It is currently Tue, 21-11-17, 11:49 GMT

All times are UTC




Post new topic Reply to topic  [ 32 posts ]  Go to page Previous  1, 2, 3  Next
Author Message
PostPosted: Fri, 27-06-14, 19:19 GMT 
Offline
User avatar

Joined: Tue, 04-09-07, 21:55 GMT
Posts: 773
Location: N 42.38846 W 83.45456
just a means of having repeatability
how this is done ????????

in a "cel" url might be a good idea
or in a setting window with a check box and input line

a option for using urand or typing in a seed

_________________
"I don't pitch Linux to my friends, I let Microsoft do that for me."
Using OpenSUSE 42.1 & Scientific Linux 6.7


Top
 Profile  
 
PostPosted: Sat, 28-06-14, 9:46 GMT 
Offline

Joined: Fri, 13-06-14, 17:45 GMT
Posts: 59
Dear Fridger!

I found very interesting your argument and I must agree with you.

Please don't misunderstand me, IT IS a fantastic work you are doing, there is no doubt about it. I would REALLY like to use celestia.Sci (I will be VERY happy on that day, when I can do that).

But please try to understand my concerns too.

So I beg to differ in one question.

Earlier you wrote "the locations etc of the random stars change" and later you wrote "By using probabilistic approaches along with the measured data on relevant probability densities, one can make VERY solid statements about such objects, DESPITE their elusiveness and far distances.".

First of all, I would not call a statement solid, if it changes every time I start celestia.Sci.

You wrote "HENCE...-- with position data from these catalogs taken at face value -- the respective stars are probably not THERE! Still with a correct interpretation of these data i.e. by properly including the error bars, the contained information is most valuable.".

I think the difference of opinion is about the connection of probability, predictions, measurement, uncertainties and reality.

I think it seems - and I might be wrong on this - that you are not differentiating consciously between two different types of distinct uncertainties.

First type of uncertainty shows up when we are predicting objects which are not measured directly - for example stars which cannot yet be studied individually. The second type shows up when we are talking about objects which are measured directly but the measurement contains errors or it is not sufficient.

As I wrote, my concern is psychological, or you call my concern philosophical (as in philosophy of science).

Your world view may be based on more platonic ideas (which comes from your excellent knowledge and use of mathematics), mine is based on more empirical and psychological ideas.

For me measurement means empirical proof, probabilistic prediction without empirical proof means realistic fiction.

Obviously your method is scientific and it works, but I think there is a difference between for example predicting the Higgs boson and measuring the Higgs boson...

So I beg to differ...I think there is a difference between predicted stars and measured stars...even if uncertainties exist in both cases.

I'm not saying that there is a better way to make a scientific software about the Universe - your method is ingenious and almost perfect - but there should be a visible border between measured and predicted parts of the software - in my opinion.


Top
 Profile  
 
PostPosted: Sat, 28-06-14, 15:34 GMT 
Offline
User avatar

Joined: Fri, 03-04-09, 8:21 GMT
Posts: 211
sjohn wrote:
For me measurement means empirical proof, probabilistic prediction without empirical proof means realistic fiction.

sjohn,

For me, the fact that you are able to post on this forum (using a computer) is empirical proof that probabilistic prediction can mean realistic fact. ( Thanks, Mr Schrodinger ;) )

Would you agree that it's impossible to post on this forum without a working computer?
Would you agree that a computer will not work without electricity? (Note: as well as a few other quantum properties at the smallest scales).
Would you agree that electricity is the result of the flow of electrons?
Would you agree that the location of an electron can only be described as a probability density?

The Universe is a far stranger place than anyone realizes...(with the exception of particle physicists) Image

You may find the following intriguing: A surprising similarity between stars and atoms

_________________
CITIZENS OF CM - JOIN THE REVOLUTION
...black out your avatar
THE AVATARS ARE REVOLTING !!!


Top
 Profile  
 
PostPosted: Sat, 28-06-14, 18:09 GMT 
Offline

Joined: Fri, 13-06-14, 17:45 GMT
Posts: 59
chuft-captain,

Are you saying that there is no qualitative difference between the quantum realm and the realm of conscious perceptions (observations, macroworld)?

I don't think that any interpretation of quantum mechanics is saying that.

A star is a conscious perception (observation), the quantum realm is the method behind it.

I think you are mixing up the question of "what" with the question of "how".

A star is a "what", quantum mechanics are the "how".

Every measurement is a "what", probabilistic determination is only a "how".

Measured stars are "real", generated stars are "fake", but of course can be very realistic.

But I think "real" is not "realistic". Your consciousness is not uncertain. The outside world is uncertain, but not the measurement.

