[extropy-chat] sjbrain calcs

[Spike]

> Eugen Leitl
> Subject: Re: [extropy-chat] sjbrain calcs
> 
> 
> Mike Lorrey wrote:
> > 
> > I am wondering, since we are dealing with micron sized particles,
> > whether you could actually have a sperical cloud rather than a dust
> > ring without lots of collisions of particles. I am wondering if
> 
> You can't, of course. Spike, don't you know what happens at 
> the poles, where all orbits intersect?

The orbits need not have a common intersection point.  In fact
the orbits need not intersect at all.  

As a thought experiment, consider the MBrain scenario Robert B. 
and I calculated last time he was here.  You have a number of nodes
orbitting the sun, flat disks a centimeter across and perhaps
100 microns thick.  These could orbit the sun in a number of 
concentric but not coplanar rings, with no two rings having
exactly the same radius.  Imagine each ring consisting 
of nodes spaced about a meter apart.  Inside that ring is another ring
with 1 meter smaller radius, with the orbit plane tilted a
microradian, inside that another ring tilted another microradian,
etc.  From the point of view of any node, the nodes to her right
or left are stationary, whereas the nodes a meter sunward are
passing south to north at a relative velocity of about 6 cm per
second (if I recall correctly) faster than a snail but slower
than a tortoise.  Any inavertent collision need not be a violent
event.  The nodes one meter anti-sunward would be passing
6 cm north to south, and those 2 meters outboard, 12 cm/sec
and so on.  This is a traditional Bradburian MBrain, since
the orbits envelope the star.  

The SJBrain is a related idea:
a mass of nodes that orbits a common center in about a 
minute-diameter sphere an hour out from the star.  It too
would consist of many concentric rings, noncoplanar and
with no two rings having the same diameter. 

> I don't see how you can actively control particle orbits 
> without using solar sails (+PV panels as dual use), and this implies
pretty big 
> ones. At least somethign the size of your palm, depending on how it
scales a 
> lot bigger.

Not necessarily.  The station-keeping ability of a solar
sailing node is not dependent upon its area but rather upon
its thickness.  For a given thickness of material, the
mass and the area (thrust available from light pressure)
scale directly together, or cancel.  If I can make nodes
1000 atoms thick, it doesn't matter if they are a square 
micron or a square millimeter in area, right?

So I am proposing station-keeping in the SJBrain using 
light pressure from the sun, but now I am going on to 
propose that this same trick still works even if the nodes 
are in a roughly spherical region of space an hour from the 
star.  In fact the station-keeping might be easier out there, 
since the forces involved are much smaller.

I am further suggesting that the common-center orbit speed could
be made fairly low, since the resistance to gravitational
collapse of the nodes could be assisted by electromagnetic
repulsion.  But even if it isn't, consider traditional orbit-
mechanics, not assisted by either light pressure or EM forces.
Jupiter's moon Metis (the closest one) orbits about a quarter
of a light second from the CM of Jupiter and goes around in
about .3 days, so an object orbitting 30 light seconds out
would do it in about .3/(.5/30)^1.5= 140 days.  Then as one
goes inboard toward the SJBrain center of mass, the orbit 
tangential velocity actually goes down linearly, assuming no 
central mass.

So no Brownian motion, Mike, no need for chaos, if we know how 
to manipulate matter at the atomic scale.  Those 30-50 billion
atom nodes will figure out where they want to be, then go
there and stay there.

Then of course if you made use of the available light pressure 
and EM forces, the relative velocities of the nodes could be 
made smaller than they would be otherwise, if one should decide
that large relative velocities are undesireable.  I suppose one
could use light and EM forces to *increase* relative velocities
of the nodes too, should Robert or someone explain a reason
why such a thing would be desireable.

Robert and other nano-groksters, how do we balance the desireability
of more nodes vs larger nodes?  What kinds of energies do you 
suppose we would need to run the kinds of calculation that would
be done on those nodes, and what kinds of distance-related
latencies could we tolerate?  I suppose it depends on what it
is we want the SJBrain to do.  Certain tasks, such as searching
for record primes, would favor fewer, farther spaced nodes, being
highly tolerant of communications delays between nodes.  But
other tasks need faster communications between nodes.

spike