Posted by
Hugh Trenchard on
Apr 13, 2010; 2:09am
URL: http://friam.383.s1.nabble.com/leadership-in-flocks-tp4868514p4893829.html
Just as an update and a follow up note on the
cyclist/sperm aggregations, I've developed (and am continuing to
develop) a simple computer peloton simulation and am working through a
series of experiments. Aspects of the simulation apply to certain
sperm aggregates as well (at least I will suggest this).
The plan at this point if for the the sim to
involve these sets of experiments:
1. a "no drafting" set where two or more sets
of agents proceed according to their own intrinsic max sustainable speed with no
capacity to match the speed of agents of other sets;
2. a "weak drafting" set, where weaker
agents can match the speeds of others if in a certain proximity
of faster agents;
3. a "strong" drafting set where agents
actually seek to match speeds of others by following behind others.
I've completed a set of experiments for number 1,
which is the obvious case in which group sorting occurs according
to maximum sustainable output, and the easiest to simulate.
The others are still underway.
My aim is to demonstrate that:
a. group sorting does in fact occur
according to relative differences in power output, and that aggregates occur
because their effective fitness levels are narrowed by a "drafting" effect so
their speeds are identical;
b. there is a correlation between aggregate size
(ie. number of agents), differences in relative maximum sustainable output, and
the time to which group sorting occurs.
The prediction is that the time required
for group sorting to occur increases as aggregate size
increases and relative differences in max output capacities
("fitness") become smaller. In other words where agents are
identically fit, they will all stay in one group and will never sort
(generally); where fitness levels are different, they will sort
rapidly. Drafting facilitates the narrowing of fitness
levels, so even if there are intrinsic differences in output, agents
will aggregate if they can draft such that they travel at the same speed
and at effectively identical output levels.
At the moment, I have in mind that sperm aggregates
fall under the category of "weak" drafting, whereby they randomly/accidently
draft, but are incapable of seeking out drafting positions as are agents in
bicycle pelotons. Pelotons exhibit strong drafting. So, under a
weak drafting model, sperm sorting should occur at some rate faster than
pelotons. In a peloton, especially one in which the entire group consists
of riders of closely matched sustainable outputs (such as a group of
professional cyclists), the group will stay together to the finish (on a
uniform course). In a peloton, the primary cause of group
sorting is the occurrence of points in which drafting benefit is
reduced such that drafting no longer equalizes the entire range of output
levels, such as hills, course obstacles, and cross-winds (I have refered in the
past to these as instabilities in the system). The nature of the
proximity of sperm to one another may mean that their weak drafting results
in very long sort times, such that sort times are nearly the same as would
occur under a strong drafting situation, and that is something I can look for as
well. There may be some surprises along the way.
If I can complete the experiments, I may seek to
get the results published, or I may aim to present them first at the
2010 AAAI Conference and seek publication afterward. Basically the idea is
to establish a model by which predictions can be made for real aggregations, and
the model should be applicable essentially to all aggregates which move at
maximum sustainable outputs and where there is some energy saving component
involved in coupled motion. It won't apply where aggregates move at outputs
significantly below max sustainable outputs, because in those situations even
the weakest agent can keep up with the strongest (the extreme situation is where
they all stand still, or move at the equivalent of a slow walking pace). It can
apply, however, in situations where there is broad range of fitness
levels and only some agents move at maximum sustainable output,
because a small increase in output among the group causes those already at max
output to be sorted "off the back", even when many among the group are not at
max output.
Vladimyr - sorry I haven't responded yet on a
couple of your posts - I've realized I need to knuckle down and really work
through a simulation, and have become focussed on that at this point. I
hope to be involved in this discussion more, however, over time.
Hugh
----- Original Message -----
Sent: Saturday, April 10, 2010 10:16
AM
Subject: Re: [FRIAM] leadership in
flocks
How similar to the
sperm peloton and the cyclist peloton, now we have flocks with leaders and
cliques?.
If each model has a
different organizing principle then why does my simple mind think there are
similarities?
I liked
Hugh Trenchards ideas
the best, there was no need for more than a simple available power assessment
on the part of the individual agent. Sticking the term leadership into the
discussion really puts a strange twist to everything.
Trenchards ideas
would have probably worked for the flocks equally well, and that is truly
interesting. Craig Reynolds 1982? wrote his early Boids paper with only
very simple principles none of which included power or
aerodynamics.
Same organized
behavior but completely different principles. Do we force complex
interpretations where simple ones suffice.
A leader in a
cycling peloton is such a temporary phenomenon that one has to be very careful
how the term it is used. But in the bird flock the leader seems to be part of
a social dynamic which might not actual exist but in the minds of the writers?
Inventing complex
explanations for simple situations seems similar to what conspiracy theorists
practice.
Dr.Vladimyr Ivan
Burachynsky
Ph.D.(Civil Eng.),
M.Sc.(Mech.Eng.), M.Sc.(Biology)
120-1053 Beaverhill
Blvd.
Winnipeg,
Manitoba
CANADA R2J
3R2
(204) 2548321
Phone/Fax
[hidden email]
-----Original
Message-----
From:
[hidden email] [mailto:[hidden email]] On Behalf Of Ted Carmichael
Sent: April 10, 2010 5:39 AM
To: The Friday Morning Applied Complexity
Coffee Group
Subject: Re:
[FRIAM] leadership in flocks
I haven't read the papers all the way
through, but on first blush, I don't see them as contradictory. Either
could be correct.
A "leader" - whether bird or person -
could act first due to internal traits (inclination, ability, imagination) or
external influence. The first implies that the leader is different from
the others in some way, while the second implies only a situational
difference: circumstance rather than inherent traits.
Once the leader acts, this creates space
for the other birds/people to act similarly, and follow the leader. The
followers must have had the same inclination towards this action, because they
end up doing it, too ... they just weren't over the tipping point yet.
There was something missing that kept them from acting first. The
leader's action clearly provides the missing element, and so all the followers
perform the same action.
The remarkable thing about the flocking
models, such as the one in JASS, is that they show that leadership doesn't
have to be due to an internal trait. It may simply be a situational
difference among very similar agents. Before these models were put
forth, the prevailing view was that leadership is always endogenous to the
leader. Now, at least, we can consider other possibilities, whether or
not they end up being correct.
-t
On Fri, Apr 9, 2010 at 8:57 PM, glen e.
p. ropella <[hidden email]>
wrote:
sarbajit roy wrote circa 10-04-09 06:34
AM:
> The religious
grouping I belong to had cause to study/discuss this about 150
> years
back (concerning flocks of men not birds). The leader of the
faction
> in opposition to mine (which means my faction vehemently
disagrees with his
> view) had this to say
That quote from your opposition seems to
fall in line with the nature
article, the idea that particular birds/humans
(presumably with
particular traits, inbred or learned) turn out to be
leaders. I take it
from your statement that you agree more with the
jasss article, that
leaders with no particularly exceptional traits emerge?
Right?
Of course, to even have this discussion, we have to allow
ourselves the
metaphor between human cliques and bird flocks...
--
glen e. p. ropella, 971-222-9095, http://agent-based-modeling.com
============================================================
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
============================================================
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
============================================================
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