🚀 go-pugleaf

> From:	UPSALA::"rodrigue-joe@CS.YALE.EDU" 13-SEP-1992 04:31:21.96
> To:	magnus@solaris.astro.uu.se (Magnus Selhammar)
> CC:	
> Subj:	Re: parallel search
>
>>> I am looking for references to Parallel search algorithms on
>>> ordered/partially ordered structures.
>
>> I am also looking for search algoritms, even though my interest is in almost
>> ordered structures.  What surprises me is that it's so difficult to find
>> such methods on searching. Yes, the field is new, but it's a very general
>> problem that should interest many persons.
>
> how do you define this problem?  thanks.


I hope you don't mind that I post your reply. I am very grateful for it since 
you are the only one who replied.

In a volume of N particles, find the n spatially nearest neighbours, 
to every particle. Assume that the nn's to a particle p are ordered 
around p in the computer memory according to, say, a Gaussian law. This
implies a higher probability to find a nn close to p in the memory than far
away in the memory. 
This is a problem in N-body methods where p interact with all other particles
in a finite volume around p, where this volume depends on the distances to
these nn's.
I am doing this on a Connection Machine. With almost ordered I mean that it's
possible to order the particles in the memory according to their spatial
coordinates, but that it is not possible to do it exact.

This is a description of my special problem. I am interested to hear or read
posts from others on searching in structures.

Magnus Selhammar


From: magnus@solaris.astro.uu.se (Magnus Selhammar)
Path: solaris.astro.uu.se!magnus
Newsgroups: comp.parallel
Distribution: world
Followup-To: 
Organization: 
Subject: 
Keywords: 

> From:	UPSALA::"rodrigue-joe@CS.YALE.EDU" 13-SEP-1992 04:31:21.96
> To:	magnus@solaris.astro.uu.se (Magnus Selhammar)
> CC:	
> Subj:	Re: parallel search
>
>>> I am looking for references to Parallel search algorithms on
>>> ordered/partially ordered structures.
>
>> I am also looking for search algoritms, even though my interest is in almost
>> ordered structures.  What surprises me is that it's so difficult to find
>> such methods on searching. Yes, the field is new, but it's a very general
>> problem that should interest many persons.
>
> how do you define this problem?  thanks.


I hope you don't mind that I post your reply. I am very grateful for it since 
you are the only one who replied.

In a volume of N particles, find the n spatially nearest neighbours, 
to every particle. Assume that the nn's to a particle p are ordered 
around p in the computer memory according to, say, a Gaussian law. This
implies a higher probability to find a nn close to p in the memory than far
away in the memory. 
This is a problem in N-body methods where p interact with all other particles
in a finite volume around p, where this volume depends on the distances to
these nn's.
I am doing this on a Connection Machine. With almost ordered I mean that it's
possible to order the particles in the memory according to their spatial
coordinates, but that it is not possible to do it exact.

This is a description of my special problem. I am interested to hear or read
posts from others on searching in structures.

Magnus Selhammar


From: magnus@solaris.astro.uu.se (Magnus Selhammar)
Path: solaris.astro.uu.se!magnus
Newsgroups: comp.parallel
Distribution: world
Followup-To: 
Organization: 
Subject: 
Keywords: 

> From:	UPSALA::"rodrigue-joe@CS.YALE.EDU" 13-SEP-1992 04:31:21.96
> To:	magnus@solaris.astro.uu.se (Magnus Selhammar)
> CC:	
> Subj:	Re: parallel search
>
>>> I am looking for references to Parallel search algorithms on
>>> ordered/partially ordered structures.
>
>> I am also looking for search algoritms, even though my interest is in almost
>> ordered structures.  What surprises me is that it's so difficult to find
>> such methods on searching. Yes, the field is new, but it's a very general
>> problem that should interest many persons.
>
> how do you define this problem?  thanks.


I hope you don't mind that I post your reply. I am very grateful for it since 
you are the only one who replied.

In a volume of N particles, find the n spatially nearest neighbours, 
to every particle. Assume that the nn's to a particle p are ordered 
around p in the computer memory according to, say, a Gaussian law. This
implies a higher probability to find a nn close to p in the memory than far
away in the memory. 
This is a problem in N-body methods where p interact with all other particles
in a finite volume around p, where this volume depends on the distances to
these nn's.
I am doing this on a Connection Machine. With almost ordered I mean that it's
possible to order the particles in the memory according to their spatial
coordinates, but that it is not possible to do it exact.

This is a description of my special problem. I am interested to hear or read
posts from others on searching in structures.

Magnus Selhammar