Grenier D.
(2000)
|
|
|
II. A new mathematics field for proof and news problems |
|
We push to the forefront these three aspects of proof pratice - status of truth, conjectures and modeling - in the learning objectives of our research work. We will detail and illustrate them from suitable typical problems. II.1. Devolution of problems, stake of truthOne certainly cannot conceive a proof without its relation to a question to be solved, located within a given field of research. In particular, the mathematical situations take a meaning in the course of the various steps : The devolution of a problem is of course directly linked
to the meaning it can take. Probably, for a majority of
pupils, this meaning is very related to the Didactical
Contract. For example, the problem is attached to a concept,
to a chapter of the textbook. The interest of problems which
are not already modelled in a mathematical form is that they
make it possible to consider this meaning outside a precise
mathematical context. We will illustrate our matter by one example. Let us consider the construction of the " Torricelli Point " : Consider the equilateral triangles constructed on each edge of a given triangle. The goal is to show that the three lines joining a vertex of the triangle to the opposite vertex of the opposite equilateral triangle are concurrent. The truth of that result can hardly be questioned,
especially if the figure is drawn with more accuracy. If a
geometry software is used, it can be legitimate to regard
the result as undoubtedly true. The proof then cannot be
regarded any longer as the best way to establish the truth.
However, its role as an explanation remains. " What is the more economic way to link thre points, like below : It would be then necessary to study different occurrences, and thus, one can see that the Torricelli point is not always the solution to the question. II. 2. The specificity of some combinatorial argumentsWe present now some methods which play a central role in discrete mathematics, but which, in our opinion, also have a more general range of validity, such as decomposition/recomposition or structuring of objects, in particular by colouring, or the pigeons' holes principle. We will see that a combinatorial point of view makes it possible to develop a new analysis of certain concepts and tools, as induction - which is a fundamental tool for proving. II. 2.1. Decomposition/recomposition Decomposition/recomposition is a central method in
combinatorics. Paving problems (of which two examples have
already been given) are naturally a privileged field of
application of this method. The principle of "decomposition/recomposition" applied to the concept of area (and more generally of volume), which is then seen as a measurement attached to surfaces and defined by its properties of additivity and invariance by recomposition, allows to solve many situations in a simple manner. One can see the example of Euclid's rectangle, in Grenier (1999). II.2.2. Induction Induction rests on a particular decomposition : the object is divided in two objects : a smaller object satisfying the same hypotheses, plus another "piece". The basic mathematical object under consideration is the set N of integers. In discrete mathematics, proof by induction is very often constructed according to the following scheme : - argue by contradiction : there exists a counterexample (of size n) ; This scheme of proof may allow to obtain an inductive construction of a class of objects starting from more basic terms. On the other hand, to be able to use the " ascending " scheme of the type : P(n0) it is necessary to know how can constructed all the (n+1)
objects from a n-object.. " For what values of n, does it exist a regular n-polygon with integer coordinate vertices (of which one can put the n vertices on the squared grid) ? " The proof consists in establishing the following result : " In the plane, there are no regular polygons with integral coordinate vertices, except squares. "
For this, notice first of all the equivalence between the
existence of equilateral triangles with integral coordinate
vertices and the existence of regular hexagons with integral
coordinate vertices. Proof : II.3. A " fundamental " situationWe suggest here a situation which gives a possibility, in
a spontaneous or guided way, that some of the main
combinatorial methods emerge. We make the assumption that
this situation can be regarded as a fundamental situation
(in the meaning of Brousseau). If it is possible to find a tiling, the question is
solved, but only for the given particular occurrence.
However, in general cases, after several unsuccessful
attempts, a proof is needed. The problem suggested to pupils is whether there exists a paving of a square polymino with an elementary square removed. This situation has been tested at various levels : collège (2nde, 1ère, Terminale), DEUG, teachers during their trainingcurriculum, teachers in activity. The strategies used for solving, the difficulties encountered during various experiments are rather similar. The initial steps adopted to attack the problem are not very distant from those of the professional researcher : experimentation on simple cases, formulation of conjectures and, more rarely, attempts of modeling and structuring. Let us give some more details. a) Tests on small cases (squares of side 2, 3, 4...). While this is not necessarily regarded as a "result", it appears very quickly that a polymino can be paved only when the number of elementary squares is even. One thus considers only the squares of odd side in which an elementary square has been removed. After some tests, it is realized that certain squares do "work": if an elementary square is removed, the remaining part can be paved. These squares are shown in black on the figure. This leads to the conjecture : " when the removed square belongs to the set of black boxes ("good" boxes), it is possible to pave by dominos. Otherwise, it seems that paving is not possible. " b) Modeling and structuring The problem is to pave by dominos; however, what is a
domino? Two adjacent squares. This relation of vicinity can be modelled by the graph : These two polyminos are equivalent with respect to the
problem of paving by dominos. c) Necessary / Sufficient conditions and Decompositions One can thus considerate the question of the diffrence between necessary condition (NC) and sufficient condition (SC). A first necessary condition had appeared from the very beginning : the parity of the number of squares. But this condition was too simple to be regarded as a true NC. In any case, it did not have any chance to be perceived as a good candidate to be a NSC (necessary and sufficient condition). Counterexamples are in fact quickly found, the one with smallest size being : The second NC - equality of the number of black squares and the number of white squares - seems more serious. Amongst other things, it gives a proof which is different from the examination of all the cases, of the non-pavability of found counterexamples. Is it sufficient ? Non connected counterexamples are hardly convincing : However, separate pieces can be joined to obtain "balanced" connected polyminos which are not pavable. One will see, by decomposition, that this condition is also sufficient for solving the case of a square polyminos in which an elementary square is removed. Break up the square into smaller pavable parts (for example in rectangles having an even side). Several partitions can appear, the simplest being the following :
Other partitions leading to smaller squares are also proposed |