Earth's solar system is special

There's been several stories recently about how a recent computer
simulation determined that Earth's solar system was actually quite
unique among the extra-solar systems that have been discovered in the
last decade. The articles basically say that this solar system started
out with exactly the right amount of dust and gas in its disk to create
a highly serene solar system where planets don't knock each other out.

SPACE.com -- Study: All Planets Are Born in Killer Environments
"By comparison, our solar system's gas giants – Saturn and Jupiter –
have nearly circular orbits that suggest less violent interaction. The
two planets also appear to have stayed close to where they grew up,
instead of migrating into the sun."
http://www.space.com/scienceastronom...evolution.html

On Aug 10, 9:02*pm, Yousuf Khan wrote:
There's been several stories recently about how a recent computer
simulation determined that Earth's solar system was actually quite
unique among the extra-solar systems that have been discovered in the
last decade. The articles basically say that this solar system started
out with exactly the right amount of dust and gas in its disk to create
a highly serene solar system where planets don't knock each other out.

SPACE.com -- Study: All Planets Are Born in Killer Environments
"By comparison, our solar system's gas giants – Saturn and Jupiter –
have nearly circular orbits that suggest less violent interaction. The
two planets also appear to have stayed close to where they grew up,
instead of migrating into the sun."http://www.space.com/scienceastronomy/080807-planet-evolution.html

From my simulations, combination of giants and small planets is fairly
common. Current observations also support that combination of giants,
and small planets is fairly common. Hint a gas giant at high
eccentricity kills surrounding gas halo/kuiper belt like. (Thought,
systems with low eccentricities doesn't need to have surrounding halo
like objects.)

If he had problems with creation of diverse star system, there might
be problem with his simulations. Did he computed energy radiation and
chemical interaction in the system? Unlikely. However these things are
important.

Did he wrote that program he used for simulation himself? Any
researcher should write very carefully his own program, so he could
know it's disadvantages, and decide on type of error cancellation.
Current CPUs accelerate 64 bit FP numbers, so algorithms must be
modified to work well with this restriction. (Which basically means
write a new algorithm.) Error propagation in long term simulations
could be a bitch.


He probably also missed the problem with stellar nurseries. A start
that lost 90 from its 98 planets, and caught 4 from planets that lost
other star would have quite different composition than a star from his
simulation. And yes stellar nurseries are that scary. (Thought I
didn't had computer power/time to create them with all necessary
stars. Few million particles are not sufficient, few tenths-hundred
might be.)

(BTW a hot Jupiter isn't a problem. A hot Jupiter at distance "a"
falling to "a/2" doesn't prevent a smaller object at distance "5*a" to
fall to "2*a". A hot Jupiter then just regularize the orbit of that
planet.)

Yousuf Khan writes:

Well, maybe the Earth system isn't special to the level of being rare,
but still among 300 systems found, how many are of this type, perhaps
3 of them? That makes 1%, which is still a small amount.

There are tremendous selection biases that skew the observed
distribution. If we look really hard at a population with equal
numbers of systems with rocky planets and gas giants, many more gas
giants will be detected. That's just a consequence of how the
detection techniques work. I.e. stellar transit observations are
biased toward large planets which occult a large fraction of the
star's light; doppler planets are biased towards massive planets which
can affect the central star significantly. Small rocky planets are
neither.

CM

On Aug 13, 5:48 pm, Craig Markwardt
wrote:
There are tremendous selection biases that skew the observed
distribution. If we look really hard at a population with equal
numbers of systems with rocky planets and gas giants, many more gas
giants will be detected. That's just a consequence of how the
detection techniques work. I.e. stellar transit observations are
biased toward large planets which occult a large fraction of the
star's light; doppler planets are biased towards massive planets which
can affect the central star significantly. Small rocky planets are
neither.

The bias towards large mass planet detection is well known, but I
don't think their simulation is actually modeling actual star systems,
just theoretical star systems given variable disk mass ranges. So
given a certain disk mass, after the modeling has finished, does the
theoretical system look like any real systems, and if so which ones,
and how many real systems match the modeled system. So by the end of
it, model A might most closely look like 3 actual systems, mode B
might look like 1 actual system, model C might look like none, model D
might look like 2, etc., etc. Of course the most interesting model
would be the one that most closely resembles our own system, and
associated with that is how many other systems look similar to our
own.

As for the the large planet detection bias, it wouldn't come into play
here, as the models would probably account for small planets that we
haven't even detected yet within these systems. So it might be
interesting to note if their planet models are right, then whether
they've predicted the right number of small planets in certain star
systems where we've already detected their large planets?

