MODEST meeting 2009-10-24
From Micasim
Contents |
Summary
Preliminary foo-fa
Scilands Move
Piet and Rob kibbitzed about the recent move of the SciLands (the group of regions that includes MICA's StellaNova).
Real Names
Rob has set up some groups with roles that will allow people to have their real names floating over their head in the place of their group titles. If you're a MICA professional member and want to have your real name floating over your head, please e-mail him.
Hydro in AMUSE
Background
Inti started with a background of AMUSE. AMUSE is intended to be MUSE 2.0, a package to connect codes for doing astrophysical simulations using different physics. Today he'll talk about hydro codes to do the ISM or stellar collisions or the like.
Currently, it's possible to couple particle hydro codes, since they're similar in their needs to the N-body codes that are already implemented in AMUSE.
There was some discussion of timestepping; as with all AMUSE codes, you give the code a target time, and it does however many steps it needs to advance its simulation up to that time.
Gravity from other simulations into Hydro & Efficiency
There were questions about specifying external gravity. If a code does self-gravitating fluids, it takes into account the gravity of the fluid it's simulating, but we need a way to tell the code to also include the "external" gravitational field from the particles being simulated by a joined N-body code. Inti says that in the interfaces there are methods defined along the lines of get_gravity_at_point. That will get the gravitational field at a given point from one code, which can presumably then be passed to another code. There was some discussion about the efficiency of this. There will be overhead in pulling that data from one code into the python layer and then pushing it into the next code, as is how things are done in AMUSE. How bad the overhead is will depend on the problem. If there's a good separation of timescales, so that there are a lot of timesteps and a lot of computation between this pulling and pushing, the overhead won't be so bad. (This is where AMUSE is the most efficient in general.) Steve worries, however, that if you consider the gas in a cluster, the dynamical timescales for the gas and stars will be similar. In any event, this would be something that's good to profile.
There was some question about using shared memory or other IPC tricks. The way AMUSE is designed right now (with "legacy" codes being as loosely coupled as possible) makes this very tough. Steve also asked about using some sort of extrapolation scheme for the external forces, as in Ahmad Cohen.
Status of the current SPH interface
Inti is currently building the interface by adding an interface function when he needs something. He is working with his SPH code (I think -- is it Inti's code?). The interface he's been building is not checked into the SVN yet.
Other Hydro Interfaces
The interface may be different for a mesh hydro code. Something like get_gravity should be generic. Jason Maron said that his code is also particle-like, so should have the same interface as an SPH code.
Piet suggests having an utterly simple hydro code, mostly to test the interface with a minimum of other baggage; perhaps a sticky particle code, or even something simpler. Arjen says that they are looking at some very simple codes.
Meeting at Drexel
We discussed having an in-person day of meetings at Drexel University, and settled on November 18 as the day. We don't know if anybody form Leiden can be there, but we certainly hope at least one or two of them can.
Chat Log
[2009/10/14 6:00] Pema Pera: Good morning/afternoon/everning everybody!
[2009/10/14 6:00] Linguini Mexicola: hi there
[2009/10/14 6:00] Makino Magic: Hi
[2009/10/14 6:01] Prospero Frobozz: Howdy all
[2009/10/14 6:01] So Kuu: heya
[2009/10/14 6:01] Prospero Frobozz: Linguini Mexicola : do we know you by another name?
[2009/10/14 6:01] Prospero Frobozz: Let's wait a couple of minutes before we start
[2009/10/14 6:01] Lagrange Euler: Mornin' all
[2009/10/14 6:01] Prospero Frobozz: I'm hoping and expecting that Inti Pelupessy will come today
[2009/10/14 6:01] Linguini Mexicola: thats me
[2009/10/14 6:02] Pema Pera: Rob, I heard that the Scilands have moved (again, I believe) -- how unsettling! Do you know why?
[2009/10/14 6:02] Prospero Frobozz: Yeah -- they moved because they wanted to add a new island, but there wasn't room off the edge.
[2009/10/14 6:02] Prospero Frobozz: As such, they had to move to a clearer spot on the grid.
