PDA

View Full Version : At very high stellar density



PetTastic
2014-Mar-25, 12:33 PM
So this started out as a Sci-fi idea.

An alien race finds their star system is falling into the core of a small dense globular cluster.
An argument starts. Do they spend all their planetary resources digging down into the crust of the planet and hope to survive the ride through the cluster's core? Or do they build solar sail generation ships and use the combined radiation pressure of their star and the cluster to escape?

However, I got a bit carried away modelling the scientific argument to the point it killed the story, and I never even got to look at the solar sail option. :o

The mean radiant temperature inside the cluster's core depends greatly on star density, effecting how much of the sky is black and how much is multi-thousand degree star surface.

At the starting numbers I picked for the density of the cluster, I soon realised even their star was going to be cooked with a mean radiant temperature well over 4,000K.
The star's surface would not cool normally. The stars in the cluster's core would be cooking each other in a runaway process. The radiant temperature at the core of the cluster increases exponentially as the cooked stars surface temperature rises and they puff up to cover more sky.

So, after all that the question is what happens when the star density at the core of a cluster or starburst galaxy gets too high? :think:

kzb
2014-Mar-25, 01:00 PM
Even at the centre of a globular cluster, the average separation between stars is many AU's. So I think it is very much a work of fiction what you propose.

Their biggest problem (in a real globular cluster) would be stellar encounters, which could disrupt the planetary orbits in their system. Almost certainly they would be in for a busy time with comets. On our time scales though, you would see what stars are coming your way generations ahead.

ngc3314
2014-Mar-25, 01:51 PM
There has been at least one study (http://adsabs.harvard.edu/abs/1983ApJ...272..390M) suggesting that this can happen close enough to an active galactic nucleus; theorist Bill Mathews termed it reverse stellar evolution. For low-mass stars in particular, the downward energy flux exceeds the upward flux and eventually bloats the outer layers enough to be lost to heating by X-ray absorption.

PetTastic
2014-Mar-25, 11:42 PM
There has been at least one study (http://adsabs.harvard.edu/abs/1983ApJ...272..390M) suggesting that this can happen close enough to an active galactic nucleus; theorist Bill Mathews termed it reverse stellar evolution. For low-mass stars in particular, the downward energy flux exceeds the upward flux and eventually bloats the outer layers enough to be lost to heating by X-ray absorption.

Interesting paper, thanks.

Could this effect put a limit on the density of globular clusters?
If all the stars in the core started to lose mass, the cluster would be less gravitationally bound and expand.

It would be interesting to know if this process comes into play in core collapse globular clusters, increasing the brightness of the stars near the core by heating and expanding their convection layers.

kzb
2014-Mar-27, 06:58 PM
That study is about the effect of quasars and active galactic nuclei on the surrounding stars. It's not the density of stars per se that causes the effects. If you have a bog standard globular cluster or galactic centre you won't see this happening.

In the situation in paper, the planet would be entering into a very hazardous region dominated by the radiation from a large black hole as it takes in matter.

chornedsnorkack
2014-Mar-27, 07:53 PM
If you are passing through a core then you may be more interesting not in space density of stars (it gives the average distance to nearer stars while inside the core, but not the minimum distance on nearest passage) but rather the surface density (First, it is easier to establish by observation. Second, while a star passing through the core would go on a curved path and see the stars themselves move, unlike a light ray, the probability of encounter would be only slightly increased).
See the collapsed core of M15:
http://archive.seds.org/hst/95-06.html
HST is unable to resolve the inner 4000 AU radius of M15.

Is there any estimate of the number of stars there?

PetTastic
2014-Mar-27, 08:38 PM
My original modelling was based on using python scripts in blender (free 3d package) to render fragments of sky at different star densities and star size distributions. I got mostly black sky at 0.004 stars per cubic light-year going up very bright, mostly white at 2.0.
So as far as my aliens were concerned there was no point digging into the planet's crust for a minimum ten thousand year ride through the cluster.
As chornedsnorkack (I think I spelt that ok) says it would not be a straight ride, and the alien's best hope would be looping around a star and being flung out quickly.

From reading up on core collapse and mass segregation, I now find it odd that some clusters don't have very very dense cores.