Yousuf Khan

On Aug 12, 6:25 pm, Raghar wrote:
If he had problems with creation of diverse star system, there might
be problem with his simulations. Did he computed energy radiation and
chemical interaction in the system? Unlikely. However these things are
important.

The article does not mention any of that, perhaps you should contact
the authors yourself?

Did he wrote that program he used for simulation himself? Any
researcher should write very carefully his own program, so he could
know it's disadvantages, and decide on type of error cancellation.
Current CPUs accelerate 64 bit FP numbers, so algorithms must be
modified to work well with this restriction. (Which basically means
write a new algorithm.) Error propagation in long term simulations
could be a bitch.

The 64-bit FP number is currently the highest precision standardized
FP number system available (aka double-precision FP), with 53
fractional bits available, which corresponds to around 15-16 decimal
places. If they are limited by that, then everybody is limited by
that.

Anyways, they aren't trying to model actual solar systems, just
theoretical solar systems which they can then fit to actual solar
systems. So the in-between precision won't matter that much since they
are just interested in which real solar systems fit their model solar
systems.

He probably also missed the problem with stellar nurseries. A start
that lost 90 from its 98 planets, and caught 4 from planets that lost
other star would have quite different composition than a star from his
simulation. And yes stellar nurseries are that scary. (Thought I
didn't had computer power/time to create them with all necessary
stars. Few million particles are not sufficient, few tenths-hundred
might be.)

Yeah, it doesn't look like he was attempting to model complex systems
like this. I guess they are mainly interested in simpler models which
they can then fit to existing real-world systems.

(BTW a hot Jupiter isn't a problem. A hot Jupiter at distance "a"
falling to "a/2" doesn't prevent a smaller object at distance "5*a" to
fall to "2*a". A hot Jupiter then just regularize the orbit of that
planet.)

Can smaller objects actually form outside the orbit of hot Jupiters?

Yousuf Khan

On Aug 14, 10:37*pm, wrote:
On Aug 12, 6:25 pm, Raghar wrote:

If he had problems with creation of diverse star system, there might
be problem with his simulations. Did he computed energy radiation and
chemical interaction in the system? Unlikely. However these things are
important.

The article does not mention any of that, perhaps you should contact
the authors yourself?

Would they have computing power and will to do these simulations?
These are quite expensive (in terms of computing power), and special
hardware doesn't help much.

Did he wrote that program he used for simulation himself? Any
researcher should write very carefully his own program, so he could
know it's disadvantages, and decide on type of error cancellation.
Current CPUs accelerate 64 bit FP numbers, so algorithms must be
modified to work well with this restriction. (Which basically means
write a new algorithm.) *Error propagation in long term simulations
could be a bitch.

The 64-bit FP number is currently the highest precision standardized
FP number system available (aka double-precision FP), with 53
fractional bits available, which corresponds to around 15-16 decimal
places. If they are limited by that, then everybody is limited by
that.


There is always 128 bit integer, or better. Basically arbitrary
precision is just slow, however not impossible. (128 bit numbers are
quite nice for finding artefacts of simulation.)

Anyways, they aren't trying to model actual solar systems, just
theoretical solar systems which they can then fit to actual solar
systems. So the in-between precision won't matter that much since they
are just interested in which real solar systems fit their model solar
systems.

However when they'd have bug in program they are using because of one
of pure computer engineering/programming problems, theirs models would
be quite far away from reality, and they will not know about that.

(BTW a hot Jupiter isn't a problem. A hot Jupiter at distance "a"
falling to "a/2" doesn't prevent a smaller object at distance "5*a" to
fall to "2*a". A hot Jupiter then just regularize the orbit of that
planet.)

Can smaller objects actually form outside the orbit of hot Jupiters?


Yes. Why not? In fact they have an advantage of regularization of
theirs orbit. Basically, it's the same problem as creation of planets
in a close binary system.

Raghar wrote:
On Aug 14, 10:37 pm, wrote:
On Aug 12, 6:25 pm, Raghar wrote:

If he had problems with creation of diverse star system, there might
be problem with his simulations. Did he computed energy radiation and
chemical interaction in the system? Unlikely. However these things are
important.
The article does not mention any of that, perhaps you should contact
the authors yourself?

Would they have computing power and will to do these simulations?
These are quite expensive (in terms of computing power), and special
hardware doesn't help much.

Well, perhaps you can contact the original authors from the article I
posted? Perhaps you can find them through a google search?

There is always 128 bit integer, or better. Basically arbitrary
precision is just slow, however not impossible. (128 bit numbers are
quite nice for finding artefacts of simulation.)

The 128-bit FP number data type has not yet been standardized by the
IEEE. Until it is, you can't meaningfully compare simulations with it.