[2009/10/14 6:02] Pema Pera: but didn't they move a year ago or so?
[2009/10/14 6:02] Prospero Frobozz: Yeah, for the same reason
[2009/10/14 6:03] Pema Pera: so we're going to be forced to move once a year???
[2009/10/14 6:03] Prospero Frobozz: Probably; who knows
[2009/10/14 6:03] Pema Pera: glad to hear that our landmarks still work
[2009/10/14 6:03] Prospero Frobozz: I don't know if Linden has a spot anywhere that's open enough to prevent any forseeable future moves
[2009/10/14 6:03] Pema Pera: maybe we need a non-Euclidean metric!
[2009/10/14 6:04] Prospero Frobozz: Heh
[2009/10/14 6:04] Prospero Frobozz: Well, I think the real solution is Diva Canto's solution -- stop putting everything on a single grid.
[2009/10/14 6:04] Lagrange Euler: I was thinking the same thing -- sounds like a problem in topology
[2009/10/14 6:04] Pema Pera: yes
[2009/10/14 6:04] Prospero Frobozz: And have interoperability swo that you can jump your account from grid to grid
[2009/10/14 6:04] Prospero Frobozz: So Scilands can be their own grid
[2009/10/14 6:04] Pema Pera: yes, exactly
[2009/10/14 6:04] Prospero Frobozz: Buuuuut, that's OpenSim only, and doesn't even function fully as we'd want there. But itw ill.
[2009/10/14 6:04] Pema Pera: we never swim to neighboring islands :)
[2009/10/14 6:05] Prospero Frobozz: And, someday, I suspect that will be the defacto standard; if Linden doesn't get no board with it, SL will become a shrinking walled garden cut off from the "world" like CompuServe was
[2009/10/14 6:05] Prospero Frobozz: But that's still a couple of years down the line, I think
[2009/10/14 6:05] Prospero Frobozz: Anyway, one moemnt....
[2009/10/14 6:06] Linguini Mexicola: so - shall we start?
[2009/10/14 6:06] Lagrange Euler: yes!
[2009/10/14 6:06] Prospero Frobozz: I want to do one admi thing first
[2009/10/14 6:06] Prospero Frobozz: Inti, I just sent you a group invitation
[2009/10/14 6:06] Prospero Frobozz: As many of you noticed, I invited lots of you to groups
[2009/10/14 6:06] Prospero Frobozz: What I did was create a couple of groups, "MICA Realnames 1" and "2"
[2009/10/14 6:07] Prospero Frobozz: Then I made group roles that have the group titles that are your eral names.
[2009/10/14 6:07] Prospero Frobozz: This will allow you to have your real name floating over your head
[2009/10/14 6:07] Prospero Frobozz: Piet -- I put your name in the main "MICA" group-- there's a Piet Hut role you can use there.
[2009/10/14 6:07] Prospero Frobozz: If you ever want to have your real name floating over your head, just make the realnames group your active group, and then make sure that's selected as the group title
[2009/10/14 6:07] Prospero Frobozz: Will -- I'll invite you
[2009/10/14 6:07] Eamu Godenot: Thanks, Pros.
[2009/10/14 6:08] Lagrange Euler: That's very handy. Thanks Rob!
[2009/10/14 6:08] Pema Pera: how do I activate the "Piet Hut" role, Rob?
[2009/10/14 6:08] Prospero Frobozz: Piet : Edit, Groups
[2009/10/14 6:08] Prospero Frobozz: Select the MICA group from the list
[2009/10/14 6:08] Prospero Frobozz: click "Info"
[2009/10/14 6:09] Prospero Frobozz: There's a dropdown there, "My Active Title"
[2009/10/14 6:09] Prospero Frobozz: Select Piet Hut from the dropdown
[2009/10/14 6:09] Prospero Frobozz: There you go
[2009/10/14 6:09] Pema Pera: thank yuo!
[2009/10/14 6:09] Prospero Frobozz: Does anybody else want a real name that I haven't invited yet?
[2009/10/14 6:09]Prospero Frobozz: Anyway, that was the admin thing.
