_nt)|u_n>$, so now anything like $<\psi(t)|A|\psi(t)>$ will involve objects of the form $\sum_{m,n}\exp(-it(E_m-E_n))c^*_m c_n A_{m,n}$
I just send a pendulum demo to WRI for the demo site.
which for large $t$ you can see will be well approximated by $\sum_{n}|c_n|^2 A_{n,n}$...
8:51 PM
damn I haven't installed TeX here
...which is time independent!
about the entropy it is always zero as I am describing a pure state.
A: Should chat have TeX support?

robjohnThis bookmark processes the current page with MathJax. It has been modified from the bookmark on this page to handle $$...$$ and to handle \\[...\\] properly, to include AMS additions, and to update automatically. The COPY TO CLIPBOARD link on pastebin.com should copy the bookmark to your clipbo...

ok one more sentence and I'll shut up
I am just worried my physics model might be not quite right now. There are physics gurus here I know. Can I ask something about it?
@NasserM.Abbasi (leonid, RM, JM, Verde/belisarius, MrW, the various WRI guys, Verbeia, rojo, wreach, etc etc etc etc etc :) I'd say I am average)
@NasserM.Abbasi I doubt I'll know the answer but maybe someone else will so go ahead
Ok, I had mass-spring system. basic thing. But it is in a tube. When the mass is moving to the right and it reaches the end of the tube, I set the speed to zero
@Verde anyway upshot is: what i said demonstrates (if it doesn't fail for some models) that you go to some time-indep state; but there is one state for a given initial condition. this is not how you envision stat mech. this can be fixed, by using maxent aproaches, but I guess I am pushing people's patience if I start on that now :)
8:56 PM
I am thinking now that I should have set the speed to the negative of the last speed. i.e. treat it as perfect elastic collision with the wall of the tube.
I am using NDSolve to solve one step at a time. I use the last solution as the initial conditions for the next solution
@NasserM.Abbasi I suppose it depends if you want your collision to be elastic or inelastic; if the wall is marble it'll be closer to v->-v, but if rubber, v->0 I guess
@acl thanks for the link!
@acl, yes, good point. But I did not explain this in the demo. I just set speed=0
@acl the demo is still not published. I can fix it and send an update before if I have to
@acl It is very interesting, but perhaps this is not the right room :)
@Verde indeed
9:01 PM
The problem is the mass is attached to a sping and not free mass. So when it hit the wall, I was not sure what to do, so I set the speed to zero for the next integration step. But now I think I should have made the speed negative, so it is more realistic I think
@NasserM.Abbasi If you set v=0, for an enough high speed you will end up with the mass stuck there
centrifugal force > spring force, and the mass is dead stopped
@Verde is the thing rotating? can I take a look at this demo?
@acl Nasser showed it to me yesterday
@acl, the mass actually does not get stuck there. Can you please try it and see if the physics makes sense to you? it is in CDF, here it is 12000.org/my_notes/mma_demos/slot_on_disk/index.htm this one sets V=0 when bob hits the wall of the disl
It does stop there for sometime, but after integration for more steps, it evantually leaves the wall back to the center of the disk. It does not get stuck there.
yes it does
with parms
9:07 PM
it depends on the spring stiffness if it gets stuck there or not. You can try test case 3 and then change the spring stiffness, then you'll see it does not get stuck for long
ok. what I said is that if you set v=0 at the wall you could get a dead mass
@NasserM.Abbasi I think that's his point, suppose you are sitting on that wall with no velocity towards the centre; the thing is rotating, pushing you to the wall with some constant apparent force. if the spring stiffness is less, you'll stay there, right?
there's nothing wrong with this happening
Ok, I'll make a new version now, where when the bob hits the wallk, OR hits the center of the disk, I'll reverse the speed. Is this ok? Should be better physics
@NasserM.Abbasi or claim it's a rubber wall and save effort
@Verde correct. FOr small spring stiffness, it gets stuck
9:10 PM
@NasserM.Abbasi you could also "perforate" the disk center and let the mass go through
@acl, yes, but it is easy for me to change it to make it bounce back. And I think it is more realistic that way?
@NasserM.Abbasi you could put a switch
@Verde, yes, I thought of that, but that is too complicated :)
@acl, good points, but too many options, and WRI editor complain allready I have too many options :)
hehe a frictionless mass, an ideal spring, and you ask for "realistic" :D
Whatever they say, claim it's all due to the Coriolis effect and be done with it. No one will check ;)
9:16 PM
@R.M like entropy:
I thought of calling it 'information', but the word was overly used, so I decided to call it 'uncertainty'. [...] Von Neumann told me, 'You should call it entropy, for two reasons. In the first place your uncertainty function has been used in statistical mechanics under that name, so it already has a name. In the second place, and more important, nobody knows what entropy really is, so in a debate you will always have the advantage.'
Conversation between Claude Shannon and John von Neumann regarding what name to give to the or attenuation in phone-line signals[37]
I just uploaded a new version now, where I flip the speed when it hits the wall or the center. 12000.org/my_notes/mma_demos/slot_on_disk/index.htm please see if this looks better. I think it does myself, it feels more real now that it bounces back. I can send this version to WRI before they test it
@NasserM.Abbasi Are you creating these out of personal interest or are you TAing/teaching a physics course?
@R.M No! I am sure it is a monopole!
@R.M. just for me to learn dynamics. I can't learn dynamics by just looking at equations, I have to see the thing running
Mathematica makes it very easy to make simulations and see things running.
9:21 PM
Yes, it helps in visualizing things
@NasserM.Abbasi But ... you should try to draw the curves without Mma
9:42 PM
Ok, this is what I did in this version here 12000.org/my_notes/mma_demos/slot_on_disk/index.htm when the bob hits the edge of the disk, I flip the sign of the speed for the next initial conditions. When it gets to the center of the disk, I set the speed to zero. The reason is: There is a spring attached at the center of the disk. So at the center of the disk, it is not the same as the edge. I think setting it to zero there makes more sense. Please try this. Use test case (3) and hit run.
I think there is something not right in this model When I let it run for long time, it becomes unstable. Selecting test (3), set simulation speed to max, and hit run. After 1-2 minutes, the disk starts spinning much faster and things do not look right. I think I have something wrong somehere :(
I think now that flipping the speed sign when the bob hits the wall is not relastic physically since the bob is on a spring. The system is gaining kinetic energy causing the disk to spin faster now. This is not right.
10:24 PM
Ok, now it is ok. Now it runs ok, flipping the speed at the edge. I had a bug. I was integrating one extra step, then fliping the speed after words. Now I flip the speed before. Now it is stable.
10:43 PM
@acl I wouldn't count myself amongst some of the experts in that list. What I will say is that I have gotten a lot, LOT better in the past year and a bit since I (a) started working on my chart-drawing prototype project and (b) started participating on StackOverflow and then here.
10:59 PM
but it is harder to get better at Mathematica compared to other languages. But if one gets better in Mathematica, they can do more with it than with other languages.
@acl, I took your advice, I added an option to select the wall type: elastic vs inelastic. The CDF is here 12000.org/my_notes/mma_demos/slot_on_disk/index.htm now when select elastic, the bob will flip speed when it hits edge. else it have speed zero there.

« first day (209 days earlier)      last day (3273 days later) »