Welcome to MilkyWay@home

Posts by Nathan

21) Message boards : Number crunching : No new WU (Message 4023)
Posted 9 Jul 2008 by ProfileNathan
Post:
I'm really sorry about this. I'm the only one running searches at the moment and I've been sick the past few days, and the searches I started before the weekend died out before I could start more.

I just started 2 more searches 3710182 and 3711182.

Again, sorry.
22) Message boards : Number crunching : New Searches (Message 3987)
Posted 3 Jul 2008 by ProfileNathan
Post:
Added searches 3703082 and 3704082 to make sure there's enough work over the weekend.
23) Message boards : Number crunching : No new WU (Message 3986)
Posted 3 Jul 2008 by ProfileNathan
Post:
And I'm starting more today to make sure there's enough work to last the weekend.
24) Message boards : Number crunching : No new WU (Message 3985)
Posted 3 Jul 2008 by ProfileNathan
Post:
I apologize if I helped in the lack of WUs, didn't realize that my searches were that close to done.

But I started new ones yesterday.
25) Message boards : Number crunching : New Searches (Message 3980)
Posted 2 Jul 2008 by ProfileNathan
Post:
I just started two new searches (3701082 and 3702082). Hopefully, this will have fixed the couple problems with the results that I was having caused by the limits of some of the parameters being too large.
26) Message boards : MilkyWay@home Science : SDSS Stripe 82 searches (Message 3946)
Posted 27 Jun 2008 by ProfileNathan
Post:
Figured I'd give you guys a heads up on the status of the searches and how they're doing.

In short, things are going great. There are a couple of things I need to get ironed out to make my life easier and to optimize the search space (the BOINC code uses a different optimization algorithm than what I previously used), but just with the two searches I've ran so far I can see that I can reproduce the results that we got in the ApJ paper! I'm not getting quite the accuracy I was getting before, but we expected this due to the optimization method we're using.

At any rate, this is great news! It shows that we can do real science using Milkyway@home and not just use simulated data sets!

Keep up the good work guys, and I'll be continuing to start new searches on new data sets, and I'll keep you updated as to when I start them and what they are.
27) Message boards : Number crunching : Has the nature of the work units changed (Message 3945)
Posted 27 Jun 2008 by ProfileNathan
Post:
Yeah, it's a different data set (this is real observational data) that is about half the size of what you guys are used to crunching. However, this should have very little effect on the timing of the run (ever so slightly faster, but it should be hardly noticeable). I can't see why this would cause the discrepancy in time, but I guess it could be that I'm using slightly different settings than Travis had.
28) Message boards : MilkyWay@home Science : SDSS Stripe 82 searches (Message 3882)
Posted 19 Jun 2008 by ProfileNathan
Post:
I just started a new search today (about 20 minutes ago actually). The initial error caused a lot of bad WUs to need to get worked out and then we had an issue with the assimimilator. However, Travis has told me that things should be back to running smoothly now, and in response I started a search using the stripe 82 data. Hopefully I can get results back quickly, and we can move on to new data sets.
29) Message boards : MilkyWay@home Science : SDSS Stripe 82 searches (Message 3815)
Posted 17 Jun 2008 by ProfileNathan
Post:
What are we searching for?

AIUI we're not searching for anything, but modelling interactions between stars (or clusters or associations thereof) and comparing the results to observations of stellar distribution within our Galaxy.


This is actually incorrect, we're not modeling interactions. We're actually not doing any "modeling" at all. What we are doing is looking for over-densities in the spatial data (positions of stars) that correspond to tidal disruptions of dwarf galaxies, specifically the Sagittarius Dwarf Galaxy (Sgr dSph) in this case.

With this data, the people that actually do do the modeling work will be matching their simulations to OUR RESULTS!

We are actually studying the observational data taken by the SDSS, so when you download a file called stars_82.txt or some similar things you are actually getting a list on coordinates for real objects in the sky!
30) Message boards : MilkyWay@home Science : What's on the agenda? (Message 3814)
Posted 17 Jun 2008 by ProfileNathan
Post:

For anyone that is interested in a complete discussion of what is going on here at Milkyway@home and how the results can be used, you can find a complete discussion in there. I do warn that it's pretty dense, though.