[2009/10/14 6:10] Prospero Frobozz: Inti, you're in charge now -- he's going to tell us what the current status of the Hydro interface in AMUSE is.
[2009/10/14 6:10] Linguini Mexicola: allright
[2009/10/14 6:10] Prospero Frobozz is sad that Jason Maron isn't here
[2009/10/14 6:10] Linguini Mexicola: So let me start by giving you a quick background overview of AMUSE
[2009/10/14 6:10] Linguini Mexicola: I guess most people here would have heart about MUSE
[2009/10/14 6:11] Linguini Mexicola: or the talk about AMUSE that Arjen and me gave some time ago
[2009/10/14 6:11] Linguini Mexicola: so the basic idea of AMUSE is to be MUSE 2.0
[2009/10/14 6:12] Linguini Mexicola: ie a package to connect different codes for doing astrophysical simulations using different physics
[2009/10/14 6:12] Linguini Mexicola: One of the important ingredients will be a interface to connect hydrodynamic codes
[2009/10/14 6:13] Ico Telling: Hi sorry for being late
[2009/10/14 6:13] Icarusfactor Scientist: cool
[2009/10/14 6:13] Linguini Mexicola: to do model the ISM or stellar collisions and the like
[2009/10/14 6:13] Linguini Mexicola: so everything that goes beyond stellar dynamics and stellar evolution, the two
[2009/10/14 6:14] Linguini Mexicola: ingredients that so far have been most succesfully coupled in MUSE
[2009/10/14 6:15] Linguini Mexicola: Now, ofcourse hydrodynamics is a different beast altogether to couple
[2009/10/14 6:15] Linguini Mexicola: and the current status of hydro is such that it is possible to couple
[2009/10/14 6:15] Linguini Mexicola: particle hydro codes - this is because particle hydrocodes are close
[2009/10/14 6:16] Linguini Mexicola: to ordinary gravitational dynamics in the code setup; practically speaking it
[2009/10/14 6:16] Linguini Mexicola: suffices to give the particles an internal energy or temperature in the interface
[2009/10/14 6:16] Linguini Mexicola: and then let the code do its thing.
[2009/10/14 6:17] Eamu Godenot: Who controls the timestepping in this scenario?
[2009/10/14 6:17] Linguini Mexicola: as usual, the code does
[2009/10/14 6:17] Linguini Mexicola: you give a target time and the details of timestepping are hidden
[2009/10/14 6:17] Prospero Frobozz: don't you have to step the n-body and the hydro separately?
[2009/10/14 6:17] Lagrange Euler: The time stepping should be controlled by the top level scheduler.
[2009/10/14 6:18] Lagrange Euler: Every module has a notion of its "next" time, and that should be all the scheduler needs
[2009/10/14 6:18] Prospero Frobozz: That is, jump into the n-body code, step a certain number of steps, then jump into the hdyro code, step a certain number of steps, etc?
[2009/10/14 6:18] Eamu Godenot: And that polls the N-body and hydro codes to see who's next, so to speak?
[2009/10/14 6:18] Lagrange Euler: More or less.
[2009/10/14 6:19] Linguini Mexicola: well yes - so a code will have its locally adaptive timestep criterion, but all that is
[2009/10/14 6:19] Linguini Mexicola: hidden
[2009/10/14 6:19] Prospero Frobozz: do typical N-body and SPH codes have a way of specifcying "external" sources of gravity? so that the gravity of the SPH partciles can affect the N-body particles, and vice versa
[2009/10/14 6:19] Prospero Frobozz: Probably N-body affecting SPH being more important
[2009/10/14 6:19] Lagrange Euler: Inti, are you talking about coupling codes with global gravity, or something more like collisions in an N-body context?
