MODEST meeting 2009-11-11

From Micasim

Project	SL
Project Lead	Rob Knop
Project Homepage	Virtual World Workshops

Transcript

[6:01] Mister Jurassic: Is this the location of the MODEST meeting?

[6:02] Prospero Frobozz: Yes

[6:02] Pan Numanox: I hope so :-)

[6:02] Prospero Frobozz: You're Nathan Leigh, yes?

[6:02] Mister Jurassic: Yes.

[6:02] Prospero Frobozz: Are you in one of the "MICA Names" groups yet?

[6:02] Mister Jurassic: hello!

[6:02] Prospero Frobozz: I'm Rob Knop -- we've talked before, of course

[6:02] Mister Jurassic: I am not sure.

[6:02] Mister Jurassic: Yes, hi Rob.

[6:02] Prospero Frobozz: The MICA Names groups are something we use so that our professional members can have their name floating over their head in the "group title" as i do

[6:03] Mister Jurassic: OH, then I don't think so.

[6:03] Prospero Frobozz: OK, I've just invited you

[6:04] Kerry Giha: It is a good idea

[6:04] Mister Jurassic: Great. Thanks!

[6:04] Prospero Frobozz: If you set "MICA realnames2" to your active group, then your name will be over your head in the group title

[6:04] Prospero Frobozz: Pan -- are you an astronomer? (Have we talked before and have I forgotten?)

[6:04] Pan Numanox: Yes, Rob. We've met too. Alf Whitehead.

[6:04] Prospero Frobozz: I was really hoping some of the Leiden guys were going to show up as I wanted to quiz them about some AMUSE things :)

[6:04] Prospero Frobozz: Ah, OK!

[6:04] Prospero Frobozz: Is Steve going to make it?

[6:05] Rocky Sittingbull: hello? :)

[6:05] Pan Numanox: As far as I know. I haven't talked to him in the last couple of days, though.

[6:05] Prospero Frobozz: I haev this fear that Steve won't come after Piet's email last night saying he wasn't going to make it this morning....

[6:05] Prospero Frobozz: Hi Rocky

[6:05] Rocky Sittingbull: ah downstairs

[6:06] Rocky Sittingbull: i was pointed to this place by my professor

[6:06] MICA seminar seat whispers: Hi Rocky Sittingbull! Touch me for Menu. Say /1a to Adjust.

[6:07] Prospero Frobozz: Rocky -- cool... who's your professor? And who are you? :)

[6:07] Rocky Sittingbull: i'm a student from Leiden

[6:07] Rocky Sittingbull: and Simon pointed me to it

[6:07] Prospero Frobozz: I'm Rob Knop -- MICA member, itinerant astrophysicist and computernerd, one-time professor. Alf is a grad student with Steve McMillan at Drexel

[6:07] Rocky Sittingbull: Portugies Zwart

[6:07] Prospero Frobozz: Yes, I've worked with Simon

[6:07] Prospero Frobozz: I was out in Leiden back at the beginning of October -- did I meet you then?

[6:07] Rocky Sittingbull: no i don't think so, i actually just began my Masters

[6:08] Prospero Frobozz: Ah, OK. Do you know what you'll be working on yet"?

[6:08] Rocky Sittingbull: well there is a possibility i might do my Major Research with Simon, starting in april or may 2010

[6:08] Prospero Frobozz: Kerry -- I believe we've met before, haven't we? You're not a professional astronomer, if I remember correctly... do I?

[6:08] Rocky Sittingbull: and he was mention self scheduling routines...(?) lol

[6:09] Prospero Frobozz: Self-scheduling routines... OK. That could be a lot of things :) Alf, do you know if that's a term that gets used in this game for particular things right now?

[6:09] Kerry Giha: I am not a professional

[6:09] Prospero Frobozz: Rocky, are you familiar at all with AMUSE?

[6:09] Prospero Frobozz: Or with "grid" computing?

[6:09] Kerry Giha: I am more of a student getting into the physics field

[6:09] Kerry Giha: I have a strong computer background

[6:10] Pan Numanox: Not exactly...

