Friam

Query about rating agencies/groups

Classic

List

Threaded

32 messages Options

Owen Densmore

Re: Query about rating agencies/groups

Administrator

Well, lets start the usual friam whining. Hey, you left out foobar++! But to be fair, I'd be fine if you replaced java with c/c++.

The point, obviously, is to give a span of languages that hit the main points.

-- Owen

On Jul 31, 2010, at 5:15 PM, Douglas Roberts wrote:

I guess you're not interested in teaching languages appropriate to HPC implementations, Owen. C++ and MPI...

--Doug

On Sat, Jul 31, 2010 at 4:47 PM, Owen Densmore <[hidden email]> wrote:

Given the constraints and goals, my approach would be to teach that there are many languages in different environments, but that they share many features (loops, conditionals, types, ...).

Then I'd pick the following areas:
Command-line programming: Bash & Python
File/Text manipulation, ssh login, regular expressions, commands

System Programming: Java
objects, GUI, Applets, types
Web Server Programming: PHP
client - server networking architecture, http requests, how it won over java

Web Client Programming: Javascript
DOM, AJAX, html, css

That may look like a lot, but it covers most programming environments and goals. And the design issues would pop out when discussing the environments in which these languages excel.

I would NOT go into a lot of detail (clearly!). Instead I'd generalize what they have in common, and possibly use cheat-sheets which have 80% of the important syntax.

The bash/python initial work would also have a lot of pragmatic elements: how to login and use a remote unix box (bash), and the historic evolution of awk, perl, and now for many, python. I'd note that python does not have a native gui (but is considered the best "pseudo-code" by theoreticians -- see sagemath.org) thus the transition to java would have even more meaning. The two web languages would clarify the tcp/ip world we live in and most do not understand.

The goal is to leave the students with a language framework from which they can choose how to proceed in future work.

-- Owen

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Douglas Roberts-2

Re: Query about rating agencies/groups

I guess I was whining. While I'm at it, I'd like to note (whine) that very few universities seem interested in, or capable of, teaching high performance computing methodology.

--Doug

On Sat, Jul 31, 2010 at 10:55 PM, Owen Densmore <[hidden email]> wrote:

Well, lets start the usual friam whining. Hey, you left out foobar++! But to be fair, I'd be fine if you replaced java with c/c++.

The point, obviously, is to give a span of languages that hit the main points.

-- Owen

On Jul 31, 2010, at 5:15 PM, Douglas Roberts wrote:

I guess you're not interested in teaching languages appropriate to HPC implementations, Owen. C++ and MPI...

--Doug

On Sat, Jul 31, 2010 at 4:47 PM, Owen Densmore <[hidden email]> wrote:

Given the constraints and goals, my approach would be to teach that there are many languages in different environments, but that they share many features (loops, conditionals, types, ...).

Then I'd pick the following areas:
Command-line programming: Bash & Python
File/Text manipulation, ssh login, regular expressions, commands

System Programming: Java
objects, GUI, Applets, types
Web Server Programming: PHP
client - server networking architecture, http requests, how it won over java

Web Client Programming: Javascript
DOM, AJAX, html, css

That may look like a lot, but it covers most programming environments and goals. And the design issues would pop out when discussing the environments in which these languages excel.

I would NOT go into a lot of detail (clearly!). Instead I'd generalize what they have in common, and possibly use cheat-sheets which have 80% of the important syntax.

The bash/python initial work would also have a lot of pragmatic elements: how to login and use a remote unix box (bash), and the historic evolution of awk, perl, and now for many, python. I'd note that python does not have a native gui (but is considered the best "pseudo-code" by theoreticians -- see sagemath.org) thus the transition to java would have even more meaning. The two web languages would clarify the tcp/ip world we live in and most do not understand.

The goal is to leave the students with a language framework from which they can choose how to proceed in future work.

