8. Crown Prince Shotoku Said that Peace Is the Most Supreme
How far have we progressed?
This
essay is composed of merely a part of results of our study by the present of March 1, 1996
and it is not only that the study is not completed but also that there remains plenty of
pages not described except chapter titles. Of course we are not in the stage to write a
postscript then providing a simple summary at the present and a prospect from now on, let
lay down the pen for the time.
The
figure below shows rough conceptual scheme of our Hamiltonian experimentation at the
present stage.
- Terminal graph division is a method dividing a graph into right subgraph and left
subgraph and proceeding experiments recursively, and it brings drastic effects to reduce
experimentation costs. For the reason of searching cost and others, it may be implemented
in such restriction that only 2 or 3 terminal should be selected. Terminal graph division
is executed on the top layer of Hamiltonian experiments.
- Three modes experiment includes three kinds of experiment of IO-division experiment,
mesh-diagram experiment and series-parallel experiment and it is implemented principally
to support short-circuit graph method. (1)Due to the three modes experiment, some kinds of
graphs are separated as non-Hamiltonian graphs by judgment with non-retrial. (2)Making
three modes experiment for the original graph or short-circuit graphs, it can detect early
contradictions which cannot be detected by factor experiment in the stage. (3)Further, it
promotes auto-determination process in operations of arbitrary selection with constraint
conditions on three modes experiment.
- IO-division experiment is a method dividing a graph into left and right subgraph and
examining interfaces of passage points or branches of these two subgraphs and also it
includes terminal division experiment and cut-set division experiment. As either of them
cannot dispose all divisions on the graph for the reason of costs, it experiments
divisions of such branches or points as just determined or removed by arbitrary selection.
- Mesh-diagram experiment, giving each points of a graph point value and calculating
points score and branches score, can regulate distribution of branches of a graph
conforming to the constraint conditions. It is adoptive to experiment a graph of terminal
division difficult.
- Series-parallel experiment detects topological contradictions of a graph by
experimenting a reduced graph obtained by Hamiltonial series-parallel conversion.
- Short-circuit graph method is a fundamental experiment for Hamiltonian experimentation
based upon factor experiment and owing to the short-circuit method, since each stage of
Hamiltonian experiment can be executed recursively as a Hamiltonian experiment of
short-circuited and reduced graph, it makes possible to implement three modes experiment
into experimental process. Short-circuit method makes thorough examination in all numbers
in the worst cases.
- Non-temporal experimentation, using retrieval transformation on revival trees, breaks
through the limitation of short-circuit method and realizes the experiments never
retrograding. In non-temporal experimentation, it is able to generate a retrieval map
constituted as a semi-point graph from deadlocked determinative paths of short-circuit
graph, to derive revival trees from this map to resolve the contradictions and further to
obtain retrieval paths without deadlock. In the case of such experiment as deviates from
the range of deadlock map, non-temporal experimentation is executed recursively.
- In short-circuit graph method, when it lockouted at the stage where just obtained a
Hamiltonian path, after taking out maximum Hamiltonian ties, the switch panel game
introduced in the 1st chapter is played out to inquire a Hamiltonian tie from these
materials. It is not clear whether switch panel game is able to be realized or not at the
present. Switch panel game should be formalized as one small problem of Hamiltonian
experiments. It is to be considered that the Hamiltonian experiment problem has it's most
peculiar characteristic at the point where dividing the problem into sub-problems does not
necessarily make the problem small. In this meaning, we cannot stop regarding that it is a
completely non-deterministic problem. However, we suppose that the switch panel game is
obviously transferred to other dimension than Hamiltonian experiment problem, and so we
confirm that it is hopeful enough to resolve this small problem.
- If it is possible to carry out the re-experiments never retrograding by non-temporal
experimentation on revival trees, there exists a possibility that we can constitute
Hamiltonian experiments with the combination of only short-circuit graph method and
non-temporal experimentation (without using three modes experiment). At the present, there
remains some parts not formalized yet in the ending phases of non-temporal experiment.
- Possibly, such a suggestion might be considered that can't we play the switch panel game
itself on the retrieval map? (In other words, to implement the game as non-temporal
experiment). We somewhat worry that the composition of the small problem of switch panel
game seems resemble to an non-decidable problem known as Post's correspondence problem.
What shall we do from now on?
There
exist so many matters left behind. Problems not yet resolved (switch panel game, complete
judgment with non-retrial), problems with unfinished parts (experiment deviated from the
range of deadlock map), problems not formalized yet (re-experimental graph method
including non-temporal experiment, cross IO contradiction), problems not obtained their
proofs (Hamiltonian series-parallel conversion, etc., problems belong to such category as
unsolved, uncompleted, not formalized are of course problems unproved), problems not
disposed (experimentation for undirected graphs), problems without formalized algorithms
(search algorithms of terminal division, cut-set division), etc. To resolve all of these
problems is the present subject.
In
these problems, there are such two classes as is able to reach a settlement in time and
needs something than time. According the arrangement of the chapter titles of this essay,
we anticipate that non-determinability problem of chapter 3, algorithms of chapter 4 and
calculation costs of chapter 5 are able to be resolved basically only if times are given.
In
parallel with formalizing algorithms, we should compose programs to demonstrate them, and
both of such works must be done simultaneously, as actual proofs of rightness of
algorithms and measurements of practical abilities. Experiments for such graphs that rare
to occur in hand-writing scale sample like cross IO contradiction may be easily realized
on a computer. Simulations on computer might be useful to extract peculiar problems not
found yet. To optimize the algorithms, especially to inquire the resolutions of graphs
with Hamiltonian ties as rapid as possible, some kind of optimization or improvement would
come to be necessary.
Dealing
with those subjects, how long time should we estimate to solve completely Hamiltonian
Circuit Problem? What we understand just now is only that at least the time must be in the
order of polynomial time algorithms. On the stage when this essay is completed, we must
compose a formally publishable revel paper and publish it. The greatest difficulty at the
present is that we are in such status that we lack all of primary elements indispensable
to accomplish them and yet we cannot abandon these things.
Assuming that the Hamiltonian Circuit Problem is a problem equals to
the method to control one's fortune and a Hamiltonian circuit represents a ring = peace,
what we have to prove from now on are enough with the following two propositions.
- Only one method to control the fortune is to select the best way at every moment.
- If there is a problem to be solved for bringing peace, it is the problem of
resource distribution.
If it is so, to solve the Hamiltonian Circuit Problem is indeed a matter of time for
us.