[6:10] Rocky Sittingbull: i can only see i've heard of it, Simon showed the MICA site, and told me about this place

[6:10] Rocky Sittingbull: *say

[6:10] Rocky Sittingbull: and ACS site

[6:10] Prospero Frobozz: OK. AMUSE is the evolution of MUSE, which is a "metapackage" of numerical codes. There are astronomical numerical codes for doing various different things.

[6:11] Prospero Frobozz: For example, gravitational n-body codes handle point masses orbiting around each other -- stars, often, or particles to represent dark matter (which is collisionless)

[6:11] Prospero Frobozz: Then there are hydro codes for dealing with gas

[6:11] Prospero Frobozz: Stellar evolutino codes for dealing with stars changing over time

[6:11] Rocky Sittingbull: yes

[6:11] Prospero Frobozz: Radiative transfer codes for light going through gas

[6:11] Prospero Frobozz: etc.

[6:11] Prospero Frobozz: Often, people have just done one thing, but of course, in reality, all of this physics comes together in many of the systems of interest.

[6:11] Prospero Frobozz: So, what MUSE did was provide a framework that made it relatively easy to stictch together different codes

[6:11] MICA seminar seat whispers: Hi Cale Jaxa! Touch me for Menu. Say /1a to Adjust.

[6:11] Prospero Frobozz: And AMUSE is the evolution of the same thing

[6:12] Prospero Frobozz: It wraps various codes, and lets you run one for a while (e.g. N-body to move the stars around), then call another (e.g. evolution to let your stars advance in age),t hen back (N-body to orbit them some more), etc.

[6:12] Prospero Frobozz: The whole thing is a python shell around all of these various other codes, the codes that do the heavy lifting

[6:12] Prospero Frobozz: "Grid" computing is sort of a buzzword -- it's been around for at least 10 years, although 10 years ago I think it was more a dream. Nowadays people really do it

[6:13] Prospero Frobozz: The idea is that you have jobs you need done, and you can "easily" move them to different sites, clusters all over the place.

[6:13] MICA seminar seat whispers: Hi Linguini Mexicola! Touch me for Menu. Say /1a to Adjust.

[6:13] Prospero Frobozz: So you can find the resources you need for whatever problem you have at hand-- some kinds of clusters may be better at some things-- and have the jobs move between them.

[6:13] Prospero Frobozz: I'm *guessing* that the self-scheduling code Simon is talking about is related to this

[613] Prospero Frobozz: The idea of having codes that figure out what they need to do next, where they need to run, and running there, and then freeing up the resources when they don't need them any more

[6:14] Prospero Frobozz: But.... that's me just guessing. :)

[6:14] Prospero Frobozz: Hey Inti!

[6:14] Prospero Frobozz: Hey Arjen!

[6:14] Rocky Sittingbull: yea it was...taking care of particles individually

[614] Linguini Mexicola: hi there

[6:14] Remy Vespucciano: Hi!

[6:14] Prospero Frobozz: As you guys can tell, I've been playing with AMUSE, that's why I keep sending you email....

[6:14] Linguini Mexicola: great

[6:14] MICA seminar seat whispers: Hi Remy Vespucciano! Touch me for Menu. Say /1a to Adjust.

[6:15] Prospero Frobozz: I have a question -- if one of the legacy codes is intrinsically parallel itself, how do you tell it which machines to run on? If you were running from the commandline, you could use a -machinefile argument to mpiexec. How do you do the equivalent from the python wrapper?

[6:15] Prospero Frobozz: Hey Cale -- are you a prof. astronomer?

[6:15] Cale Jaxa: hello

[6:15] Prospero Frobozz: Re: my questino -- that was about AMUSE, in case it wasn't obvious :)

[6:16] Remy Vespucciano: Rob, i've not added support for that yet

[6:16] Prospero Frobozz: ah, OK :)

[6:16] Remy Vespucciano: we are currently integrating our first true parallele code

[6:16] Prospero Frobozz: Then I wasn't being dense. I didn't see arguments for it, but I haven't grokked the full class structure yet, and I haven't figured out where any of the MPI implementation is

[6:16] Prospero Frobozz: OK -- my next qeustion was going to be if phiGRAPE when running in "emulation" mode is MPI parallel -- I would guess not, then.