The denser the core gets the higher the movement dampening factors become.
Gravitational waves from close encounters, collisions, tidal forces from randomly rotating stars approaching each other, etc.

That is why I was suggesting there needs to be some mechanism in play that stops a near-total collapse, unless that is how big black holes are created.

cjameshuff
2014-Mar-28, 02:01 AM
I think your estimate is a bit off.

In terms of area: For stars like the sun, to cover 1% of the sky at a distance of 1 light year, you'd need nearly 2 trillion stars.

In terms of the inverse square law: Alpha Centauri A and B are separated by about 20 AU. It would take about 10 million similar stars at a distance of 1 light year to equal the light received from the other member of the binary.

chornedsnorkack
2014-Mar-28, 08:00 AM
My original modelling was based on using python scripts in blender (free 3d package) to render fragments of sky at different star densities and star size distributions. I got mostly black sky at 0.004 stars per cubic light-year going up very bright, mostly white at 2.0.
Pure illusion, due to the poor range of light intensity rendering by computer screens, and all artificial light sensors.
0.004 stars per cubic lightyear means, say, 1 sunlike star at 6,25 lightyears. Which is about 1,5 magnitudes dimmer than Alpha Centauri - magnitude about 1,25.
Mere 500 times denser? This means 500 stars within that distance. The nearest single star 8 times nearer - 0,8 lightyears - and therefore 64 times brighter, or 4,5 magnitudes, at about -3,25. And the total of the 500 stars being about 6,75 magnitudes brighter, at about -5,5. Star-dotted... but only slightly brighter than ordinary starry sky.


As chornedsnorkack (I think I spelt that ok) says it would not be a straight ride,
Not exactly straight, but nearly so, precisely because of large number of stars.

and the alien's best hope would be looping around a star and being flung out quickly.

No, thatīs the worst. Because it requires a close approach to that star.


From reading up on core collapse and mass segregation, I now find it odd that some clusters don't have very very dense cores.

It takes time. Most globular clusters needed more than they had.


That is why I was suggesting there needs to be some mechanism in play that stops a near-total collapse, unless that is how big black holes are created.
It is suggested that tight binaries are that mechanism.

PetTastic
2014-Mar-28, 10:35 AM
I think your estimate is a bit off.

In terms of area: For stars like the sun, to cover 1% of the sky at a distance of 1 light year, you'd need nearly 2 trillion stars.

In terms of the inverse square law: Alpha Centauri A and B are separated by about 20 AU. It would take about 10 million similar stars at a distance of 1 light year to equal the light received from the other member of the binary.

If I was to render using just OpenGL then almost all the stars are single pixels at both 0.004(local star density) and 4.0 stars per cubic light year.
So in terms of area covered there is only a few pixels difference.

Doing a floating point render on one of my 3D engines and then normalising both equally to 24 bits per pixel, I assume would give me one image with a few bright pixels (one white 0xffffff), and the other image would be mostly black (brightest pixel 0x191919).

So I was using a photo realistic mipmap sprite renderer in blender that replaces single pixel stars with single grey pixels going up to multi-pixel four pointed star sprites.
If I start with a setting that gives a realistic sky at local star density, increasing star density a thousand fold brings all the stars ten times closer and therefor a hundred times brighter.
Every star that was brighter than 0x191919 (dark grey) becomes a multi-pixel sprite as it would in a photo, nearby stars that were already sprites get ten times bigger.
However, I could only go up to 2.0 because the sprites overlap too much.

One problem with my approach is that that I am not letting any pixels cap out beyond white 0xffffff.
However, on the other hand most of the stars in the core of a cluster are larger than our local stars.

cjameshuff
2014-Mar-28, 11:14 AM
If I was to render using just OpenGL then almost all the stars are single pixels at both 0.004(local star density) and 4.0 stars per cubic light year.

The sun only spans 30 milliarcseconds as viewed from 1 light year. Unless you were rendering with hundreds of thousands of pixels per degree of field of view, the stars are far smaller than single pixels, equating a single pixel to a star greatly exaggerates their importance.