That is why I'm saying all along that my argument is psychological and philosophical...

Measured stars are real (like the data of Gaia mission will be), predicted stars are realistic, but fictional...

Of course measurements can be inaccurate, and predictions can be accurate, but it is not my point...they are fundamentally different ontologically.

So all I'm saying is that celestia.Sci should mirror the ontological state of reality - in my opinion anyway...


Top
 Profile  
 
PostPosted: Sun, 29-06-14, 0:13 GMT 
Offline
User avatar

Joined: Mon, 03-09-07, 23:01 GMT
Posts: 388
Location: Tuscany, Tyrrhenian Sea
chuft-captain wrote:

However this would require that the actual randomly generated seed value (from a seed = time(NULL) call) is saved when fluct is set to false, and re-called on subsequent startup.



How to read before the load the part after the "else"...

Code:
if(fluct)
    srand(time(NULL);
else
 /*   srand(1); */
int seed = srand(time(NULL);
printf ("my seed: %d\n", seed);
...

_________________
Never at rest.


Top
 Profile  
 
PostPosted: Sun, 29-06-14, 1:46 GMT 
Offline
User avatar

Joined: Mon, 03-09-07, 23:01 GMT
Posts: 388
Location: Tuscany, Tyrrhenian Sea
sjohn wrote:
but there should be a visible border between measured and predicted parts of the software


Paradoxally and out of irony, this is just what Fridger is doing. Its method is "operational"; the "solid statements" based upon catalogs concerns the stellar classes: naively, how many red/blue/white giants/normal/dwarf stars are there to be OpenGLcally drawn in their colors. Thus the two uncertainties below:

sjohn wrote:
First type of uncertainty shows up when we are predicting objects which are not measured directly - for example stars which cannot yet be studied individually. The second type shows up when we are talking about objects which are measured directly but the measurement contains errors or it is not sufficient.


have to be reduced to one: the first kind, because:

sjohn wrote:
First of all, I would not call a statement solid, if it changes every time I start celestia.Sci.


is not valid, as long as the stellar classes are maintained every time celestia.Sci starts, meaning that, e.a., still naively, if the red stars are found at border of the galaxy, every time they will be at border and not one time they will be at centre.

_________________
Never at rest.


Top
 Profile  
 
PostPosted: Sun, 29-06-14, 5:52 GMT 
Offline

Joined: Fri, 13-06-14, 17:45 GMT
Posts: 59
Dear fenerit!

Are you saying that Hipparcos and Gaia satellites were unnecessary, we should just predict stars?

Should we ban experimental physics too?

Or confirmation is ontologically unnecessary?

Are we talking about a new scientific method here?

Some kind of virtual mathematical solipsism?

Because if not, I don't see that any of your arguments has any effect on the ontological problem I presented...


Top
 Profile  
 
PostPosted: Mon, 30-06-14, 6:30 GMT 
Offline
User avatar

Joined: Fri, 03-04-09, 8:21 GMT
Posts: 211
sjohn,

As Fridger has already made the decision to use stochastic methods for visualization of objects at cosmological distances (which is his decision to make), we should reserve this thread for discussion of the details of how this can be correctly achieved (scientifically) in the software.

Can I suggest that you start a new thread specifically for the purpose of your more general "psychological / philosophical" discussion. That would be a more appropriate place for anyone to discuss the validity of these methods, however I'm finding these particular philosophical discussions to be a little distracting in the context of this thread.

Kind regards
CC

_________________
CITIZENS OF CM - JOIN THE REVOLUTION
...black out your avatar
THE AVATARS ARE REVOLTING !!!


Top
 Profile  
 
PostPosted: Mon, 30-06-14, 15:55 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4508
Location: Hamburg, Germany
chuft-captain wrote:
sjohn,
As Fridger has already made the decision to use stochastic methods for visualization of objects at cosmological distances (which is his decision to make), we should reserve this thread for discussion of the details of how this can be correctly achieved (scientifically) in the software.


Let me close this not very useful debate with sjohn with just a few general comments:

After reading sjohn's posts again carefully, it seemed to me that he firmly believes that my statistical approach in celestia.Sci is pure theory, i.e. being not based on any direct experimental measurements whatsoever..

This is of course entirely wrong, since all used probability densities in celestia.Sci rely on best-fit approximations to the respective existing experimental data like luminosity densities, spacial star densities, color-magnitude distributions etc!