-- Owen

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

--
Doug Roberts
[hidden email]
[hidden email]
505-455-7333 - Office
505-670-8195 - Cell

Marcus G. Daniels

Re: Query about rating agencies/groups

Douglas Roberts wrote:
> While I'm at it, I'd like to note (whine) that very few universities
> seem interested in, or capable of, teaching high performance computing
> methodology.
Nah. In the coming years, scalability and reliability challenges of
exascale computing will challenge most HPC doctrine, anyway.

Marcus

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Owen Densmore

Re: Query about rating agencies/groups

Administrator

In reply to this post by Douglas Roberts-2

It will be nice when we can try HPC easily at home, so to speak. In order for an individual to set up a HPC system of some sort, most of us would have to buy systems we don't already have. In a class, we could have a set of 4-person projects where each person contributes their laptop as a node.

But still, it'd be awkward, and I suspect the only realistic solution would be amazon clusters:

http://aws.amazon.com/ec2/hpc-applications/

At one time, this was also true for client-server computing, but web hosting solved that nicely.

If you had to remove one of the proposed environments:

Command-line programming: Bash & Python

System Programming: Java

Web Server Programming: PHP

Web Client Programming: Javascript

to include HPC, which would it be?

I guess the obvious would be Java, as suggested before and replace "System programming" with "System and HPC programming". Hmm..that'd open up the possibility of GPU programming, which would cover both systems and graphics. Maybe we're onto something here!

-- Owen

On Aug 1, 2010, at 6:47 AM, Douglas Roberts wrote:

I guess I was whining. While I'm at it, I'd like to note (whine) that very few universities seem interested in, or capable of, teaching high performance computing methodology.

--Doug

On Sat, Jul 31, 2010 at 10:55 PM, Owen Densmore <[hidden email]> wrote:

Well, lets start the usual friam whining. Hey, you left out foobar++! But to be fair, I'd be fine if you replaced java with c/c++.

The point, obviously, is to give a span of languages that hit the main points.

-- Owen

On Jul 31, 2010, at 5:15 PM, Douglas Roberts wrote:

I guess you're not interested in teaching languages appropriate to HPC implementations, Owen. C++ and MPI...

--Doug

On Sat, Jul 31, 2010 at 4:47 PM, Owen Densmore <[hidden email]> wrote:

Given the constraints and goals, my approach would be to teach that there are many languages in different environments, but that they share many features (loops, conditionals, types, ...).

Then I'd pick the following areas:
Command-line programming: Bash & Python
File/Text manipulation, ssh login, regular expressions, commands

System Programming: Java
objects, GUI, Applets, types
Web Server Programming: PHP
client - server networking architecture, http requests, how it won over java

Web Client Programming: Javascript
DOM, AJAX, html, css

That may look like a lot, but it covers most programming environments and goals. And the design issues would pop out when discussing the environments in which these languages excel.

I would NOT go into a lot of detail (clearly!). Instead I'd generalize what they have in common, and possibly use cheat-sheets which have 80% of the important syntax.

The bash/python initial work would also have a lot of pragmatic elements: how to login and use a remote unix box (bash), and the historic evolution of awk, perl, and now for many, python. I'd note that python does not have a native gui (but is considered the best "pseudo-code" by theoreticians -- see sagemath.org) thus the transition to java would have even more meaning. The two web languages would clarify the tcp/ip world we live in and most do not understand.

The goal is to leave the students with a language framework from which they can choose how to proceed in future work.

-- Owen

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

--
Doug Roberts
[hidden email]
[hidden email]
505-455-7333 - Office
505-670-8195 - Cell

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Owen Densmore

Re: Query about rating agencies/groups

Administrator

In reply to this post by Marcus G. Daniels

On Aug 1, 2010, at 8:52 AM, Marcus G. Daniels wrote:
Douglas Roberts wrote:
>> While I'm at it, I'd like to note (whine) that very few universities seem interested in, or capable of, teaching high performance computing methodology.
> Nah. In the coming years, scalability and reliability challenges of exascale computing will challenge most HPC doctrine, anyway.
>
> Marcus

Agreed! It should be great fun when our laptops have 12 processors and 2 GPUs. Clearly that's driving many of the programming changes at Apple and MS. I understand VM designs help ( such as OS X LLVM: http://llvm.org/releases/download.html ), as do language changes such as Java's lambda expressions which are designed for MP. And of course our earlier chat about Go applies.