[2009/10/14 6:19] Linguini Mexicola: that is a good question:
[2009/10/14 6:20] Lagrange Euler: They are very different problems
[2009/10/14 6:20] Linguini Mexicola: normally SPH codes are geared towards self gravitating fluids
[2009/10/14 6:20] Lagrange Euler: (from a dataflow point of view)
[2009/10/14 6:20] Linguini Mexicola: and have their own gravity solver - so in AMUSE
[2009/10/14 6:20] Linguini Mexicola: these will be preserved (with the option of not using them)
[2009/10/14 6:21] Lagrange Euler: But if the gas is intracluster, then both the stars and the gas feel each others potential.
[2009/10/14 6:21] Linguini Mexicola: no if you want to couple the gravity of different modules this will be possible because we have defined in the
[2009/10/14 6:21] Linguini Mexicola: interface a set of function like get_gravity _at_point
[2009/10/14 6:22] Linguini Mexicola: that - if the code provides that possibility - returns the local gravity field
[2009/10/14 6:22] Lagrange Euler: How is that implemented?
[2009/10/14 6:22] Prospero Frobozz: Hurm -- but that would require you to call that a huge number of times. That is, if you want to take the N-body gravity into the SPH code, you'd have to call it once for the position of each SPH particle, and then feed it back into the SPH code. woudln't that be a lot of overhead for a lot of particles?
[2009/10/14 6:23] Prospero Frobozz: (perhaps not; perhaps the equivalent of one timestep, although there's the overhead of going in and out of the code)
[2009/10/14 6:23] Lagrange Euler: Exactly. That sounds slow!
[2009/10/14 6:23] Prospero Frobozz waves to Jason
[2009/10/14 6:23] Linguini Mexicola: that depends: in a bridge scheme like coupling you only need to couple
[2009/10/14 6:23] Linguini Mexicola: every so many timesteps
[2009/10/14 6:23] Makino Magic: You coud extend the interface so that it receives a number of positions and return many forces.
[2009/10/14 6:23] Prospero Frobozz: Right
[2009/10/14 6:24] Linguini Mexicola: the alternative of having a function that can be called was
[2009/10/14 6:24] Prospero Frobozz: Is that functionality that typical codes have?
[2009/10/14 6:24] Linguini Mexicola: explored using the f2py interface using callbacks
[2009/10/14 6:24] Prospero Frobozz: Or is that something we're likely to have to implement inside the legacy codes to get the AMUSE interface working?
[2009/10/14 6:24] Lagrange Euler: But you are still sending the data back and forth between independent parallel modules
[2009/10/14 6:24] Linguini Mexicola: that worked, but we have moved to MPI interface and then this becomes , well not impossible maybe,
[2009/10/14 6:25] Linguini Mexicola: but more difficult and much slower
[2009/10/14 6:25] Lagrange Euler: The main advantage of the old swig/f2py interface was that there was a common address space, so there was the possibility of passing pointers. Now you have to do MPI send/recv, right?
[2009/10/14 6:25] Linguini Mexicola: yes, exactly
[2009/10/14 6:26] Lagrange Euler: How about shared memory, or other IPC tricks?
[2009/10/14 6:26] Prospero Frobozz: Inti, when you say "bridge", that's codes communicating directly with each other? Right now, I believe, all that's implemented is communicating from the AMUSE framework layer down to the code, right?
[2009/10/14 6:26] Linguini Mexicola: At the moment, for the bridge scheme the large interval kicks are done in python space
[2009/10/14 6:27] Linguini Mexicola: with the function call down to the module for the gravity call
[2009/10/14 6:27] Prospero Frobozz: OK, so it would be the python space getting the info from one code, and then passing that info to the other code, rather than setting up direct communication, yes?
[2009/10/14 6:27] Linguini Mexicola: jep
[2009/10/14 6:28] Linguini Mexicola: to get back to the hydro codes:
[2009/10/14 6:28] Lagrange Euler: Is there a possibility of some sort of extrapolation scheme for the external forces, as in Ahmad-Cohen?
[2009/10/14 6:29] Prospero Frobozz: It would be interesting to do some profiling to see how efficient that was.