[6:17] Cale Jaxa: i'm not, i've taken some astrophysics courses, but I'm not. for those of you who haven't met me, I'm a human-computer interaction researcher--George invited me, so if I'm not intruding I'd like to sit in occasionally on different MICA gatherings where appropriate

[6:17] Linguini Mexicola: at the moment the mpd ring determines which machines will be used

[6:17] Prospero Frobozz: Cale, ah! Good to have you here.

[6:17] Prospero Frobozz: Where do you do your research?

[6:17] Prospero Frobozz: Init : OK... I need to learn more about configuring mpd then.

[6:18] Pan Numanox: Inti, does each machine need a config file to tell Sapporo which GPUs to use, then?

[6:18] Cale Jaxa: looking at how scientific collaboration can happen over virtual worlds, and seeing how it's different from other types of virt. collaboration. I'm with the department of Informatics at Univ. California, Irvine

[6:18] Prospero Frobozz: Oh, you're the colleage of Crista, yes? You were on that grant together?

[6:18] Linguini Mexicola: pan: I think so - at the moment we have only tested machines with identical configurations though

[6:18] Cale Jaxa: I'm a second-year Ph.D. working with Gloria, who's on the grant with Crista

[6:18] Prospero Frobozz: OK :)

[6:18] Pan Numanox: Cool. Sounds like something interesting to play with :-)

[6:19] Cale Jaxa: I've always been interested in both physics and astronomy, though, so it'll be interesting to listen to you guys :)

[6:19] Prospero Frobozz: At the moment, we're talking pretty hard core computers....

[6:19] Prospero Frobozz: I don'tk now if you were here when I was briefly describing AMUSE to Rocky

[6:20] Prospero Frobozz: Remy/Arjen is a software engineer out in Leiden who is one of the main AMUSE Programmers, and Inti is a post-doc out in Leiden who is also one of the core AMUSE people.

[6:20] Cale Jaxa: i don't think I was... I'd appreciate the chat transcript from when I wasn't here earlier in the meeting, if it's okay with everyone, though

[6:20] Rocky Sittingbull: ok

[6:20] Cale Jaxa: i see

[6:20] Prospero Frobozz: Cale -- yes, after the meeting I'll post the transcript (and last week's transcript -- I'm slow!) to the MODEST wiki

[6:21] Cale Jaxa: oh, excellent

[6:21] Prospero Frobozz: The MODEST wiki is on the sign at the front of the room, or if you just want to click : http://www.manybody.org/wiki

[6:21] Cale Jaxa: thank you

[6:21] Prospero Frobozz: http://www.manybody.org/wiki/index.php/Second_Life_Workshops

[6:21] Prospero Frobozz: is where chat transcripts show up

[6:22] Cale Jaxa: great, thanks!

[6:22] Prospero Frobozz: Inti : how are things going with your SPH code? Is the interface stabilizing at all yet?

[6:22] MICA seminar seat whispers: Hi ffd2 Cyberstar! Touch me for Menu. Say /1a to Adjust.

[6:22] Prospero Frobozz: Hey Marcell!

[623] ffd2 Cyberstar: Hi Rob!

[6:23] Prospero Frobozz: I am visiting the American Museum of Natural History right now, trying to help them get their GRAPEs runnign. Man, those machines are fussy bits of hardware...!

[6:24] Prospero Frobozz: There's a bunch of hydro guys up here -- Mordecai-Mark Mac Low is the senior guy, and Jason Maron is his grad student. Jason's been to these meetings before, and is supposed to give a MICA seminar sometime in November.

[6:24] Prospero Frobozz: I've been hearing them bad-mouth SPH :) (They have a Lagrange Particle code, I believe.)

[6:24] MICA seminar seat whispers: Hi Cale Jaxa! Touch me for Menu. Say /1a to Adjust.

[6:24] Prospero Frobozz: I am not familiar at all with Lagrange particle codes; one year I will learn.

[6:24] Linguini Mexicola: At the moment we are working on a radiative transfer code; I get back to fixing the SPH interface after that (we would work towards coupling the two so once rad trafo is somewhat working we would want to do finish and stabilize the hydro sph)

[6:25] Prospero Frobozz: Cool -- I forget the name of the two rad transfer guys who were at the AMUSE meeting. Is this one fo their codes?

[6:25] Pan Numanox: Does "coupling" imply some sort of back-door communication between the SPH and Rad codes?

[6:25] Pan Numanox: Or is this at the interface level?

