[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Distributed agents
From: |
Marcus G. Daniels |
Subject: |
Re: Distributed agents |
Date: |
21 Jun 1999 13:28:21 -0700 |
User-agent: |
Gnus/5.070084 (Pterodactyl Gnus v0.84) Emacs/20.3.10 |
>>>>> "NG" == Natalie Guay <address@hidden> writes:
NG> Does someone knows if I can distribute agents on several
NG> computers and keep control on another one???
The Holy Grail for Swarm is to realize our `logical model of
concurrency' in a system integrated with popular thread and message
passing libraries. The result of such a system would be that Swarms
could be transparently and efficiently (even automatically) migrated
across heterogenous computing resources.
Since a basic design objective of Swarm is to preserve the visibility
of all the fine-grained discrete event structure of a simulation, this
requires a lot of hacktivation energy to get done and get done right.
If your main objective is to explore a large parameter space using a
bunch of computers, you might investigate writing an executive
`experiment Swarm' and then use object serialization to move agents around.
While this is a significant amount of work, it might make sense if it
is otherwise infeasible to explore the space using a tool like Drone,
remote execution shell scripts, or just manual oversight
(e.g. throwing xterms and watching with `top').
==================================
Swarm-Support is for discussion of the technical details of the day
to day usage of Swarm. For list administration needs (esp.
[un]subscribing), please send a message to <address@hidden>
with "help" in the body of the message.