Doug: would you be OK if we did not include C/C++ if we did include Go? Have you got it running on your linux boxes? It builds nicely on Darwin, so I'm sure it'd be a snap. I'd like your view on how it would fit into HPC environments like you work with.

-- Owen

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Marcus G. Daniels

Re: Query about rating agencies/groups

In reply to this post by Owen Densmore

Owen Densmore wrote:
> It will be nice when we can try HPC easily at home, so to speak.
AMD's OpenCL toolkit scales well on Opteron system. Also, Snow Leopard
comes with an OpenCL implementation and scales well for small problems
on the GT9400 and GT9600 found in a MacBook Pro.

http://developer.amd.com/gpu/atistreamsdk/pages/default.aspx

From a educational point of view, the problems of GPUs (high latency)
are a benefit because an experimenter has to learn how to deal with an
`unbalanced' design. In the future, more and more communication will
be on chip, but asynchronous. Finding ways to deal with that will be a
challenge, and the people that have the skills to do it will be at an
advantage in the HPC market.

There is a lot of useful investment that can be done at a small scale,
and doesn't require having access to tens of thousands of boxes and
fancy interconnects. The hard issues for coarse-grained communication
will be related to resilience, when nodes go down or give bad results --
in large part figuring out algorithms for adapting to a compute fabric
that is not fixed. Even researchers that do have access to such
fabrics don't have direct control over them. A given, more or less,
and you deal with it.

http://www.nvidia.com/object/cuda_opencl_new.html
http://www.alphaworks.ibm.com/tech/opencl/faq
http://www.pgroup.com/support/downloads.php

Marcus

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Douglas Roberts-2

Re: Query about rating agencies/groups

OpenCL doesn't do distributed message passing parallelism, only data parallel. From Wikipedia: http://en.wikipedia.org/wiki/OpenCL

OpenCL provides parallel computing using task-based and data-based parallelism.

I doubt that message passing distributed computing will be easy in our lifetime(s). Which is just fine by me -- job security!

--Doug

On Sun, Aug 1, 2010 at 9:25 AM, Marcus G. Daniels <[hidden email]> wrote:

Owen Densmore wrote:

It will be nice when we can try HPC easily at home, so to speak.

AMD's OpenCL toolkit scales well on Opteron system. Also, Snow Leopard comes with an OpenCL implementation and scales well for small problems on the GT9400 and GT9600 found in a MacBook Pro.

http://developer.amd.com/gpu/atistreamsdk/pages/default.aspx

>From a educational point of view, the problems of GPUs (high latency) are a benefit because an experimenter has to learn how to deal with an `unbalanced' design. In the future, more and more communication will be on chip, but asynchronous. Finding ways to deal with that will be a challenge, and the people that have the skills to do it will be at an advantage in the HPC market.

There is a lot of useful investment that can be done at a small scale, and doesn't require having access to tens of thousands of boxes and fancy interconnects. The hard issues for coarse-grained communication will be related to resilience, when nodes go down or give bad results -- in large part figuring out algorithms for adapting to a compute fabric that is not fixed. Even researchers that do have access to such fabrics don't have direct control over them. A given, more or less, and you deal with it.

http://www.nvidia.com/object/cuda_opencl_new.html
http://www.alphaworks.ibm.com/tech/opencl/faq
http://www.pgroup.com/support/downloads.php

Marcus

Marcus G. Daniels

Re: Query about rating agencies/groups

Doug writes:

> /OpenCL provides parallel computing using task-based and
> data-based parallelism. /
>
> I doubt that message passing distributed computing will be easy in our
> lifetime(s).
OpenCL provides a partial ordering of event dependencies, or in the
simplest case, an in-order queue of tasks or data parallel operations.
These operations can be kernel invocations (function calls) or
asynchronous memory transfers between address spaces. A device is free
to implement the latter using remote directory memory access (RDMA)
transfers, such as modern MPI implementations would do. A PCIe
connected GPU is in some ways a standalone system, and I don't see a
reason why a moderate sized cluster couldn't be completely handled by
OpenCL. There have been some projects in that direction like clumpi
(OpenCL under MPI).