However when they'd have bug in program they are using because of one
of pure computer engineering/programming problems, theirs models would
be quite far away from reality, and they will not know about that.

I doubt that precision will make a whole lot of difference to their
simulations. As the original authors said, they did 100 different
simulations of something like 1 million particles each. They were just
roughly trying to correlate their simulations real world solar systems,
for categorization purposes. They weren't saying that those solar
systems evolved *exactly* like that.


Yousuf Khan

On Aug 24, 5:54 pm, Yousuf Khan wrote:
Raghar wrote:
On Aug 14, 10:37 pm, wrote:
On Aug 12, 6:25 pm, Raghar wrote:

If he had problems with creation of diverse star system, there might
be problem with his simulations. Did he computed energy radiation and
chemical interaction in the system? Unlikely. However these things are
important.
The article does not mention any of that, perhaps you should contact
the authors yourself?

Would they have computing power and will to do these simulations?
These are quite expensive (in terms of computing power), and special
hardware doesn't help much.

Well, perhaps you can contact the original authors from the article I
posted? Perhaps you can find them through a google search?

There is always 128 bit integer, or better. Basically arbitrary
precision is just slow, however not impossible. (128 bit numbers are
quite nice for finding artefacts of simulation.)

The 128-bit FP number data type has not yet been standardized by the
IEEE. Until it is, you can't meaningfully compare simulations with it.

However when they'd have bug in program they are using because of one
of pure computer engineering/programming problems, theirs models would
be quite far away from reality, and they will not know about that.

I doubt that precision will make a whole lot of difference to their
simulations. As the original authors said, they did 100 different
simulations of something like 1 million particles each. They were just
roughly trying to correlate their simulations real world solar systems,
for categorization purposes. They weren't saying that those solar
systems evolved *exactly* like that.

Yousuf Khan

Are you aware that the _actual basis, for the nebular hypothesis, you
know the current theory on how planets form, was falsified, many many
years ago, when Edwin Hubble discovered, that the so called Nebular
Clouds, which formed the basis of said hypothesis, were in fact, not
Nebular clouds as first thought, but were instead, galaxies?

No, I didn't think so.
walking away (shaking head)

On Aug 24, 12:54*pm, Yousuf Khan wrote:
Raghar wrote:
On Aug 14, 10:37 pm, wrote:
On Aug 12, 6:25 pm, Raghar wrote:

If he had problems with creation of diverse star system, there might
be problem with his simulations. Did he computed energy radiation and
chemical interaction in the system? Unlikely. However these things are
important.
The article does not mention any of that, perhaps you should contact
the authors yourself?

Would they have computing power and will to do these simulations?
These are quite expensive (in terms of computing power), and special
hardware doesn't help much.

Well, perhaps you can contact the original authors from the article I
posted? Perhaps you can find them through a google search?

There is always 128 bit integer, or better. Basically arbitrary
precision is just slow, however not impossible. (128 bit numbers are
quite nice for finding artefacts of simulation.)

The 128-bit FP number data type has not yet been standardized by the
IEEE. Until it is, you can't meaningfully compare simulations with it.

Well, but, the reason adaptive A.I., microcomputers, parallel
processors,
CD, DVD+rw,, holograms, lasers, masers, fiber optics, PV Cells,
optical computers, GPS, laser printers, Ebooks, Blogs, USB, and
Robots were invented,
is because if you wait for IEEE to standardize data typing, you
might
just as well wait for the sun to explode.




However when they'd have bug in program they are using because of one
of pure computer engineering/programming problems, theirs models would
be quite far away from reality, and they will not know about that.

I doubt that precision will make a whole lot of difference to their
simulations. As the original authors said, they did 100 different
simulations of something like 1 million particles each. They were just
roughly trying to correlate their simulations real world solar systems,
for categorization purposes. They weren't saying that those solar
systems evolved *exactly* like that.

* * * * Yousuf Khan

On Aug 24, 4:59 pm, Rick wrote:
Are you aware that the _actual basis, for the nebular hypothesis, you
know the current theory on how planets form, was falsified, many many
years ago, when Edwin Hubble discovered, that the so called Nebular
Clouds, which formed the basis of said hypothesis, were in fact, not
Nebular clouds as first thought, but were instead, galaxies?

No, that's not right at all! There were certain objects thought to be
nebulas which were later proven by Hubble to be actually outside of
our own galaxy; they turned out to be separate galaxies of their own.
Hubble by no means disproved the existence of all nebulas. All Hubble
did was prove that some of the nebulas were other galaxies; in the
process he also discovered that other galaxies exist. For example, the
Andromeda Galaxy was for ages known as the Andromeda Nebula.

Yousuf Khan