abstract = "Recent work showed how an expression in a functional
programming language can be compiled into a massively
redundant asynchronous spatial computation called a
distributed virtual machine. A DVM is comprised of
bytecodes reified as actors undergoing diffusion and
communicating via messages containing encapsulated
virtual machine states. Significantly, it was shown
that both the efficiency and the robustness of
expression evaluation by DVM increase with redundancy.
In the present work, spatial computations that become
more efficient and robust over time are described. They
accomplish this by self-replication, which increases
the redundancy of the elements of which they are
comprised. The first and simplest of these
self-replicating DVMs copies itself by reflection; it
reads itself from a contiguous range of memory. The
remainder are quines. As such, they reproduce by
translating and transcribing self-descriptions. The
nature of the self-descriptions and of the translation
and transcription processes differ in each case. The
most complex self-replicating DVM described represents
a fundamentally new kind of artificial organism, a
machine language program reified as a spatial
computation that reproduces by compiling its own
source-code.",
notes = "Department of Computer Science, University of New
Mexico, Albuquerque, NM 87131