[6:26] Linguini Mexicola: these guys were Kees dullemond and rowin meijerink, the code I am talking about is from another guy here in Leiden

[6:26] Linguini Mexicola: (Jan-Pieter Paardekooper)

[6:26] Prospero Frobozz is Offline

[6:26] Linguini Mexicola: Kees' code (radmc-3d) we will also work on

[6:28] Linguini Mexicola: Alf: no backdoor coupling ( we will probably want to communicate heating rates or photon flux) it remains very much to be seen how far we will be able to get with the interface

[6:28] Prospero Frobozz is Online

[6:28] Prospero Frobozz: That was very sad

[6:28] Rocky Sittingbull: welcome back

[6:28] Prospero Frobozz: This is the last thing I saw:

[6:29] Prospero Frobozz: [6:26] Linguini Mexicola: these guys were Kees dullemond and rowin meijerink, the code I am talking about is from another guy here in Leiden

[6:29] Prospero Frobozz: Can somebody cut-n-paste what happened after that?

[6:29] Prospero Frobozz: (Sorry)

[6:29] MICA seminar seat whispers: Hi Prospero Frobozz! Touch me for Menu. Say /1a to Adjust.

[630] Prospero Frobozz: It'd be neat if you could do it all with the interface -- just for cleanliness.

[6:30] Prospero Frobozz: Although -- if there was a clean way to set up communication channels directly between codes, that'd be "even better"

[6:30] Prospero Frobozz: The danger is writing ad-hoc communicatino channels between two specific codes

[6:30] Prospero Frobozz: that becomes an N^2 interface building problem

[633] Prospero Frobozz: One thing I do want to talk about since the three most important Leiden guys are here:

[6:33] Prospero Frobozz: Next Wednesday (November 18) is the AMUSE meeting at Drexel.

[6:33] Prospero Frobozz: Alf, you'll be there, yes???

[6:33] Prospero Frobozz: I will be there

[6:33] Prospero Frobozz: Steve and Piet will be there

[6:33] Prospero Frobozz: I suspect Enrico will be there

[6:33] Prospero Frobozz: I *think* Jason Maron will be there

[6:33] Prospero Frobozz: I'm not sure who else.

[6:34] Prospero Frobozz: However, since nobody from Leiden will be there, and Leiden is the center of mass of AMUSE, I'd love it if we could do some "mixed reality" things

[6:34] Pan Numanox: I will be there, save for a class I have to get to.

[6:34] Prospero Frobozz: What are your guys' time constraints?

[6:34] Prospero Frobozz: Pan : is that Steve's class?

[6:34] Pan Numanox: No. I'm in his Tues/Thurs class.

[6:35] Linguini Mexicola: probably the afternoon (local time) is ok for me

[6:35] Prospero Frobozz: Arjen, Marcell, Linguini, I don't suppose we can talk you into working next Wednesday night :) I'll write a note to your families....

[6:35] Prospero Frobozz: :D

[6:36] Prospero Frobozz: Alf : what time is Steve's Wednesday class?

[6:36] ffd2 Cyberstar: :)

[636] Pan Numanox: One sec. I'll get that from the registration system.

[6:38] Pan Numanox: 10-11am EST

[6:38] Prospero Frobozz: OK, and I'm guessing we're all going to want to do lunch... hurm.

[6:38] Prospero Frobozz: How late are you Leiden guys willing to meet?

[6:38] Linguini Mexicola: I can not promise anything after about 18:00, but I may get some time at 23:00 or so

[6:38] Prospero Frobozz: How far is Ledien off of UT right now?

[6:38] Prospero Frobozz: EST is -5 hours from UT

[6:38] Linguini Mexicola: +!

[6:39] Linguini Mexicola: +1

[6:39] Prospero Frobozz: OK -- so 13:00 EST is 19:00 Leiden.

[6:39] Prospero Frobozz: Very sad

[6:39] Prospero Frobozz: Stupid Round Earth

[6:39] Prospero Frobozz: right

[6:39] Prospero Frobozz: So, here's what I propose.

[6:39] Prospero Frobozz: We'll attempt at least two "mixed reality" events.