Before entering such a discussion on inherent uncertainties, It is indispensable to know the precise meaning of experimental error bars that accompany EVERY kind of experimental measurement! Firstly, there are only two types of experimental errors:

  • so-called statistical errors (with identical definition in conventional experimental measurements and stochastic methods). These decrease with the number of measurements or with the number of objects involved.
  • so-called systematic errors. These typically characterize (hardware) imperfections of the measuring apparatus or other deteriorationg effects that do NOT decrease with the number of measurements.

The basic "axiom" in case of statistical errors is that all measurement errors are distributed in a Gaussian manner (<=> Normal distribution). The error bars then result as the standard deviation of this distribution. Thus experimental uncertainties quantify the spread of the results of measurements when these are repeated many times or when many objects are involved. The familiar error bars are defined in terms of the width of the underlying Normal distribution.

Since systematic errors are NOT normal distributed, it is generally not allowed to add systematic and statistical errors in square to form the total experimental error.

Last not least:
A great concern of mine is to implement an intuitive and transparent visual scheme of displaying experimental uncertainties in celestia.Sci. This is all but easy and good ideas would be much appreciated!

A typical example are the orbits of multiple star systems. Since the measured orbital parameters partly involve large errors, the graphical display of the resulting orbit lines should make this unambiguously clear, without messing up the clarity of the display!

Fridger

_________________
Image


Top
 Profile  
 
PostPosted: Mon, 30-06-14, 22:41 GMT 
Offline
User avatar

Joined: Fri, 03-04-09, 8:21 GMT
Posts: 211
Quote:
A great concern of mine is generally to implement an intuitive and transparent visual manner of displaying experimental uncertainties in celestia.Sci. This is all but easy and good ideas would be much appreciated!

Perhaps you can employ a variation on the technique used to display the rings of Saturn.
In this case the maximum opacity of the texture would be in the center and the texture would be drawn in the plane of the orbit, centered on the orbit line, and the opacity would mirror the y-axis value of the gaussian bell-curve, hence reducing gradually to complete transparency at perhaps 5 std deviations either side of the (mean) orbit line.

Of course, this may be difficult to implement correctly using textures, as that approach may be too static for your purposes (I suspect you would want to render this in 3D - i.e. above and below the plane of the orbit, as well as in the plane, and that the size of the error may vary at different points in the orbit), however maybe you can achieve the "look" of this approach (if you can imagine it) in a dynamic fashion by applying the same effect in principle by varying the opacity of the "error" region in code.

CC

_________________
CITIZENS OF CM - JOIN THE REVOLUTION
...black out your avatar
THE AVATARS ARE REVOLTING !!!


Top
 Profile  
 
PostPosted: Mon, 30-06-14, 23:31 GMT 
Offline
User avatar

Joined: Mon, 03-09-07, 23:01 GMT
Posts: 388
Location: Tuscany, Tyrrhenian Sea
I think you could do that following the way in which math softwares does: a new menu item named, say, "Statistics" then sub-menus like, say, "Deviations..." and... plop! ... a window with literally the error bars concerning the selected subject; histograms, plots, "spreadsheets looking data" or whatelse is seen in math softwares. This should be a fine evolution of Celestia and in specific an unique features of celestia.Sci. (if QT environment allow such plotting tools).

_________________
Never at rest.


Top
 Profile  
 
PostPosted: Tue, 01-07-14, 15:01 GMT 
Offline

Joined: Fri, 13-06-14, 17:45 GMT
Posts: 59
Dear Fridger!

Just one more comment:

You wrote:

"After reading sjohn's posts again carefully, it seemed to me that he firmly believes that my statistical approach in celestia.Sci is pure theory, i.e. being not based on any direct experimental measurements whatsoever.. This is of course entirely wrong, since all used probability densities in celestia.Sci rely on best-fit approximations to the respective existing experimental data like luminosity densities, spacial star densities, color-magnitude distributions etc! "

You absolutely misunderstood my point.

Never questioned your scientific method and I never denied that your statistical approach is based on empirical knowledge.

My point was about direct/indirect knowledge and the ontological/epistemological difference between the two.

For example:

Direct knowledge: a.) discovering America b.) completely mapping America with ships c.) mapping the Galaxy with the Gaia satellite.

Indirect knowledge: a.) making a theory about that there must be more landmass "somewhere", b.) making a probabilistic approximation about the shape of America (based on empirical facts and natural laws) c.) predicting stars with random star generator using a scientific method based on empirical facts.

If you are saying that this problem is irrelevant in a topic about "Do's and Don'ts" of random star generation...well...then I really rest my case...


Top
 Profile  
 
PostPosted: Tue, 01-07-14, 16:19 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4508
Location: Hamburg, Germany
CC & fenerit,

thanks for your thoughts about visualizing measurement uncertainties in celestia.Sci.!