Even though the paper is technical, I suggest you post a summary here that crunchers and the general public can understand.



The paper describes in rigorous detail (and in mathematical detail) the method and models in which we do all of our calculations: a very brief summary is that we use a very small volume to study spatial distributions of stars. This allows us to model the stream debris as a cylinder within some background distribution which are described by a number of parameters. We then use a technique called maximum likelihood to "search" the likelihood "space" and find the values of the parameters that have the highest likelihood. These correspond to the "correct" values for those parameters.

Using this method we were able to reproduce the parameters for a simulated data set, and closely reproduce earlier results on the stripe 82 data that we used. However, we were able to greatly reduce the errors regarding these values and determine new quantities as well. All total we were able to present values for the position, orientation, and size (spatial and number) of the stream and the values that best define the background distribution.

If there's any questions I'll answer them, and there is much more detailed explanation in the paper.
31) Message boards : MilkyWay@home Science : SDSS Stripe 82 searches (Message 3565)
Posted 29 May 2008 by ProfileNathan
Post:
Thanks! I figured you guys would like to know a little more precisely where we're looking. I'll try and get a less ambiguous picture format to show where in relation to the rest of the Milkyway we're working.

We started with this specific piece of data because it is the best studied in regards to the Sagittarius Tidal Stream. Therefore, we would have a way to validate that we were getting the correct results when using non-simulated data. If we would have started with some other set of data, there would be no way to say it was correct, but at the same time not know it was incorrect either. Since we were able to replicate the results seen from previous work with smaller errors we were able to show that the method works and better than previous methods.

This is the first set of non-simulated data for Milkyway@home for the same reason. We need to show that we can replicate the results using BOINC on non-simulated data before we can move on to other data sets.

Once we have the results back from this first stripe, we then want to systematically cover the rest of the area seen in the second figure from my original post with similarly sized wedges. By doing this we'll be able to build up a complete image/map of the tidal stream, instead of just having a tiny piece of it. This will not be trivial though as there is obviously other structures present within the data that we either have to remove or simultaneously fit.
32) Message boards : MilkyWay@home Science : What's on the agenda? (Message 3552)
Posted 28 May 2008 by ProfileNathan
Post:
Thanks for the update, but I don't think this paper qualifies to be there. Our paper wasn't to the point of using the Milkyway@home data yet as we were still in the process of getting it up and running while we did the work in the paper on a cluster. The next paper would definitely work though as a of the work should come from you guys.

Thanks again, and if I am misunderstanding what the qualifications for the papers page is please let me know.
33) Message boards : Number crunching : New searches on observational data (Stripe 82) (Message 3524)
Posted 27 May 2008 by ProfileNathan
Post:
Just started new searches on the data explained here.
34) Message boards : MilkyWay@home Science : SDSS Stripe 82 searches (Message 3523)
Posted 27 May 2008 by ProfileNathan
Post:
I just started a search using data from SDSS stripe 82 which is the set of data we used in the paper we just got accepted to ApJ. This is a well studied piece of data that is along the Celestial Equator in the South Galactic Cap.

The Celestial Equator is an extension of the Earth's equator and defines the Equatorial Coordinate system seen below,

The piece of data can be seen below (the green strip) with respect to the rest of the SDSS survey.


What I'm looking at doing here is replicating the results seen in the paper before I move on to the rest of the data seen in the second image here.
35) Message boards : MilkyWay@home Science : ASTRONOMY PICTURE OF THE DAY™ (Message 3522)
Posted 27 May 2008 by ProfileNathan
Post:
SDSS does an image of the week taken from the SDSS' own data: they can be found here.

The archives of previous images is here.
36) Message boards : MilkyWay@home Science : What's on the agenda? (Message 3521)
Posted 27 May 2008 by ProfileNathan
Post:
First off I'd like to apologize for my horrendous lapse in checking these boards and posting: I've been working hard at getting a paper out and we finally got it accepted into the Astrophysical Journal. I know this is not an excuse, but I hope you can all understand.

You can view/download the the paper here.

For anyone that is interested in a complete discussion of what is going on here at Milkyway@home and how the results can be used, you can find a complete discussion in there. I do warn that it's pretty dense, though.