[639] Prospero Frobozz: The first one will be in the morning EST, starting at 9AM EST and going until 10 or 11AM EST based on how it's going

[6:39] Prospero Frobozz: Steve will have to miss the 10-11AM EST, and HOPEFULLY we'll get people up for 9AM.

[6:40] Prospero Frobozz: 9AM EST will be 15:00 Leiden

[6:40] Prospero Frobozz: Arjen, Marcell, Inti, would any of you be able to make a 22:00 Leiden meeting?

[6:41] Connecting to in-world Voice Chat...

[641] Connected

[6:41] Prospero Frobozz: Either a 22:00 or 23:00 meeting

[6:41] Prospero Frobozz: I know it's late... but I really would love to have some real time communcation between you three and the folks in Philly

[6:41] Remy Vespucciano: Yes, 22:00 would be just ok, 23:00 bit too late for mee

[6:42] Linguini Mexicola: 23:00 could work for me...

[6:42] Prospero Frobozz: OK -- so, let's do the second one at 16:00 EST / 22:00 Leiden

[6:42] Prospero Frobozz: And we will go 1 or two hours based on how things are going

[6:42] Prospero Frobozz: Arjen will head out when he needs to, and Inti will come when he can

[6:42] Prospero Frobozz: We'll "do the best we can given the stupid round Earth"

[6:42] ffd2 Cyberstar: 2200-2400 ok for me

[6:42] Prospero Frobozz: Great, Marcell

[6:43] Prospero Frobozz: I am not sure exactly how we'll do tihs mixed reality thing. I have done mixed reality meetings before.

[6:43] Prospero Frobozz: I did one with the DNOC group at Linden when I worked there; that one worked pretty well

[6:43] Prospero Frobozz: What we did was :

[6:43] Prospero Frobozz: everybody who was at the physical meeting all logged into Second Life.

[6:43] Prospero Frobozz: We used Voice, but we turned down the volumes on each other.

[6:43] Prospero Frobozz: So we heard those in the room in real time.

[6:43] Prospero Frobozz: And everybody who was present only through Second Life could hear everybody in the room... PLUS we each had our own presence.

[6:44] Prospero Frobozz: That, I think, is the best way to do it, but it's fiddily to get set up.

[6:44] Prospero Frobozz: You have to remember to turn downt he voices for everybody in the room with you, or you get "double talking" which is confusing... AND you have to remember ot turn them back up later so that you'll hear them next time you meet in SL

[6:44] Rocky Sittingbull: and hope nothing crashes...

[6:44] Prospero Frobozz: Well, let's not bring realism into this

[6:44] Prospero Frobozz: The other way I've seen it done is to have one person as the "interface" -- an avatar that represents the physical meeting, with a microphone everybody can talk to.

[6:45] Prospero Frobozz: I've found that less satisfying -- it's more like a phone conference call, and doesn't have the" immersion" advantages of virtual worlds

[6:45] Prospero Frobozz: (Cale: this is probablys tuff you'll be able to tell us about as your research advances :) )

[6:45] Prospero Frobozz: Those who aren't in the room feel more disconnected, as there's this "hive mind" avatar that does all the speaking for everybody, and there's no visual feedback (the "green wiggly lines") to help you figure out who's talking

[6:45] Cale Jaxa: haha, i'm learning from _you guys_

[6:46] Prospero Frobozz: Plus, the people in the room aren't all "present" in the virtual world, so they don't feel as rpesent with the remote peopel

[6:46] Pan Numanox: Hmmm... I will see how many headsets I can round up here. Having a good number would help with the Voice Chat.

[6:46] Prospero Frobozz: Anyway, this will be a useful experiment. I expect some resistance, but I'm giong to try to convince folks to try the "everbody jacks in" approach.

[6:46] Prospero Frobozz: At the very least, this will be an experiment in the kind of thing that MICA wants to figure out hwo to do...!

[6:47] Prospero Frobozz: I'll send out a reminder next week

[6:47] Prospero Frobozz: The first "mixed reality" meeting will take place at the time of this workshop -- 6AM SLT / 9:00 EST / 15:00 Leiden

[6:47] Prospero Frobozz: so, no need to cancel the workshop!