[2009/10/14 6:29] Prospero Frobozz: goign through the python
[2009/10/14 6:29] Prospero Frobozz: is compared to various different timestep sizes
[2009/10/14 6:29] Prospero Frobozz: s/is/as/
[2009/10/14 6:29] Linguini Mexicola: it should be reasonably efficient for problems where
[2009/10/14 6:30] Linguini Mexicola: there is seperation of scales
[2009/10/14 6:30] Prospero Frobozz: Yeah
[2009/10/14 6:30] Prospero Frobozz: That's when AMUSE in general is the most efficient
[2009/10/14 6:31] Lagrange Euler: Suppose you take the gas in cluster problem. Is there a clear separation of scales there? I'd have guessed that the dynamical time scales for stars and gas would be the same.
[2009/10/14 6:31] Linguini Mexicola: that depends on the sound speed of the gas
[2009/10/14 6:31] Lagrange Euler: Yes, but if everything is in ~virial equiibrium
[2009/10/14 6:32] Linguini Mexicola: I think outflows from evolved stars will be interesting
[2009/10/14 6:32] Lagrange Euler: Yes, I agree
[2009/10/14 6:33] Linguini Mexicola: In any case, this related to the SPH
[2009/10/14 6:33] Linguini Mexicola: but ofcourse there are a lot of problems where different
[2009/10/14 6:33] Linguini Mexicola: gas dynamics methods are more appropiate
[2009/10/14 6:33] Prospero Frobozz: Is the SPH amuse interface already defined? Have you built the interface for your SPH code?
[2009/10/14 6:33] Linguini Mexicola: and one may want to do grid hydro
[2009/10/14 6:34] Linguini Mexicola: the SPH interface is WIP (meaning I am doing the simulations and adding functionality
[2009/10/14 6:34] Linguini Mexicola: whenever I need something
[2009/10/14 6:34] Linguini Mexicola: eventually the interface will have to be refactored and simplified
[2009/10/14 6:34] Linguini Mexicola: )
[2009/10/14 6:35] Prospero Frobozz: Heh, OK
[2009/10/14 6:35] Lagrange Euler: So you have implemented the "get_gravity()" functions?
[2009/10/14 6:35] Prospero Frobozz: (Is it checked in?)
[2009/10/14 6:35] Linguini Mexicola: no, not yet checked in
[2009/10/14 6:36] Prospero Frobozz: I'm guessing that the interface right now is pretty SPH specific, yes? Rather than being a general Hydro interface
[2009/10/14 6:36] Linguini Mexicola: the get_gravity I think may be - Arjen?
[2009/10/14 6:36] Prospero Frobozz: Jason : your code is a mesh code, not a SPH code, isn't it?
[2009/10/14 6:36] Linguini Mexicola: (Arjen is waking up)
[2009/10/14 6:37] Remy Vespucciano: Yes, nice lunch :-)
[2009/10/14 6:37] Corwin Taurus: the interface is identical to SPH
[2009/10/14 6:37] Prospero Frobozz: Heh -- middle of the afternoon naptime over there?
[2009/10/14 6:37] Corwin Taurus: comoving particles
[2009/10/14 6:37] Prospero Frobozz: Corwin : Ah, OK... so it's not a mesh code
[2009/10/14 6:37] Prospero Frobozz: Inti, Arjen, do you guys have any mesh codes on your radar for what you'll think about with AMUSE first?
[2009/10/14 6:37] Prospero Frobozz: It probably makes sense to focus on SPH first just because it's easier, and it *does* get Hydro in!
[2009/10/14 6:38] Remy Vespucciano: The get_gravity is implemted for the phigraphe nbody code
[2009/10/14 6:38] Prospero Frobozz: SPH or other particle type codes like Jason's
[2009/10/14 6:39] Linguini Mexicola: It is our goal to have grid hydro codes in,
[2009/10/14 6:39] Lagrange Euler: Have you profiled the combined phiGRAPE/SPH code to see what the overhead is?
[2009/10/14 6:39] Linguini Mexicola: no
[2009/10/14 6:39] Lagrange Euler: It would be a worthwhile exercise
[2009/10/14 6:40] Pema Pera: have you thought about having one utterly simple hydro code, mostly to test the interface with a minimum of other baggage?