This publication proves that what we're doing here works and is the best means we have to do these types of studies. This is good and means that we can move on to doing science as opposed to proving that the method we've set up is valid.

In the coming days/weeks I'm going to start running searches using actual astronomical data taken by the SDSS. Up to this point Milkyway@home has been running on simulated data in order to get the infrastructure up and running and to determine the means and methods to make it work.

I'll keep you all updated as to when I start these searches and the results of them. Also, I'll try to post pictures that highlight the area that I'm actually looking at with a given search.

Again, I apologize for my lapse earlier, and I'll gladly answer any questions anyone has about the paper or the work that's going to start showing up.
37) Message boards : MilkyWay@home Science : Absolute magnitude distribution (Message 999)
Posted 10 Dec 2007 by ProfileNathan
Post:
There should be no break between the change, it should be nothing more than an update of the binaries.
I don't know about the WU designation, I'll have to talk to Travis about that.

We will have the ability to run both convolved and unconvolved searches, however, and it will be of interest to see how much of a time increase there is based on how many points we use, etc.

38) Message boards : MilkyWay@home Science : Absolute magnitude distribution (Message 954)
Posted 7 Dec 2007 by ProfileNathan
Post:
So in the very near future (as soon as Travis/Joe get the binaries updated again) there's going to be a change in the WUs. Namely they are going to be longer and I wanted to give you guys an idea as to why.

First of all, it's imperative to understand that it's immensely difficult to determine the distance to a star. This is because we can only measure how bright something is: this is know as the apparent magnitude. This wouldn't be a problem but all stars are not the same and so a really bright star that's really far away may look the same as a really dim star that's really close. So we developed a scale of the intrinsic brightness of stars. So we say how bright would a star be if it was a fixed distance away: we call this the star's absolute magnitude. Only if we have both the apparent and absolute magnitude can we then determine the distance to the object, and even then there is a large amount of error.



This figure is an HR diagram. They come in many differing forms but always show the temperature/color (x-axis) versus brightness/absolute magnitude (y-axis). So if you make a plot of all stars within a galaxy, etc., you will get something that looks like this. All stars traverse this plot throughout their life-cycles and so if you look at a population of stars you will see stars at all stages and thereby fill the diagram. If you follow the above link you can see more about what types of stars fall where and gain a little bit better understand of what you're looking at. Now if you divide up the the color/temperature correctly you are able to determine what type of star is within that range. This becomes very useful.

Right now with our algorithm we are looking at what's known as F-turnoff stars. These are the F stars that fall right at the point that the stars "turn off" the main sequence and start to move up the giant branch. And I've approximated this on the figure above with the green lines, so we are only looking at the stars between those lines. You can see that all stars we are looking at are not at the same y-value (absolute magnitude) but actually differ quite greatly over the small range in color.

The current WUs have been assuming a fixed value for the absolute magnitude and calculating distance for all stars using this value. As you can see, this is a rough estimate at best and not very good. The updated WUs will be assuming a Gaussian distribution of stars with a maximum at that fixed value. This then says that most of the stars will have this fixed value but theres some chance it's spread out from the value. Therefore, when we combine this with our algorithm we get a much more accurate representation.

Now, you may be thinking "that's great but why is it making the WUs take longer" well that's because we have to numerically integrate this Gaussian distribution and so we have to actually do about 30 times as many calculations as before. You shouldn't see this large of an increase in time as we've decreased some of the other values as well, but on average a WU should run about 4 times longer now. This may increase a little more later, too, if we see we need to be more accurate.

Hope this helps in letting you know what's going on, and let me know if you have any questions/comments.
39) Message boards : Cafe MilkyWay : Hi Folks! (Message 442)
Posted 14 Nov 2007 by ProfileNathan
Post:
I'll talk to the others today, and see if we can get some graphicy goodness up for you all. ^_^

40) Message boards : MilkyWay@home Science : Graduate Student Paper (Message 426)
Posted 12 Nov 2007 by ProfileNathan
Post:
I'm the astronomy PhD student :-). Yeah, it will be for all three on this end as well. There will be a lot of publications on the results we get from different data sets of the same structure, data sets of different structures, different models, and even on the method itself. So, as Travis said there is a lot of work to be done, and a lot of things to learn from it.


Previous 20 · Next 20

©2020 Astroinformatics Group