Wrongly rendered colliding galaxies

[ANDREA]

For my “Violent, strange, stupend Universe” script, I was searching data regarding colliding galaxies, so I reached this very recent and beautiful link from HUBBLE (April 24, 2008)
http://hubblesite.org/newscenter/archive/releases/2008/2008/16/image/a/
After this I searched for the shown galaxies in Celestia, using the deepsky.dsc 1.10 (last one, 11 months old, available in sourgeforce here http://celestia.cvs.sourceforge.net/cel ... stia/data/),
Using Celestia 1.5.0 with added celestia-svn-r4317 by cartrite.
And here the surprises: most of the galaxies couples are not shown as in the images, but as distant or very distant objects.
Pay attention, we are not speaking of fake couples, with different distances, but of truly contacting or merging objcts, in collision each other.
I don’t know if I’m missing or mistaking something, but if I’m right I think there is something wrong in the deepsky.dsc data, at least in those regarding this particular type of objects.
Here the images of the couples I checked, on left Celestia's, on right the Hubble’s one.
I had to split them in three posts, too many to enter in a single message, sorry.
Any idea?
Bye

Andrea

[ANDREA]

Here the second set of images.
Sorry, another one left for the next message
Bye

Andrea

[ANDREA]

Here the last image.
Bye

Andrea

[t00fri]

I think there is something wrong in the deepsky.dsc data,

Before writing such an "exploded" thread, it's always good to check once more...

Here is what I get from the current deepsky.dsc for the galaxy triplet IC 4686, IC 4687 and IC 4689 (your first example above):



I think only this SPECTACULAR agreement between the Hubble imaging and my galaxy rendering would have deserved a special post! It is definitely a triumphal confirmation of my automatic rendering method based on catalog data alone, which I have pursued over years.

Now since already your first "bug" example was incorrect, please understand that I did not take the time to check the others...
I think you simply forgot to turn up the brightness enough.

But in general, common sense should tell that colliding galaxies are asking too much as to the method I realized in Celestia: When galaxies collide, they exercise a tremendous gravitational pull on each other, which tends to disrupt or badly disturb the standard shapes.
Here is a corresponding animation that I showed in CM:
http://www.mpa-garching.mpg.de/galform/ ... galaxy.mpg

Hence the naive Hubble templates that we are forced to use, can hardly be an accurate representation of this situation. Moreover, the templates are extremely low-resolution, little (128x128) images, only. It makes little sense to try and compare the corresponding rendering of Celestia with super hires Hubble images...

If you have enough patience, I can always recommend using custom templates to display the galaxies of your choice in more detail and with corrected shapes.

Fridger

[t00fri]

Of course, I was curious enough to check another one of your said "wrong galaxy renderings": The pair NGC 6050 A and NGC 6050B sometimes also
designated as the pair NGC 6050/IC 1179. It's the first example of your second group:

Unlike your claim, I get another spectacular agreement with the Hubble imaging. Find the comparison below:

Fridger

[t00fri]

Finally, I checked your last image, the agreement of which is also as good as it can possibly be (modulo gravitational distortions, of course):

The colliding pair NGC 6621/NGC 6622

Find the comparison below:



++++++++++++++++++++++++++
You see, Andrea, this is what comes out if one puts scientific precision first
++++++++++++++++++++++++++

[ANDREA]

You are right, Fridger, the images you show are almost perfect.
But could you please kindly explain me what is that makes my views so different from yours?
I told you that I'm using the latest deepsky.dsc so, or you are using another one (but not yet published in sourceforge), or there is something that I don't know, so that I'm making some mistake.
Here are the data from my deepsky.dsc, regarding NGC 6050A and 6050B:

Galaxy "NGC 6050A:IC 1179A:UGC 10186:MCG 3-41-93"
{
Type "SBc"
RA 16.0897
Dec 17.7572
Distance 1.938e+08 # distance uncertain!
Radius 2.537e+04
AbsMag -19.07
Axis [ -0.4596 -0.8537 0.2451]
Angle 43.2454
InfoURL "http://simbad.u-strasbg.fr/sim-id.pl?Ident=NGC 6050A"
}

Galaxy "NGC 6050B:IC 1179B:UGC 10186:MCG 3-41-92"
{
Type "SBc"
RA 16.0894
Dec 17.7544
Distance 4.918e+08 # method: NED (batch)
Radius 2.861e+04
AbsMag -19.79
Axis [ 0.4298 -0.8555 -0.2887]
Angle 111.9348
InfoURL "http://simbad.u-strasbg.fr/sim-id.pl?Ident=NGC 6050B"
}

Distances are VERY different, so, if they are given as "colliding", how can the 6050A distance be shown as "uncertain" ?