[2009/10/14 6:40] Pema Pera: perhaps a sticky particle code
[2009/10/14 6:40] Pema Pera: or even something simpler
[2009/10/14 6:40] Linguini Mexicola: In any case, we had a discussion during last weeks workshop about grid codes
[2009/10/14 6:40] Remy Vespucciano: Yes, we are looking at some very simple codes
[2009/10/14 6:41] Prospero Frobozz: It would be nice if there could be a "usable" overlap between grid and SPH type codes, so that (at least for some problems) would could plug and unplug different hydro modules to see how different the results are. I don't know how practical that will really be.
[2009/10/14 6:41] Remy Vespucciano: collisions (like sticky speres) are part of hydro
[2009/10/14 6:41] Linguini Mexicola: (and we had talks with developers of grid codes about importing their codes)
[2009/10/14 6:41] Lagrange Euler: In principle, the I/O issue is the same as when coupling two N-body codes -- stars and dark matter, perhaps.
[2009/10/14 6:41] Prospero Frobozz: That is, a usable overlap between the itnerface for the two different types of codesw
[2009/10/14 6:42] Linguini Mexicola: Grid codes have a number of issues which makes them stand apart
[2009/10/14 6:42] Pema Pera: or even two parts in one code, Steve, like Ahmad-Cohen
[2009/10/14 6:42] Prospero Frobozz: Steve : although when you're linking two different N-body codes, it's probably because you either are on two very different timescales -- big system and embedded cluster -- or because you have two regimes -- loosely coupled and tightly coupled -- you're swapping between. So you don't have to swap back and forth all that often. With hydro and N-body, I can see wanting to keep the two timescales closer.
[2009/10/14 6:42] Linguini Mexicola: and it will be a bit more involved to have them interchangeable with SPH type codes
[2009/10/14 6:42] Lagrange Euler: Yes, but you want to separate the codes into two modules to test the parallel structure
[2009/10/14 6:42] Lagrange Euler: (Sorry, two threads... My comment was to Piet)
[2009/10/14 6:43] Linguini Mexicola: ok
[2009/10/14 6:43] Prospero Frobozz: Inti, Arjen : is there anything others of us here can do to help with the hydro (at least SPH) interafcing work?
[2009/10/14 6:44] Lagrange Euler: Presumably the get_gravity interface could be implemented with any code, grid or not, if you aren't too concerend about efficiency. Do you have a grid code in mind?
[2009/10/14 6:44] Remy Vespucciano: If you know of a really simple hydro code....
[2009/10/14 6:45] Linguini Mexicola: I wouldn't on the spot think of something....
[2009/10/14 6:46] Prospero Frobozz: Ok; I may dig a bit to see if I can find a freely available "really simple hydro code"
[2009/10/14 6:46] Linguini Mexicola: Maybe you have suggestions of what would be the most interesting test
[2009/10/14 6:46] Linguini Mexicola: / application
[2009/10/14 6:46] Linguini Mexicola: that you would want to have working
[2009/10/14 6:46] Prospero Frobozz: Well, my own personal interest is in two colliding galaxies, wanting to see the gas get shocked and compressed so that it forms nuclear starbursts and starbursts elsewhere in the system :)
[2009/10/14 6:47] Lagrange Euler: You need a grid code with a Poisson solver. I have a flash expert in house. Not a simple code, but one capable of providing the information the interface needs...
[2009/10/14 6:47] Prospero Frobozz: That's probably a fairly hard problem, as the timescales are probalby not easily separated between N-body and hydro
[2009/10/14 6:47] Linguini Mexicola: flash has licensing issues (very strict)
[2009/10/14 6:48] Pema Pera: if I can briefly start a different thread: shall we pick a day in November to all get together at Drexel to talk about these issues in person, where "all" is whoever is able to drop by for a day? -- it may be easier to pick a day while we're all online now, rather than doing it by email
[2009/10/14 6:48] Linguini Mexicola: wouldn't use that for AMUSE
[2009/10/14 6:48] Prospero Frobozz: Yeah... that it's partially DoD code makes it tough, and probably Flash and AMUSE can't be linked. But, the Flash developer may have the knowledge that will tell you what the interface iwll need to look like for a general grid code
[2009/10/14 6:48] Lagrange Euler: No, only as provate proof of concept.
