705 Allotted calculations, their possibilities and difficulties.
Nuenen, 8 March 1979. 2 pages. transcription
A von Neumann computer consists in a calculator and a memory, close to each other to avoid wiring time delays. One can think of dividing the calculations in parts and have them executed by more machines and have information sharing between them.
One must distinguish between a few machines at large distance (example monetary transactions) and many small machines as close as possible (example wind tunnel simulation). In the latter case wiring problems will only allow limited connections.
Not all types of calculations are fit: mathematical ingenuity will be needed to perform iterative calculations differently.
Geographically distributed systems will have difficulties with timing, central clock signals will take time to reach each machine, and the state of the system is undefined. Different proof techniques are required.
—MB
733 Reaction on account of "Experimental post-graduate education for informatics engineer at the THE (Eindhoven February 1980)"
Nuenen, 12 March 1980. 4 pages. transcription
Report written by the "Management commission ad hoc curriculum experimental discipline informatics science".
Dijkstra first slogs the very bad Dutch in which the report is presented. It should have been a policy account, the result is a poor simulation, or a persiflage, of that. The language is so bad, that often it looks as if the text has to hide the fact that there is no contents.
He treats the report paragraph per paragraph, attacking what he considers the misunderstandings among the commission's members, who in general consider the computer science fundamentally similar to those of well established sciences: you study the matters and then you can go out engineering. Computer science is not very much fact based, but this intellectual image never even occurred to the members. He one by one treats and criticizes the proposed qualifications and the sections with examples.
Finally he asks what one should do with such a piece of snot? It's no use doing it all over, the commission is unable to do better. He advices to discharge the members with dishonour and find a job for the secretary where he cannot do such harm.
—MB
753 On a theorem by Lambek and Moser
Nuenen, 9 October 1980. 3 pages.
States a theorem of Leo Moser and Joachim Lambek (1954), about mapping an increasing sequence of integers to its functional "inverse", and a resulting partition of integers into two sets. A proof of the theorem by a picture, from a book of R. Honsberger, is given.
The subject is taken up in EWD758 "An intriguing example", EWD759 "A somewhat open letter to D.A. Turner", and EWD770 "D.A. Turner's reply", about a program for the function inversion task in the functional (or "applicative"= programming language SASL, and in EWD776 "Lambek and Moser revisited", with a different proof by Dijkstra, by transforming an imperative ("iterative") program.
—GR
776 Lambek and Moser revisited
Nuenen, 8 February 1981. 7 pages. transcription
Proves a theorem of Leo Moser and Joachim Lambek (1954) about mapping an increasing sequence of integers to its functional "inverse", and a resulting partition of integers into two sets. The theorem is proved by deriving a program for the computation of one sequence from the other, and then transforming the program.
This is related to EWD753, where the theorem has been stated, and a different proof has been reported.
The subject has also appeared in EWD758 "An intriguing example", EWD759 "A somewhat open letter to D.A. Turner", and EWD770 "D. A. Turner's reply", about a version of the program in SASL (a "functional" or "applicative" programming language).
—GR