Distance 1.938e+08 # distance uncertain!

It should be equal to 6050B,or vice-versa, isn't it?
And, BTW, do you truly think I'm so stupid to pretend Celestia rendering galaxies like Hubble?
You lost a lot of your precious time to uselessly explain that

Moreover, the templates are extremely low-resolution, little (128x128) images, only. It makes little sense to try and compare the corresponding rendering of Celestia with super hires Hubble images...

Obviously this is not what I'm asking for: I'm asking why your coupling are different from mine, or better, why I have
NO COUPLING among those galaxies, and obviously it's not a matter of

...turn up the brightness enough...

Some of the images I showed you, like the first (NGC 4686-4687-4689, and the second, NGC 2206 and IC 2163) show clearly that the galaxies are far away, and surely not colliding.
In the attached image of NGC 6050A that I obtained with my data set, 6050B is so far away that, even with maximum galaxy brightness, it’s still invisible, we can see only its mark.
This is not a flame, at least for me, it's only a kind request of explanation on how you can obtain such views, using the same database.
I've still the doubt that their distances are wrong.
Thanks.

Andrea

[t00fri]

deepsky.dsc is processed completely automatically, since I can hardly inspect in person all 10000+ sets of galaxy data. In the course of this process, the computer attaches always the label '# distance uncertain!' if the distance is NOT given in any of the 10 catalogs that I used by means of an ACCURATE method. Then a truely uncertain method must be used that can easily be wrong by a factor of two, say. Fortunately, meanwhile the number of such remaining cases has become really small.

Since I can hardly test automatically by computer, whether any two galaxies are colliding, that label of caution is indeed justified from within MY ANALYSIS. If you have the time and devotion you can adjust by hand the hundreds of colliding galaxy distances in deepsky.dsc, of course. But which one of the two distances would you want to modify? .

I ONLY use published, peer reviewed values of distances and thus would never change such values by hand due to some kind of personal prejudice that comes from outside the 10 used renowned galaxy catalogs.

Transparent documentation in science is a most crucial request.

All my deepsky.dsc files have a unique processing date AND version attached. I use

# Adapted for Celestia with Perl script: deepsky.pl Revision: 1.50
# Processed 2007-5-18 5 137 0 18:10:24 UTC

which is the current one. There were NO changes recently. Your data for NGC 6050A,B are IDENTICAL to what I used.

Given that some distances can be somewhat uncertain/incorrect ( a factor 2, say), one can compensate a bit for this by zooming in/out the angle of view with SHIFT+ mouse_Left movements ("telescope effect, telephoto lens effect!). This makes uncertainties in the distance less apparent, but leaving the relative separation and orientations untouched. That's probably what you didn't exploit.

But apart from this, I really would not know why you didn't get the correct views. I used the public code and data files, nothing special whatsoever. Some other users will certainly manage to reproduce my displayed results.

We must all live with the fact that measurements can have substantial uncertainties. This holds particularly for astronomy. We still have no way of visualizing such inherent uncertainties in Celestia. But since we are educated people we should fold this fact in from the beginning into our argumentations ...

Fridger

[ANDREA]

deepsky.dsc is processed completely automatically, since I can hardly inspect in person all 10000+ sets of galaxy data. In the course of this process, the computer attaches always the label '# distance uncertain!' if the distance is NOT given in any of the 10 catalogs that I used by means of an ACCURATE method. Then a truely uncertain method must be used that can easily be wrong by a factor of two, say. Fortunately, meanwhile the number of such remaining cases has become really small.