[2009/10/14 6:48] Prospero Frobozz: Piet : Yes, i think that's a good idea
[2009/10/14 6:48] Lagrange Euler: I think it is DoE
[2009/10/14 6:48] Pema Pera: How about 11/11 ?
[2009/10/14 6:49] Prospero Frobozz: I have to admit that myself, my life is very up int he air... I will be interviewing at the end of October for this CITA job, and if I get it, I may be moving
[2009/10/14 6:49] Prospero Frobozz: would anybody from leiden be able to come out to Drexel?
[2009/10/14 6:49] Prospero Frobozz: I don't know how much travel money is in Simon's grant
[2009/10/14 6:49] Ico Telling: I would prefer after 11/15
[2009/10/14 6:49] Pema Pera: 11/18 ?
[2009/10/14 6:49] Ico Telling: That's ok
[2009/10/14 6:49] Prospero Frobozz: Re: 11/11, I have to check with my wife, as my parents-in-law are coming. Also, the week of Thanksgiving is not good for me, as my own mother is probably coming....
[2009/10/14 6:50] Ico Telling: is 11/18 in the Thanksgiving week?
[2009/10/14 6:50] Prospero Frobozz: So 11/18 would be good
[2009/10/14 6:50] Pema Pera: no
[2009/10/14 6:50] Prospero Frobozz: Thanksgiving is 11/26
[2009/10/14 6:50] Ico Telling: That's what I thought so 11/18 would be good
[2009/10/14 6:50] Lagrange Euler: Sometime after proposal season ends (>11/15)
[2009/10/14 6:50] Linguini Mexicola: We just passed a deadline for travel grants, so it would be more difficult
[2009/10/14 6:50] Linguini Mexicola: to come from leiden
[2009/10/14 6:50] Pema Pera: Jun, would you happen to pass through the US in November, by any chance?
[2009/10/14 6:51] Lagrange Euler: 11/18 would be OK for me, apart from my 10 am class...
[2009/10/14 6:51] Lagrange Euler: speaking of which...
[2009/10/14 6:51] Makino Magic: I'll visiting SLAC for Nov 29-Dec 2
[2009/10/14 6:51] Prospero Frobozz: heh
[2009/10/14 6:51] Prospero Frobozz: We can all go and sit in on Steve's class -- what are you teaching?
[2009/10/14 6:51] Lagrange Euler: Well, that's "near" Philadelphia
[2009/10/14 6:51] Pema Pera: ah, could you come to Drexel before your SLAC visit perhaps, Jun?
[2009/10/14 6:51] Lagrange Euler: Mathematical Physics
[2009/10/14 6:51] Prospero Frobozz: SLAC is near philadelphia when compared to (say) M31
[2009/10/14 6:51] Lagrange Euler: (graduate)
[2009/10/14 6:52] Lagrange Euler: or Tokyo
[2009/10/14 6:52] Prospero Frobozz: We might even learn something! Most likely we'd have no idea what you were talking about
[2009/10/14 6:52] Lagrange Euler: I certainly hope not!
[2009/10/14 6:52] Prospero Frobozz: "What are all those d's you keep putting in front of variables? And what's witht he upside-down triangles?"
[2009/10/14 6:52] Makino Magic: Hmm, not completely impossible...
[2009/10/14 6:52] Lagrange Euler: Everyone here has taken (and probably taught) this course...
[2009/10/14 6:52] Pema Pera: but that would unfortunately be around thanksgiving . . . .
[2009/10/14 6:53] Prospero Frobozz: Unless Jun can come a week early....
[2009/10/14 6:53] Pema Pera: like 11/28 or so, which would be a Saturday, would that be impossible?
[2009/10/14 6:53] Lagrange Euler: Or a week late?
