From: Robert J. Bradbury (bradbury@aeiveos.com)
Date: Mon Oct 21 2002 - 21:02:15 MDT
Scientific American has a great article in the November issue
on the topic of colliding stars ("When Stars Collide").
Unfortunately it looks like its only available online for $.
(Though the movies might be free).
Previously I had thought the lifetime of Matrioshka Brains
was limited to the lifetime of the smallest stars, i.e. small
red dwarfs (M class). If I recall correctly thats around a trillion
years or so. Logically, one would use star-lifting or some other
method to take the sun, which a 10 billion year lifetime G class star,
down to that size to get the greatest longevity. Ok fine. But
according to the SciAm article -- two identically sized main sequence
stars that collide will "merge" within an hour. The result is
a hotter star because some of the surface hydrogen gets driven
into the resulting core so you get a hotter star with about twice
the mass and presumably a stellar atmosphere that may be richer
in metals that were formerly in the cores of the two parent stars.
So, to maximize the longevity (and memory or computational capacity)
of an MBrain here is what you do.
You find the nearest star with a slightly smaller mass than the star the
MBrain normally derives power from (presumably this involves some
consideration of the delta-v you have to apply to put the MBrain on a
collision course with said star). You aim your MBrain at the star.
Someplace far enough away from the target star, the MBrain Uber-lord
instructs all of the MBrain subunits to "Decloak the lifegiver and raise
your radiation shields". The lifegiving star collides with the
unsuspecting (dumb) target star. The merge occurs (generating one would
presume some really great fireworks) [at this point you had better have
your fingers crossed that the simulation got it right because some of the
other outcomes of star collisions of various types are *not* so pretty].
After you have finished listening to the 1812 Overture, the MBrain
Uber-lord gives the order to "Recloak the newborn lifegiver". Ok, now you
have a star about with twice the mass of the originals.
This isn't desirable, so you resume star-lifting again (hauling off the
more useful metals) until you get the star down to a size where metal
lifting isn't very productive or you get back to your original
computational capacity. (Remember the innermost shells of an MBrain have
the fastest computational capacity because they have the smallest
internode distances.) So the "merge" process gained you a lot of material
for increased memory storage but may have cost you some aggregate
"thought" capacity (in real time) because you can't get as close to the
larger star. You actually do have more material for computational nodes
if you were computationally constrained in the first place -- but these
nodes have to be added on the outside layers of the MBrain so they will
think much slower in real time (due to high internodal delays).
Deciding how much material to strip off your newborn lifegiver
gets very complex -- you don't want to plot a course to
increasingly heavy stars since their lifetimes get progressively
shorter *and* it requires wasting a lot more energy to lift
the material out of the gravity well. Presumably there may
be some interesting "pollution" problems here if you are using
a star lifting process that requires ion or molecule expulsion.
Throw too much matter around in the inner MBrain orbits and
you will pollute their internode communication beams.
But the Dyson spin-up schemes to throw off material from
the star's equator would seem to require a *really* large
amount of material to spin up a stellar mass.
I previously thought it was going to be fairly difficult to extend the
lifespan of an MBrain but using this approach its remarkably easy. [He
says tongue in cheek because if the collision changes the direction in
which the newly formed star is headed, the MBrain subunits may have quite
a job of catchup to do.] As smaller stars are more abundant you might take
the strategy of simply making your lifegiver increasingly smaller so you
have to exert less delta-V to aim the MBrain at its next "encounter".
You probably do an analysis of the general direction in which you are
headed and plot out of course that would most rapidly grow you to a
"maximum MBrain size". ["Maximum" MBrain size does vary with the star
type you "choose" to be -- larger, but shorter lived stars, can support
larger MBrains -- for G class stars the maximum size is several light
years.] Alternatively, you could harvest half of the mass of the star,
target another similarly sized star and recover your misspent youth.
If star lifting turns out to be difficult or slow, then a glancing
collision still results in a merged star but it throws up a lot of
material from the stars out into space. Presumably you might manage this
in such a way that you get back a star approximately the size of the
originals but a lot of matter getting lifted out of the gravity wells and
into space in an hour. An express star lifting so to speak.
There is another possibility of simply having the MBrain subunits
hop from one M-class star to another as each lifegiver drops below
the minimal power requirements for what it wants to think about.
In this case you wouldn't collide at all but simply aim the stars
very close to each other so hopping to the next energy source takes
a minimal amount of time.
Given hundreds of billions of possible stars to collide with and knowing
that the longest lived are the most abundant it would seem that this ups
the maximum lifetime of MBrains to at least a hundred billion trillion
(10^23) years.
Robert
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:17:42 MST