Fridger, I cannot understand why everything that is told about your work (that’s precious, and I’m very serious saying this) is always interpreted by you as a personal offense.
I never told that ALL the deepsky.dsc data are wrong, I’m only saying that for the seven galaxy groups I tested (yes, we are speaking of seven, I repeat, SEVEN, not 10000!), there is a relevant distance error (2.5x in my example).
This obviously is not your fault, but anyway the problem exists, and is big.
So, instead of explaining us how clever you have been to automatically process completely the huge quantity of data taken from a huge quantity of datasets (we know this very well, given the incredible number of times you reminded us of this), it was perhaps better to say:
“OK, you are right, probably these distances are wrong, so, thank you, as I’ll have a bit of time I’ll try to check them.”
That would have been a polite reply, not the rude, aggressive, self-conceited one you gave here.

Since I can hardly test automatically by computer, whether any two galaxies are colliding, that label of caution is indeed justified from within MY ANALYSIS. If you have the time and devotion you can adjust by hand the hundreds of colliding galaxy distances in deepsky.dsc, of course. But which one of the two distances would you want to modify? .

The uncertain one, obviously, since the other one is certain and the galaxies are colliding.
Perhaps this is not the scientific way, but it’s surely a practical and logical way.

I ONLY use published, peer reviewed values of distances and thus would never change such values by hand due to some kind of personal prejudice that comes from outside the 10 used renowned galaxy catalogs.

So for you pointing out a probable error is:
"some kind of personal prejudice that comes from outside the 10 used renowned galaxy catalogs"?
Wow!
This is megalomania!

Transparent documentation in science is a most crucial request.All my deepsky.dsc files have a unique processing date AND version attached. I use
# Adapted for Celestia with Perl script: deepsky.pl Revision: 1.50
# Processed 2007-5-18 5 137 0 18:10:24 UTC
which is the current one. There were NO changes recently. Your data for NGC 6050A,B are IDENTICAL to what I used.
Given that some distances can be somewhat uncertain/incorrect ( a factor 2, say), one can compensate a bit for this by zooming in/out the angle of view with SHIFT+ mouse_Left movements ("telescope effect, telephoto lens effect!). This makes uncertainties in the distance less apparent, but leaving the relative separation and orientations untouched. That's probably what you didn't exploit.

No, I didn’t, and still I don’t see why I should have done it, given that the data contained in deepsky.dsc are the most accurate possible available.
I only wish to see such galaxies as HST sees them from the Earth, and to do it I wouldn't need to zoom or move arount them.
BTW, who are the NASA and Hubble astronomers (scientists like you, correct?) that name the same galaxies whose we are speaking about as “Colliding”?
How can collide galaxies with a distance 2.5 times bigger?
There is something wrong: or the galaxies are not colliding (hard to believe), or the distances are wrong..

But apart from this, I really would not know why you didn't get the correct views. I used the public code and data files, nothing special whatsoever. Some other users will certainly manage to reproduce my displayed results.
Fridger

I would equally like to see someone else obtaining the same results so, please, make the same test, just for sake of knowledge and to give an end to this thread.
Thank you

Andrea

[t00fri]

Andrea,

if you like to change these seven or so distances according to your liking, please do it in your personal deepsky.dsc. For the official version we are pursuing a solid, transparent and well-documented database strategy. It is simply not relevant whether you consider changing these seven values as practical or better or whatever.... How can YOU tell which of the 10000+ galaxies in my catalog are interacting in some way? How would you tell it to the computer? But we will certainly not fiddle a few galaxy distances according to gusto leaving another undetermined number untouched!

If you manage to come forward, however, with a scientifically accepted catalog where the distances of ALL known colliding galaxies are determined from reliable measurements, I will be VERY glad indeed to add these values to the official version. Whether you like it or not, Celestia's OFFICIAL database is set up ONLY from documented sources.

Fridger, I cannot understand why everything that is told about your work (that’s precious, and I’m very serious saying this) is always interpreted by you as a personal offense.

Andrea you are misinterpreting my reaction completely. I am a professional in this kind of stuff. It's not at all an emotional issue for me. I am just trying to tell you the methodology that we apply. Nothing else.

How can collide galaxies with a distance 2.5 times bigger?

You continuously forget to attach a big error bar to these values! That's why I have added that label of caution. You are NOT allowed to conclude that the distance is 2.5 times bigger from the given number.

Fridger

[Cham]

Andrea,

just to confirm Fridger's pictures. Here are both the "colliding" NGC6050 A and B on my own system. I only used the zoom in Celestia to get this (I didn't played with the brightness) :

ngc6050.jpg

You simply have to adjust your zoom factor, from a position close to Earth.