Since GPUs have such ridiculously fast floating point performance,
learning how to program them is a relevant skill in HPC. OpenCL is not
an integrated single source solution, but at least it is an affordable
place to start (the software is free), and doesn't lock you in to one
vendor. I could imagine higher-level languages targeting OpenCL as a
sort of assembly language. But, if you want to understand how to make
performant code, one needs to have an idea how this stuff works..

Marcus

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Prof David West

Re: Query about rating agencies/groups

In reply to this post by Owen Densmore

Owen,

re: "they share many features, (loops, conditionals, types)"

Smile when I say this---

but, good Smalltalk programmers would almost never write a loop (instead they would use do: or select: or send some similar message to a collection), would use only simple conditionals (no boolean or nested IFs and never a case statement), and - because they are not brain dead - would never use types. [No, a class is not a type, even an abstract data type. At most it is a 'prototype' ala Lakoff and Johnson. It is also an "object factory." -- At least this is the way it was supposed to be. Sigh!]

davew

On Sat, 31 Jul 2010 16:47 -0600, "Owen Densmore" <[hidden email]> wrote:

Given the constraints and goals, my approach would be to teach that there are many languages in different environments, but that they share many features (loops, conditionals, types, ...).

Then I'd pick the following areas:

Command-line programming: Bash & Python

File/Text manipulation, ssh login, regular expressions, commands

System Programming: Java

objects, GUI, Applets, types

Web Server Programming: PHP

client - server networking architecture, http requests, how it won over java

Web Client Programming: Javascript

DOM, AJAX, html, css

That may look like a lot, but it covers most programming environments and goals. And the design issues would pop out when discussing the environments in which these languages excel.

I would NOT go into a lot of detail (clearly!). Instead I'd generalize what they have in common, and possibly use cheat-sheets which have 80% of the important syntax.

The bash/python initial work would also have a lot of pragmatic elements: how to login and use a remote unix box (bash), and the historic evolution of awk, perl, and now for many, python. I'd note that python does not have a native gui (but is considered the best "pseudo-code" by theoreticians -- see sagemath.org) thus the transition to java would have even more meaning. The two web languages would clarify the tcp/ip world we live in and most do not understand.

The goal is to leave the students with a language framework from which they can choose how to proceed in future work.

-- Owen

On Jul 31, 2010, at 3:22 PM, Ed Angel wrote:

Greetings from a wet Alaska.

I agree with Dave and I’d like to add a few more comments about languages, programming and design that are colored by the difficulties of teaching computer science and engineering in large universities.

Most academic computer scientists would agree with Owen about the advantages of teaching the first course using Scheme. Many schools tried it and it was a failure. The prime reason was that almost no schools have a student body as good or as uniform as MIT. In a typical state university, the first programming course known as CS 1 is taught to a very diverse group, including students who intend to major in CS, all branches of engineering, math, business and a lot of the sciences. Most of them will never take another CS course so it’s not clear that a single Scheme course is the best for all of them. What makes things more complex in NM (and I think this is typical of other states) is that you have very few students who start in a program and stay in it for four years. When you get to the second class, almost half the students took CS 1 elsewhere (CNM, some other community college, NMSU, NMT). If UNM were to use Scheme, it’s highly unlikely that any of the other would follow and in the case of most community colleges would probably have great difficulty in staffing such a course.

What further complicates the picture is the state mandates articulation. Hence, each school must accept the CS 1 that the others teach and cannot require students who transfer in to retake CS 1. This is a huge problem for students even when the two CS 1 languages were C++ and Java. It would be impossible if one school were using Scheme and not the others. At UNM, we don’t list a specific language as a prerequisite for the second CS class which is taught in Java. However, most of the CS majors take the first class is in Java although there is a python section. The engineers mostly take a first course using the programming language in Matlab. The computer engineers take the Java version of CS 1 and then quickly switch over to C and C++.