@fenerit: Sure, errors will not be displayed continuously, but rather there will be some ToolBar button & a menu entry, both with sub-entries to toggle visualizations of uncertainties in various ways.

@CC: Using the opacity degree of freedom for characterizing the continuous effects of input parameter errors along 2D orbits or even for 3D structures seems to have potential. Anyway, that is also the generic direction I was contemplating since a while. To realize something like what you propose I actually would use GLSL shaders also for reasons of speed.

I should also point out that there is a most interesting 2D plotting package (QWT 6.1.x, LGPLv.2) that offers a multitude of fancy plot types e.g. as overlays on the celestia.Sci canvas. The great thing is that QWT is fully integrated into the standard Qt design & building tools: Qt-designer and Qt-creator.
You can easily google for info about QWT and for displays of many example plots ...

This package can serve very well to illustrate particular aspects of inherent uncertainties with a click.

Cheers,
Fridger

_________________
Image


Top
 Profile  
 
PostPosted: Tue, 01-07-14, 16:45 GMT 
Offline
Site Admin
User avatar

Joined: Fri, 31-08-07, 7:01 GMT
Posts: 4508
Location: Hamburg, Germany
sjohn,

Quote:
...
My point was about direct/indirect knowledge and the ontological/epistemological difference between the two.


Apparently, this issue is not at all relevant here, since in Deep Space astrophysics and Cosmology ALL experimental data necessarily represent indirect knowledge! Unlike Christopher Columbus who sailed to America and stepped himself on the discovered land, it is clearly impossible in Deep Space physics to place a detector right where you want to measure. The distances are clearly far too large.

For example, we will probably never be able to travel to distant galaxies in order to prove directly that Black Holes exist in their centers. Still every astronomer and astrophysicist ist meanwhile convinced that Black Holes do exist in the centers of all galaxies. Of course there are many indirect evidences that all of us scientists consider sufficient.

Analogously, we cannot even prove directly that the stars which Hipparcos or Gaia satellites measure actually do exist in the way they appear to us many lightyears away! Still no-one would actually doubt the results!

Because of General Relativity (i.e. specifically gravitational lensing!), the apparent places of stars may be shifted and the apparent luminosities be increased due to big nearby masses. Because we do believe in the GR theory, we nowadays can subtract such effects. But note, all this is far from direct proves.

...And so on...

In summary: In Deep Space ALL evidences represent indirect knowledge in some way. Hence, we always need the support of theoretical considerations to gain valuable insight!

Fridger

_________________
Image


Top
 Profile  
 
PostPosted: Tue, 01-07-14, 21:52 GMT 
Offline
User avatar

Joined: Fri, 03-04-09, 8:21 GMT
Posts: 211
sjohn,
Quote:
predicting stars with random star generator using a scientific method based on empirical facts.
It's probably worth noting that "predicting stars" is not Fridger's aim.

The aim (correct me if I'm wrong Fridger) is to produce a realistic "facsimile" of the gross morphology of these deep space objects, such that the proportion and distribution of different classes of stars in the overall population of a certain type of galaxy, for example, reflects the reality we would observe. (e.g. Typically, what proportion of stars in a particular type of galaxy are red-dwarfs and how does the evidence suggest that they are usually distributed within that type of galaxy?)

So the aim is not to "predict" individual stars, but to reproduce their approximate distributions spatially according to the evidence we have. With this aim in mind, it is irrelevant where an individual star is situated exactly, or whether indeed it exists at all. What's important is that the entire population and distribution of that type of star within a galaxy, etc... is determined by the correct statistical technique for the circumstances, whether that be a modified gaussian distribution or some other distribution altogether, according to what we know about their distribution from observation and other evidence.
The hope is that by doing so the appearance from a distance of deep-space objects in Celestia will closely match their real-life appearance.
At the end of the day, there are many other factors involved, such as how well do the star-shaders replicate the color and luminosity of each star and it's contribution to the whole, and how good is the quality of the monitor you display it on, etc, etc..., so some compromises to absolute accuracy are inevitable.

The "do's and don'ts" thread should really focus on critique of the statistical techniques employed, and how well they are likely to achieve this "limited" ambition, with all these constraints in mind. The wider debate, though valid as a philosophical discussion, IMO should be the subject of it's own thread.

Regards
CC

_________________
CITIZENS OF CM - JOIN THE REVOLUTION
...black out your avatar
THE AVATARS ARE REVOLTING !!!


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 32 posts ]  Go to page Previous  1, 2, 3  Next

All times are UTC


Who is online

Users browsing this forum: No registered users and 2 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group