[t00fri]

Cham,

thanks for your test and for helping to further claifying this issue.

Actually, I am devoting some second thoughts to Andrea's proposal above, to constrain the uncertain distances of some colliding partner galaxies via that Hubblesite list
http://hubblesite.org/newscenter/archiv ... 6/image/a/
in an acceptable manner that would allow a fully traceable/transparent documentation of scientific standing.

I see two major remaining problems:
----------------------------------------
-- it is certainly not correct in general to EQUATE the distances of the colliding partners. Which of the distances are to be taken as the reference in an algorithmic sense, if they differ by a factor of two, say? Or should one take the mean square average, [tex]\langle d\rangle = \sqrt{d_1^2 + d_2^2}[/tex] . This option is bad if one distance determination is more solid... How to decide which of the two colliding galaxies is the CLOSER one. Incorrect choices often lead to qualitatively wrong visualizations! Example: The famous interacting pair M51(NGC 5194) + NGC 5195. To see what I mean just EXCHANGE the VERY similar distances of NGC 5194 and NGC 5195 in deepsky.dsc!

Moreover, the gravitational force is a LONG RANGE force (mass of graviton =0) and the masses of galaxies are huge. So gravitational pulls can act also on pretty large distances. Clusters of galaxies in the environment are also known to induce similar effects. Moreover there are images known of analogous gravitational deformations of INDIVIDUAL galaxies. This might be due to invisible black holes for example.

Hence it is clearly a speculative procedure to equate all the distances of the hubble list galaxy pairs. I need a solid conceptional algorithm that can be implemented AND documented cleanly!

-- There are of course MANY more colliding/interacting galaxies known beyond that short Hubble listing. We are certainly dealing conceptionally with more than the 7 cases pointed at by Andrea. From a scientific point of view that short Hubble listing is mainly "eye candy" and serves to underlign the importance of the Hubble telescope to the public . I know many more interacting galaxies myself from the digital sky survey (DSS), for example. The DSS imaging is much less "spectacular", but provides equally relevant scientific evidence. Hence we actually need an exhaustive scientific catalog of ALL such known galaxies, to do a good job in deepsky.dsc. So, if anyone from our "WEB wizzards" finds some such list...let me know!

Fridger

[Cham]

Fridger,

there is also the "geometric mean" defined by [tex]d_{geo}=\sqrt {d_{1}d_{2}}[/tex], with [tex]d_{1} <d_{geo}<d_{2}[/tex]. This distance could be usefull for colliding galaxies with an unreliable distance.

[ANDREA]

Andrea, just to confirm Fridger's pictures. Here are both the "colliding" NGC6050 A and B on my own system. I only used the zoom in Celestia to get this (I didn't played with the brightness): You simply have to adjust your zoom factor, from a position close to Earth.

Hello Cham, thank you for the example and information, it opened me a new approach.
So I tried first of all the NGC 6050A “follow” and NGC 6050B “lock” commands , then the zoom settings (using "," and "." buttons), starting from the Earth, and now the image I obtained is almost identical to Fridger's and your one, look below, but this has been achieved only after using “Magnitude Limit 15.0”, “Galaxy Light Gain 70%”, and “FOV 00° 00' 03"!!!”.
This solves my galaxy collision visibility problem, and many thanks to both of you, but this arises another problem: now I cannot any more “Orbit” the couple of galaxies, so the sight during my show is forcedly “STATIC”.
So, what’s the difference from instead showing the wonderful HST’s image as a Celestia billboard?
Nothing, I think.
This is the reason why I’m compelled to modify [b]“unscientifically and for personal use only, obviously”[/b] the distance of at least one of the colliding galaxies, in order to allow a “DYNAMIC” show of the phenomenon.
Thanks a lot.
Bye

Andrea

[ANDREA]

…Actually, I am devoting some second thoughts to Andrea's proposal above, to constrain the uncertain distances of some colliding partner galaxies via that Hubblesite list
http://hubblesite.org/newscenter/archiv ... 6/image/a/
in an acceptable manner that would allow a fully traceable/transparent documentation of scientific standing.

Thank you Fridger, this will be very appreciated, and surely not only by me, to avoid the problem I just addressed in my previous message to Cham, i.e. the “STILLNESS” of the actually obtained image.

