Tuesday
Nov022004

Niko's Nature by Hans Kruuk

This is an academic biography of Niko Tinbergen, the pioneeer of ethology, the study of animal behaviour.

Tinbergen made his name by devising and carrying out some beautifully simple experiments on nest-finding in wasps and birds, on pecking behaviour in gull chicks, and on aggression between male sticklebacks.  He also seemed to have a real knack for attracting first class students and inspiring them to do first class work.  The names of just a few of his students will show what I mean: Desmond Morris, Richard Dawkins, Aubrey Manning, John Krebs, Marian Stamp Dawkins.  Tinbergen also wrote many popular books and articles on animal behaviour, and even made a few films.

However, Tinbergen was somewhat flawed both a scientist and as a person.  He didn't back up his work with the quantitative and statistical arguments that are now thought to be necessary, and he suffered increasingly from clinical depression in the latter half of his career, to the extent that some of his later students hardly ever saw him.  In spite of these flaws, his research group still managed to flourish, apparently largely thanks to his assistant Mike Cullen who looked after his students for him and provided the quantitative expertise that he lacked.  Tinbergen also showed good sense in his reaction to the potentially devastating criticisms of his experiments by the American Danny Lehrman: he invited Lehrman across to Oxford to help sort out what could be salvaged.

In 1973, the value of Tinbergen's work was recognized when he was awarded the Nobel prize for medicine, jointly with his old friend Konrad Lorenz, and Karl von Frisch.  Ten years later, Tinbergen suffered a couple of strokes but these had the fortunate side-effect of curing his depression and so he was able to enjoy himself during his last few years.  He died in 1988.

The relationship between Tinbergen and Lorenz was an interesting one, particularly in how it was affected by their experiences during the war.  In that respect it bears comparison with that between the physicists Niels Bohr and Werner Heisenberg.

A good read if you are interested in biology.  It should inspire you to search out copies of Tinbergen's papers and books.

Sunday
Oct102004

The Return of the Egyptian Geese

This morning at 09:00am I went for another walk around the Reading University lakes with Zoe.  There is much more bird activity on the lakes in mornings than later in the day, especially now that the evenings are getting darker. 

The 'resident pair' of egyptian geese (the rufous phase male with its lame grey phase female) were standing on the weir at the west end of the largest lake when they suddenly took off and flew off towards the centre of the lake.  Then we saw five rufous egyptian geese fly in overhead, circle the lake and splash down on the water.  These could well have been the resident pair's first and second generation offspring coming back to the lake on which they were reared.  We hadn't seen them for several weeks.  Maybe they have been staying on some of the other lakes in the Reading area.  Anyhow, a few minutes later there was a commotion when a rufous form one chased off several  of the others.  Maybe this was one of the parents telling its offspring to go and find their own lake - I understand that birds can be quite ruthless when it comes to encouraging their young to fend for themselves.

We also saw a lone cormorant, as we did back in August, standing on the same submerged rock.  Over the past couple of weeks I have seen a lone cormorant on three occasions, twice on the lakes between Frimley and Farnborough, and once over the Kennet in Reading.  I suppose it is possible that these sightings are all of the same bird moving around from one body of water to another.

Saturday
Oct092004

Golden Pheasant

On Friday afternoon my wife Liz, her sisters, and my daughter Zoe saw a male golden pheasant (Chrysolophus pictus) in the grounds of Reading University.  It was being slowly shooed down the drive from Foxhill House by the driver of a van.  Apparently it had escaped from a nearby garden where it was being kept as a pet.  These spectacular birds are native to central China but were introduced to Britain as an ornamental bird for aviaries and parkland.  I was still on my way home from work and so missed all the fun. 

Tuesday
Sep282004

Kingfishers Again

Last week I saw at least one of the kingfishers more or less each day as I was walking to and from the station, alongside the Blackwater.  On one occasion I came across one one of them perched still on a branch over the river.  It stayed for about ten seconds, allowing me to get a good view of its plumage, then it flew off.  It was tiny: little bigger than a robin.

On Sunday I took Zoe to see them but there was no sign of them.  Maybe that was because there were too many people around, it being the middle of the day.  Next time I will take her in the early morning. 

Today, I saw one of them again, in the morning and in the afternoon.

Wednesday
Sep222004

Undefined Values should remain Undefined

In "On Using Conditional Definitions in Formal Theories" J. R. Abrial and L. Mussat [in D. Bert et al (Eds.), ZB2002, Lecture Notes in Computer Science. pages 242-269, Springer, 2002] review various approaches to handling undefined values.  One of these, approach (3) on page 247, involves "claiming that a term of the form E/0 denotes a genuine real number, but that this number is unknown".  Abrial and Mussat go on to express the opinion that it is "hard to accept" that "3/0 denotes a real number".

I agree. In my approach 3/0 denotes a value but we cannot tell what that value is, not even whether it is a real number or not:  It is completely undefined.  The requirement that the value must be a real number appears to arise from the inappropriate enforcement of typing requirements.  The problem is that the type invariant is being (incorrectly) enforced in the post-condition, even when the pre-condition does not hold.  For example, consider:

∀x,y. x∈R ∧ y∈R
    (x/y)∈R

Here x/0 is required to always be in R (assuming x and y are in R), whereas what we really want is for the (x/y)∈R to be guarded as follows:

∀x,y. x∈R ∧ y∈R
    y≠0 ⇒
        (x/y)∈R

Here x/y is no longer forced to be a real number when y=0, indeed it is completely unconstrained.  This is obvious, isn't it?  Why then do some people find it necessary to constrain such undefined values?  Don't they realize that this unconstrainedness is necessary for composing specifications and for reducing the complexity of the rules to reason about them?

Actually the approach (3) in Abrial and Mussat is due to D. Gries, "Foundations of Calculational Logic" [in Mathematical Methods in Program Development, M. Broy and B. Schneider (Eds.), Springer, 1996].  It can also be found in C. Morgan, Programming from Specifications, Prentice Hall, 3rd Edition, 1998, section 6.7.