[647] Linguini Mexicola: ok

[6:48] Prospero Frobozz: anyway, back to geek-talk

[6:48] Prospero Frobozz: Inti -- how are things going with that radiative transfer integration? There were some questions before that I don't know if you saw about whether you're going to be able to do it all through the python interface, or if you're going toh ave to implement back-door pipes between the codes directly. Do you know yet if you'll need to? Arjen, Marcell, Inti, have you guys given any thought to serious back-door pipes, or are you currently focusing on doing all of AMUSE through the top-level interface?

[650] Linguini Mexicola: its very close to working, we have imported particles, run stuff with a minimal interface, but still some scaling issues, so we have not yet verified that it actually works correctly

[6:51] Pan Numanox: Sorry folks, need to run to a class. See you all next week!

[6:51] Prospero Frobozz: Is this the hydro code you're interfacing the radiative transfer with?

[6:51] Prospero Frobozz: See you alf!

[6:51] Linguini Mexicola: no the rad transfer itself: it is a 'particle' based rad transfer code

[6:52] Prospero Frobozz: Ahhhh, OK

[6:52] Prospero Frobozz: Heh

[6:52] Prospero Frobozz: particle = photon?

[6:52] Linguini Mexicola: (actually based on delaunay triangulation)

[6:52] Prospero Frobozz: Or is it something more than that?

[6:52] Linguini Mexicola: particles are vertices in a triangulation, the radiation is transported along the delaunay lines

[6:52] Prospero Frobozz: That is... does each particle represent a single frequency, or does the aprticle have a rich state (e.g. I_nu as a function of nu)

[6:53] Prospero Frobozz: OK... I don't fully understand what that is, but I've never devled seriously into radiative transfer codes.

[6:53] Prospero Frobozz: my knowledge of radiative transfer comes from the 1d equations at the rybicki and lightman level

[6:53] Linguini Mexicola: the particles are points in space (so SPH particles could be mapped directly)

[6:54] Linguini Mexicola: the radiation transport is at the moment ly-alpha continuum transport

[6:54] Prospero Frobozz: Are you aming at reioinization simulations or some such?

[6:55] Linguini Mexicola: that is were the code comes from, yes, but it will have more general application

[6:55] Prospero Frobozz: Right

[6:55] Prospero Frobozz: Hydrogen is everywhere :)

[6:55] Prospero Frobozz: Well, OK, it's not at the pre-quark-freezout early universe

[6:55] Prospero Frobozz: But everywhere after that

[6:55] Prospero Frobozz: Hurm

[6:55] Prospero Frobozz: OK

[6:55] Prospero Frobozz: It's not before recominbation

[6:55] Prospero Frobozz: Much

[6:55] Prospero Frobozz: because we're talking atoms

[6:56] Prospero Frobozz: But what's a few hundred thousand years of error?

[656] Linguini Mexicola: ;-)

[6:56] Rocky Sittingbull: times goes fast

[6:56] Prospero Frobozz: Rocky : how old are you? :) It only goes faster as you get older....

[6:56] Mister Jurassic: lol

[6:56] Prospero Frobozz: I'm 41 years old, and am still wondering when I'm going to become an adult

[6:56] Rocky Sittingbull: eehm 24 :p

[6:57] Rocky Sittingbull: but it goes fast already

[6:57] Prospero Frobozz: OK... well, one parting thought

[6:57] Prospero Frobozz: Thought for the day : go to google and type in the following search "speed of light in feet per nanosecond"

[6:58] Prospero Frobozz: all you europeans : a "foot" is a unit of measure that the folks in the dark ages (i..e the USA) use. It's 12 inches, and 1 inch is 2.54 centimeters

[6:58] Prospero Frobozz: So it's a bit less than 1/3 of a meter

[658] Prospero Frobozz: And with that, let's close. I'll see y'all next week at the mixed reality event!

[6:58] Mister Jurassic: Sounds good. Thanks Rob!

[6:58] Kerry Giha: ok See ya later

[6:59] Rocky Sittingbull: have fun!

[659] Linguini Mexicola: cheers

[6:59] ffd2 Cyberstar: bye

[6:59] Prospero Frobozz is going to go compressed air at GRAPE power supplies

[659] Prospero Frobozz: *go blow compressed air

[6:59] Kerry Giha: Sounds like making wine :)

[6:59] Prospero Frobozz: heh

[6:59] Prospero Frobozz: if only

[6:59] Prospero Frobozz: see y'all later