Welcome to MilkyWay@home

Posts by Eric

1) Message boards : News : Visualization/Screensaver Work (Message 38608)
Posted 11 Apr 2010 by Eric

I'm sorry to say that the screensaver work is stalled currently, but I can tell you why.

Lately, we've been putting a lot of time into getting some n-body simulation code running on BOINC. You'll hear about it when it's finished. This has proven non-trivial, since the code base we have to use (the NEMO Stellar Dynamics Toolbox) is very linux specific but needs to be ported to run on the many architectures that BOINC supports. Even if we were to only run it on linux, it's proving more and more complicated as we discover some interesting things, e.g. it compiles and runs things using system("make whatever whatever") during program execution. It's pretty hairy at some spots.

Travis, as you can tell, has been very busy for awhile, so when this came up (a week after the screensaver idea did), we decided I should take on this project since he's pretty busy balancing the world on his shoulders :)

We didn't mean to mislead anyone or create a little instant hype with this post a few months back, and if you feel that way we apologize. We've just reallocated resources given a change in goals and this n-body stuff has taken much longer than expected. The complexity of the code base is greater than average and thus greater than I expected.

I'm leaving when the summer starts for a software gig in Boston, but as I understand another student will be hired full time. I can't promise that he'll be working on the screensaver, but from my understanding once this n-body stuff is finished, the screensaver will once again be a top priority. I'm not in a position to guarantee that that's the next step in the course of the project, though.

I hope that clarifies things well for you - I appreciate your interest in the project, and we're doing our best with the man-hours we have.

2) Message boards : News : Visualization/Screensaver Work (Message 37576)
Posted 19 Mar 2010 by Eric
This is just a post to let everyone know that this visualization project is still ongoing. In truth, I haven't started on it beyond this initial design phase, as I've been working on something else that you'll hear about at some point. That's very vague, but I don't want to give a time estimate or any details since I can't guarantee much accuracy on either front.

Stay tuned.

3) Message boards : News : Visualization/Screensaver Work (Message 36811)
Posted 27 Feb 2010 by Eric
Here are some details on the visualization:

Zoomed out view: This view will show the spiral of the Milk Way galaxy along with highlighted wedges which are currently computing. I'm not sure how much each of you understand regarding the science, but the wedges are very large and easily visible from a galaxy-wide view. My initial post in this thread mentioned showing the "best current model" of the galaxy - I didn't word that very well, as, truthfully, in a zoomed out view where you could (potentially) see over a million stars, a perfectly accurate scientific view is visually indiscernible from a randomly generated star field. Not to mention, theres no need to require more computation or data transfer than is necessary. The overall view won't be based on specific data points from the project, BUT, the wedges will be accurate to the project data.

Zoomed in view: This is the more interesting view, which will show the current work being done. The screen will zoom into a wedge, and you can think of a wedge as a pizza slice - they have a large surface area along one plane, but are very thin. As such, this view will be orthogonal to the face of the main plane (the top of the pizza slice). As each star in the wedge is processed, it will be shown in the wedge. Thus, visually, you will see the wedge filling up with stars.

Now, in reality, there are 100,000 stars in a wedge and showing each one individually would just lead to a mess of color. As such, a sort of heat map will be shown, where red may indicate a high density of stars and blue indicates a low density. After these computations are finished, we then determine what stars are part of the background and which are part of the star stream. At this point, the background will fade and the stream will be shown in a brighter color in the foreground.

That's the basic gist of things - I didn't go into deep deep detail, e.g. there can be multiple streams per wedge which would be dealt with by coloring each stream differently. Hopefully this gives you a good idea of where the visualization is headed.
4) Message boards : News : Visualization/Screensaver Work (Message 36757)
Posted 24 Feb 2010 by Eric
Thanks for all the input guys - I've been silent for a little while but it's all very useful. We have meetings today and tomorrow at which I'm going to pin down the details on this, and I'll let you know what's decided. It'll be an iterative process, of course, and intermediate releases will come out whenever possible.

We've gotten some emails asking about a visualization - there currently is nothing, hence this thread. Sit tight, there'll be something out. I'm an undergrad/grad student (long story, just search "coterminal degree rpi" on google) and so am doing this in addition to a full class load - it won't be coming out as quickly as if I was working 40 hours/week, since I won't be.

I would LOVE to see the Idea No2!
But the other would be great too!
Is there a need to help you with this extension?

We always love your help :) I need to consult with the rest of the guys to get details on how they handle this usually.

5) Message boards : News : Visualization/Screensaver Work (Message 36590)
Posted 16 Feb 2010 by Eric

I would like either #1 or #2. but since most of the respondents have stated their desire for screensaver #2, I will support Item #2, as long as there is, at the zoomed-in level, the same data that is present in #1, and there should be a preference to determine whether it zooms in and how long it stays at either level, so both groups can be satisfied. It seems like the best of both worlds. Also, it would be really cool to have a LCARS-style interface : )

As it develops further I'll share what customization options we're considering. I'm going to try for as much detail as possible - Idea 2 will require time spent on a large view as well as a more detailed close-up, so I wouldn't say it will have detail identical to Idea 1. Even so, you get diminishing returns visually as you add more and more detail, so I think you'll be happy with the results at both the zoomed-out and zoomed-in levels.

A few of you seem to like this LCARS interface...I'll have to look into that.
6) Message boards : News : Visualization/Screensaver Work (Message 36549)
Posted 14 Feb 2010 by Eric
My choice is idea 2. I imagine a kind of Star Trek command screens where continuous data streams appear.

Oooh, you mean the LCARS interface look?

Only with appropriate sound effects!!

Haha, yeah, this is what we're looking for...sound effects might be a little far fetched, but visual ideas you guys want. This is really half about science and half about making things look interesting, and now's the time to throw around your eye candy ideas.

And please, show us your beta developments !

We'll keep you updated.
7) Message boards : News : Visualization/Screensaver Work (Message 36523)
Posted 14 Feb 2010 by Eric
Yeah, the whole GPU thing is certainly a consideration. The visualization would be opt-out so that power users could choose to use the full power of their GPUs for computation.
8) Message boards : News : Visualization/Screensaver Work (Message 36490)
Posted 13 Feb 2010 by Eric
Hey everyone,

I've joined the MilkyWay@Home project for the semester and have heard some rumblings of people wanting some sort of visualization, or screen saver, dare I say :-). We've come up with a few ideas and would like input on which sounds the most appealing to you.

The ideas are ordered from showing the most scientific data to the least. In other words, Idea 1 will give a very clear visualization of what the work units are doing, whereas Idea 3 is more eye candy with fewer details on the science. I want to know which end of the nitty-gritty/eye candy spectrum you prefer.

Idea 1: A zoomed in view of a wedge, displaying the streams being computed for the wedge with indications of progress so far, color-coding to indicate what stars are in what stream, etc. A very good view of what really is being computed, but not much of an overall picture.

Idea 2: An alternating view of the galaxy that starts zoomed out (showing the current best model of the galaxy), followed by a zoom or similar transfer to showing the currently computing wedge. These views would alternate at some specified time interval.

Idea 3: A constant zoomed out view of the galaxy with maybe some small pictures on the sides showing a currently computing wedge and some sort of progress indicator for current work units.

Obviously it would be ideal to incorporate aspects of all different views, but given time constraints, I'd like to focus my efforts on the most popular idea. I encourage you to throw out your own modifications on the ideas and we'll hash through this.

Let 'em rip

©2022 Astroinformatics Group