[2009/10/14 6:53] Pema Pera: a week late I will be in Japan :)
[2009/10/14 6:53] Prospero Frobozz: If we do this, and if nobody from Ledien makes it, we should try to do a "mixed reality" session in SL/RL to see if we can integrate everybody.
[2009/10/14 6:53] Lagrange Euler: Weekends aren't going to work well for me.
[2009/10/14 6:53] Prospero Frobozz: 11/28 wouldn't be good for me, I fear.
[2009/10/14 6:54] Prospero Frobozz: anything within a couple of days of Thansgiving probably doesn't work for me.
[2009/10/14 6:54] Makino Magic: Sorry a week later is out, since I have to be in Mitaka on Dec 7
[2009/10/14 6:54] Lagrange Euler: My classes are over on December 5, so I am ~free after that.
[2009/10/14 6:54] Pema Pera: Jun, would Monday, work, 11/23 ?
[2009/10/14 6:54] Prospero Frobozz: I could probably do mid-December, modulo whether or not I've moved....
[2009/10/14 6:54] Pema Pera: I'll be in Japan in December
[2009/10/14 6:54] Makino Magic: I have to be in Kobe on Nov 25...
[2009/10/14 6:55] Lagrange Euler: I'm really sorry, but I have to go. Can we continue this by e-mail?
[2009/10/14 6:55] Lagrange Euler: Bye all.
[2009/10/14 6:55] Pema Pera: well, it seems that November 18 is still the best then
[2009/10/14 6:56] Ico Telling: yes I think 11/18 is the best option
[2009/10/14 6:57] Remy Vespucciano: I'll look around if one of us can make it in person 11/18. If not I hope the
[2009/10/14 6:57] Remy Vespucciano: real / second life
[2009/10/14 6:57] Remy Vespucciano: is workable
[2009/10/14 6:57] Pema Pera: yes, we can probably set up something
[2009/10/14 6:57] Prospero Frobozz: Yeah, -- the time zones will be the biggest issue
[2009/10/14 6:57] Pema Pera: yes, with Steve teaching at 10
[2009/10/14 6:57] Prospero Frobozz: But I think we can set up some sort of mixed reality thing. It would be a good experiment, anyway!
[2009/10/14 6:57] Pema Pera: by the time he is free, it will be the end of the afternoon in Holland
[2009/10/14 6:57] Prospero Frobozz: But I'd love it if either or both of you guys could be there.
[2009/10/14 6:58] Pema Pera: yes
[2009/10/14 6:58] Pema Pera: Thank you, Inti, I hope you don't mind the many interruptions today!
[2009/10/14 6:58] Pema Pera: we're a rowdy bunch :)
[2009/10/14 6:58] Linguini Mexicola: yes, you were - I leave the rest of the stuff to some other time!
[2009/10/14 6:59] Prospero Frobozz: OK
[2009/10/14 6:59] Prospero Frobozz: well, we can continue this discussion next week, if we think it's fruitful
[2009/10/14 6:59] Ico Telling: OK Thanks. shall we continue this next Wed?
[2009/10/14 6:59] Corwin Taurus: what would everyone think about a presentation about our particle code then?
[2009/10/14 6:59] Prospero Frobozz: I'll try and figure out what the topic will be next Monday, in case Things Happen between now and then
[2009/10/14 6:59] Prospero Frobozz: Corwin : that would be good
[2009/10/14 6:59] Prospero Frobozz: Esp. if you can talk about the AMUSE interface as part of it
[2009/10/14 7:00] Ico Telling: Should we let Inty finish next Wed?
[2009/10/14 7:00] Pema Pera: maybe better yes
[2009/10/14 7:00] Ico Telling: Considering all the interruptions...:)
[2009/10/14 7:00] Prospero Frobozz: Yes
[2009/10/14 7:00] Prospero Frobozz: Let's do that first :)
[2009/10/14 7:00] Pema Pera: and Jason the week after
[2009/10/14 7:01] Pema Pera: See you all next week!
[2009/10/14 7:01] Ico Telling: OK see you Wed
[2009/10/14 7:01] Pema Pera: Thanks Inti, and thanks Rob for moderating!