Even the better schools are driven somewhat by the market. But I think once you see why Scheme (or equivalent) doesn’t work, there are no perfect choices. The market push for Java is there in some industries and not others. A lot of the market still wants C++ but it is a terrible teaching language (and a lot of industry sees C++ as the lesser of evils, not as the language they would like to have). Although some schools focus on a single language others give superficial knowledge in many.

The accrediting boards were aware of these issues and rather than requiring a particular language in CS 1, defined the requirements as proficiency in one language and familiarity with at least one other paradigm. As well intentioned as this may be, it’s not clear that is has worked out as hoped. At UNM, students become proficient at Java, learn both C and Scheme at the second level and eventually study logic and functional programming. But in the end most of them become proficient in the language they started which in the case of Java can be a problem for them in the job market depending on the industry and to working in areas such as graphics, game technology and CAD. But if we still used C++, they would have problems with other areas and other industries.

To get back to some of the issues Owen and David raised, the goal should be to teach computational thinking and design. If that is the case, starting in Scheme would be better than what we do now. When we teach it as a second level course, it doesn’t really get used nor are the ideas adequately reinforced by later courses. It’s an aspect of a much larger problem that is a focus of those of us concerned with the problems the country is facing with STEM education, namely the replacement of fundamental ideas of computer science with a focus on what is often called computer literacy. In higher ed, it manifests itself as stressing proficiency in some programming language rather than in the underlying principles. In K-12 education, where the problem is far more serious, computer science ideas have pretty much vanished from the curriculum and from teacher training. In NM, very few schools even teach a CSs course. In those that do, the course is designed to prepare students for the AP Computer Science exam which has become a Java proficiency exam. That presents computer scientists with a difficult dilemma in that while they are opposed to the course, they are hard pressed to oppose the only “computer science” in the schools.

Finally there is the design question. I spent a lot of time my last few years at UNM trying to deal with design in the CS curriculum. At the college level there are serious issues of whether you can teach design and if colleges and universities have faculty who can do it. A more serious issue is the siloing that characterizes most universities. Engineering schools tend to think they own design but the reality is that design has gotten crowded out of the engineering curriculum by the increase in the new material that most departments feel they have to teach. Accrediting agencies now require capstone courses which try to get some design back in the curriculum but it’s not the same as having design integrated throughout the curriculum. In my experience the one school that truly understands design and is good at teaching it is Architecture. Studio artists are also very good at design and problem solving but at UNM, the College of Fine Arts doesn’t see design has part of their curriculum. It will be interesting to see what happens with the new College of Santa Fe. As for CS, it always seemed to me that design was fundamental to any programming course and to our overall program. But that view met vociferous objections from the engineering faculty who believe that “design” is what separated engineering from sciences including computer science.

Sorry for the ramble. The sun just came out and I’m going back on vacation.

Ed

On 7/29/10 1:45 PM, "Prof David West" <[hidden email]> wrote:

Owen,

Speaking as an academic, I agree with you that too many schools believe they need a single language and are driven by pure market conditions - i.e. what language will look best on a graduate's resume.

In my program we require students to demonstrate proficiency (write thousands of lines of code) in four languages: a scripting language, a procedural language, a declarative language, and an object language (and no, Java is not even close to being an object language). [We have not decided if functional is a fifth group or a variation on procedural or declaritive as far as the "thinking paradigm" required.] Which language is chosed within those groups is mostly irrelevant, except as it best suits a specific problem domain (.e.g C or C++ for telecom).

Interestingly, the only empirical research of which I am aware as to what is a Good Answer - was done at Carelton (in Canada): Their research showed that students learning Smalltalk as their first language picked up a second language (did not matter if it was procedural or declarative) about 50% faster than if the first language was Pascal (remember that one?) - and Pascal was created specifically to be a good teaching language. Conversely if students took a procedural (C, Java, Pascal) as their first language, it took them almost 100% longer to become proficient in Smalltalk and about 60% longer to become proficient in Lisp. COBOL was an interesting anomaly - If COBOL was the first language, Smalltak came much easier, but Lisp was still a big cognitive leap.