…There are of course MANY more colliding/interacting galaxies known beyond that short Hubble listing. We are certainly dealing conceptionally with more than the 7 cases pointed at by Andrea.

I’m not so sure about this, considering that many of the Hubble’s images whose we are speaking of, are of galaxies not included in your 10000+ deepsky.dsc dataset.
Probably there are not so many in the dsc.

…From a scientific point of view that short Hubble listing is mainly "eye candy" and serves to underlign the importance of the Hubble telescope to the public . I know many more interacting galaxies myself from the digital sky survey (DSS), for example. The DSS imaging is much less "spectacular", but provides equally relevant scientific evidence. Hence we actually need an exhaustive scientific catalog of ALL such known galaxies, to do a good job in deepsky.dsc. So, if anyone from our "WEB wizzards" finds some such list...let me know! Fridger

Fridger, even if I’m not a scientist, from three years I use daily, repeat, daily, the DSS (here: http://stdatu.stsci.edu/cgi-bin/dss_form) for blinking the images of galaxies taken daily in search of supernovae (whose I personally discovered 6, up to now) with POSS-2 Red, Blue and IR images.
We image ALL the galaxies visible from Georgia and Arizona up to 19th magnitude, so a lot of galaxies, many more than 10000, and I can assure you that galaxies couples or triplets in APPARENT contact (apparent, obviously) are just a very small percentage.
I think that truly colliding objects in your 10000+ dataset could be less than 100, and probably many less. Thanks again.
Bye

Andrea

[ANDREA]

Just for jocking, and about colliding and merging galaxies, do you remember the "Galaxy Zoo" project, about one or two years ago, where worlwide volunteers were involved in classification of galaxies?
Well, on August 2, 2007 they reached the target of 10,000,000# images of galaxies checked , and in one of their forums they opened the "Biggest cosmic trainwrecks - Mergers" Topic, where we can find a lot of such objects, all in color.
Images are not so nice, but the galaxies yes.
Give a look here (a total of 288 pages, wow!):
http://www.galaxyzooforum.org/index.php?topic=567.0
Bye

Andrea

[ANDREA]

i modified the distance value of NGC 6050A, giving the sameof 6040B, i.e. Distance 4.918e+08
The first image shows the result, obtained with normal FOV (28°) and very lower values of Magnitude Limit and Galaxy Light Gain.
As you see we are very close to HST image, but with the most important difference regarding my previous one: now I can look all around the couple,obtaining views like the 4 shown in the second image.
But I think that the result is not identical to the HST image, the two galaxies are too distant each other, so probably something else should be done with R.A. and DEC values.
Anyhow, I'm satisfied with this, even if, alas!, all the beauty and the subtle variations of colors in the image obtained at 00° 00' 03" FOV are lost.
Fridger, surely there is some important reason about this that I don't know, but is there any hope to obtain the very nice results of your image at normal FOVs?
Thank you.
Bye

Andrea

[Cham]

Andrea,

the solution is very easy : just give to both galaxies a mean value for their distance. This way, you'll get two "interacting" galaxies in Celestia if you move around them at close range, and you'll keep the same image at small FOV when you stay close to Earth.

[ANDREA]

Andrea, the solution is very easy : just give to both galaxies a mean value for their distance. This way, you'll get two "interacting" galaxies in Celestia if you move around them at close range, and you'll keep the same image at small FOV when you stay close to Earth.

Oh, sure, it's the logical solution I didn't search for.
Thank you for pointing it, Cham.
Bye

Andrea

[t00fri]

Friends,

I think the NGC 6050A,B pair is a FIRST example of what worries me since the beginning of this discussion:

Namely that it may simply be WRONG to equate the two distances or to set them equal to some average value! See my astrophysical arguments above, why this is so...

Why?

BOTH in my Celestia Image AND in the Hubble photo you can see that the left hand galaxy 6050A actually has the BIGGER apparent size of the two. With small errors, however, the NED database says that 6050A is considerably SMALLER than 6050B. With my substantially different distances for the pair this just reproduces the RIGHT relative topology and size in agreement with the Hubble image.

If we associate a common average distance to both the apparent size relation is reversed and INCORRECT!

So this indicates that qualitatively, 6050B is considerably farther away than 6050A...

Fridger