3D renderers just aren't designed for this task, and there are much simpler ways to approach the problem.

chornedsnorkack
2014-Mar-28, 11:14 AM
However, on the other hand most of the stars in the core of a cluster are larger than our local stars.
No, they are not. Most stars in globular clusters are dimmer than Sun because the stars bigger than Sun, a minority to begin with, have burnt out long ago. Only the even smaller minority of stars that are going through red giant stage currently are larger.

PetTastic
2014-Mar-28, 11:42 AM
The sun only spans 30 milliarcseconds as viewed from 1 light year. Unless you were rendering with hundreds of thousands of pixels per degree of field of view, the stars are far smaller than single pixels, equating a single pixel to a star greatly exaggerates their importance.

3D renderers just aren't designed for this task, and there are much simpler ways to approach the problem.

The mipmap sprite renderer can't render smaller than a pixel but does appear to calculate the correct brightness for a sub pixel object.

Well, you can just say all the stars are a hundred times brighter when ten times closer.

(All of those 0x191919 in my post should be 0x020202 not that is matters)

But in my test I was only reducing the average distance between stars by a factor of 10, wikipedia (http://en.wikipedia.org/wiki/Globular_cluster#Composition) has stars 100 to a 1000 times closer in the core, so up to 1 million times brighter.

cjameshuff
2014-Mar-28, 11:59 AM
Starlight 1 million times brighter than what Earth gets would still be about 0.001 times as bright as sunlight.

PetTastic
2014-Mar-28, 03:00 PM
Starlight 1 million times brighter than what Earth gets would still be about 0.001 times as bright as sunlight.

Well, that is 1 sun type star at 31.6 AU.
At the other extreme, I would guess brightening total star light by a factor of a million would be considerably more noticeable.

cjameshuff
2014-Mar-28, 04:19 PM
Well, that is 1 sun type star at 31.6 AU.
At the other extreme, I would guess brightening total star light by a factor of a million would be considerably more noticeable.

That figure was for brightening total star light by a factor of a million.

cjameshuff
2014-Mar-28, 04:51 PM
Also note that if a particular collection of stars is scaled so the distances decrease by a factor of 1 thousand and brightness increases by 1 million, density increases by a factor of 1 billion. With the total volume constant instead of the total number of stars, brightness is directly proportional to density. Going from 0.004 stars/light year to 2 stars/light year while keeping volume constant would only brighten the night skies by a factor of 500. If done by reducing distances within a cluster without adding stars, it would only brighten the night skies by a factor of 63. This is the scenario your method gave a "very bright, mostly white" sky for.

PetTastic
2014-Mar-28, 07:05 PM
That figure was for brightening total star light by a factor of a million.

Sorry, I must be using different numbers to you.
I have ambient daylight on Earth at 17,500 lux and night sky at 0.002 lux, so with this 'magic' factor of a million I am using. I get 0.114.
However, both numbers include atmospheric effects like air glow and scattering.
In space, I get a number much closer to yours with directional sunlight at 60,000 and ambient starlight at 0.0001 lux so 0.0067.

cjameshuff
2014-Mar-28, 10:29 PM
Sorry, I must be using different numbers to you.
I have ambient daylight on Earth at 17,500 lux and night sky at 0.002 lux, so with this 'magic' factor of a million I am using. I get 0.114.
However, both numbers include atmospheric effects like air glow and scattering.
In space, I get a number much closer to yours with directional sunlight at 60,000 and ambient starlight at 0.0001 lux so 0.0067.

Sunlight at Earth's orbital distance is closer to 127000 lux, starlight 0.0001 lux. But even with your numbers, their planet will reach equilibrium at a temperature where it radiates 1.114 times as much...all else being equal, this corresponds to an increase in temperature by a factor of 1.026. Their star will be warmed by a far, far smaller amount.

PetTastic
2014-Mar-29, 10:22 AM
So the moral of the story is: Never believe impressive results from a graphical simulation and post rubbish on forums until you have double checked you are not using StarLightNormalised in the aperture calculation instead of StarLightUnNormalised. :doh:

I still can't match those lux numbers, I, have 79000 as direct sunlight on Earth, including an ambient component from blue sky. Is that much visible light really lost?