The Carleton research also showed that Smalltalk-First led to a deeper understanding of programming in general.

The biggest drawback to Smalltalk-First arises from the fact that, in Smalltalk, you never use things like nested or Boolean IFs and no loops and cyclomatic complexity was an order of magnitude lower on average - so moving to a different language almost always felt like descending into swamp of tedious and verbose complications.

[[ While I am being opinionated - Design (decomposition and distribution of knowledge and behavior), not programming language, is the real key - proper design makes the coding almost trivial. (I am in the midst of a hot and heated argument with my colleagues in the Software Craftsmanship (they are all about code and programming) movement about this right now - and doing a presentation to about a hundred professional developers in Mpls next week. ]]

dave west
============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Marcus G. Daniels

Re: Query about rating agencies/groups

Prof David West wrote:
>
> but, good Smalltalk programmers would almost never write a loop
> (instead they would use do: or select: or send some similar message to
> a collection)

As would almost any good programmer in any language. Most languages
have these in their standard libraries.

> would use only simple conditionals (no boolean or nested IFs and never
> a case statement), and - because they are not brain dead - would never
> use types.

Don't agree. All a statically typed language needs is a reflection
system, generics built upon that, and macros. From that you can make
dynamic types when you need them. For that effort, you 1) get to know,
at compile time when you've missed cases or misused interfaces, and 2)
performance. C++ programs, for example, that always use pure virtuals
instead of private member functions have more bloat and are slower.

For new users, it's useful to have a lower correctness wall and less
up-front work. For example, by up-front work, I mean the cumbersome
mechanics of declaring complex union types. Then they can make progress
right away and stay engaged. Languages like Smalltalk, Scheme, R, or
Pure are appealing for these users. For professional development, I
think having internal consistency checks, like strong typing, is
useful. Type inference and macros (e.g. Haskell) mostly eliminate the
redundant mechanical aspects.

Marcus

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Eric Charles

Agent based modelers database

In reply to this post by Prof David West

Hey all,
My former Ph.D. adviser, Jeff Schank, is trying to start a database to connect agent based modelers from a variety of different disciplines. He is putting together a very impressive website at http://www.agent-based-models.com/blog/?page_id=172

If you would like your name added to the list, or want to suggest other people who should be listed, send an email, with a link to your/their webpage to: [hidden email]

Thanks,

Eric Charles

Professional Student and
Assistant Professor of Psychology
Penn State University
Altoona, PA 16601

Eric Charles

Re: Agent based modelers database

I'm not sure how Jeff evaluated other abm-website options. He is much deeper in the abm community than I am, and felt there was a niche to fill. In particular, I believe he thought there was room for a site that listed people doing abm in as many areas as possible. One potential for the site would thus be as a sort of academic dating site - it could connect people from different academic disciplines / problem domains, whose traditional disciplinary boundaries would otherwise keep them unaware of each other. The people he has contacted thus far have been enthusiastic and supportive, so clearly he is finding other people who are not tapped into whatever other resources you are referring to.

Because I am honestly ignorant of the field, what are the major sites that you feel currently serve this need? (Or related needs?)

Thank you,

Eric

P.S. Just to plug the link again: http://www.agent-based-models.com/blog/?page_id=172 Shoot him an email if you are interested in being listed: jcschank@...

On Tue, Aug 3, 2010 01:21 PM, "glen e. p. ropella" <[hidden email]> wrote:

ERIC P. CHARLES wrote  circa 08/02/2010 11:37 PM:
> My former Ph.D. adviser, Jeff Schank, is trying to start a database to
> connect agent based modelers from a variety of different disciplines. He
> is putting together a very impressive website at
> http://www.agent-based-models.com/blog/?page_id=172

Just out of curiosity, why did he create yet-another-abm website when
there are so many he could contribute to?

-- 
glen e. p. ropella, 971-222-9095, http://agent-based-modeling.com