undreamt-of possible uses?
1 What's the problem?
The properties of physical, chemical, biological and other systems fluctuate chaotically. This sometimes becomes only clear if one increases the sensitivity of the measuring instruments or if one makes specific conditions for the system. Let's suppose, we wish a specific state of a property of a such system. One can perhaps shift this property with the following method into a wanted direction. This method imitates the natural evolution: An unwanted fluctuation direction of the property is hindered permanently (e.g. by short-time destroying energy supply) and an wanted fluctuation direction supported (e.g. by short-time reduction of energy supply).
The method
is not new. In essential points it corresponds to "Simulated Annealing"
. Read literature: genetic algorithms, chaos theory, neural nets. However,
own enquiries showed that varied (especially technical) POSSIBLE USES are
hardly known. However, the APPLICATIONS described here have hypothesis
character.
This method should be made known here so that it can be checked,
advanced and APPLIED. Therefore, the APPLICATIONS are the emphasis of this
text. Specialist of all departments (physicists, chemists, technicians,
mathematicians, information scientists, physicians, chaos researchers and
other) should be addressed. APPLICATIONS would be conceivable on many fields.
Therefore, the reader himself should think about this topic. The author
did this in several places, too. No ready theory is described here. This
would exceed the possibilities of the author.
The reader should read this text from the viewpoint of a "not
knowing one" - he needs to only have technical general education. Because
this text was written for "not knowing ones", experts who already know
this all could have the impression that the wheel should be invented here
again. That is not so. This is a (relatively) popular scientific text that
should make topic interesting for USERS. Later the reader can read the
corresponding specialist literature via genetic algorithms, chaos theory,
neural nets and so forth. The heading of text said already, that the main
topic are the APPLICATIONS and therefore the word "APPLICATION" was always
capitalized here. At first it is attempted to make the method plausible
as a imitation of the natural evolution in this text. Possible APPLICATIONS
are then discussed later. Therefore, it is a question of a simple imitation
of the natural evolution - therefore, too the name: Artificial Evolution.
Common characteristics with genetic algorithms and other procedures
are not coincidental. However, the method was named only "Artificially
Evolution" here in text because the APPLICATIONS have hypothesis character
and conflicts with the existing and recognized applications and theories
should be avoided.
For motivation, some POSSIBLE APPLICATIONS are suggested here which are explained at length later. The following examples contain the essential of the entire text.
(The TEACHER (electric switching, computer or similar things) controls energy supply into dependence of the measured values.)
1.) Changing a property of a solid state
+------------------------------------------------+
|
TEACHER (electrical circuit, computer ...) |
|
He realizes the following:
|
|
The less the difference to desired
|
|
state of the property is, the energy supply |
|
to the solid is the less.
|
+-----|--------------------------------------|---+
V
^
V +--------------+
^
+-------|-------+
| solid | +-----|----------------+
| energy supply
|->>-|(its property |->>-| measuring instruments|
|
| | fluctuates) | | (for the
measurement |
+---------------+
|
| | of the property) |
+--------------+ +----------------------+
2.) The increase of the efficacy of a transformation of energy e.g. a solar cell
+------------------------------------------------+
|
TEACHER (electrical circuit, computer ...) |
|
He realizes the following:
|
|
The higher the made voltage,
|
|
the less the additional energy supply
|
+-----|--------------------------------------|---+
V +--------------+
^
V |
energy |
^
+-------|-------+
| converting | +-----|----------------+
| additional
|->>-| solid |->>-| measuring
instruments|
| energy supply
| |(the made |
| (for the measurement |
+---------------+
| voltage | | of the made
voltage) |
| fluctuates) | +----------------------+
+--------------+
3.) Displacement of a chemical balance
+-------------------------------------------------------+
|
TEACHER (electrical circuit, computer ...)
|
|
He realizes the following:
|
|
The less the difference to desired
|
|
concentration of the material is, the additional |
|
energy supply is the less.
|
+-----|--------------------------------------|----------+
V +--------------+
^
V |materials
of a| ^
+-------|-------+
| chemical | +-----|---------------+
| additional
|->>-| reaction |->>-|measuring instruments|
| energy supply
| | (the
| |(for the measurement |
+---------------+
| concentration| |of the concentration)|
| fluctuates) | +---------------------+
+--------------+
4.) Support of healing processes in the medicine
+-------------------------------------------------------+
|
TEACHER (electrical circuit, computer ...)
|
|
He realizes the following:
|
|
Whenever an illness signal is registered,
|
|
energy supply to the solid state is increased,
|
|
otherwise reduced.
|
+-----|--------------------------------------|----------+
V +--------------+
^
V |
diseased |
^
+-------|-------+
| part of the | +-----|-----------------+
|
|->>-| body |->>-| measuring
instruments |
| energy supply
| |( illness |
|(for the measurement |
+---------------+
| signal | | of the
illness signal)|
| fluctuates) | +-----------------------+
+--------------+
5.) The increase of the emission strength of a radioactive substance
+-------------------------------------------------------+
|
TEACHER (electrical circuit, computer ...)
|
|
He realizes the following:
|
|
The greater the current emission strength is,
|
|
energy supply to the radioactive substance
|
|
is adjusted the less.
|
+-----|--------------------------------------|----------+
V +--------------+
^
V |
| ^
+-------|-------+
| radioactive | +-----|-------------------+
|
|->>-| substance |->>-| measuring instruments
|
| energy supply
| | (emission | |(for
the measurement |
+---------------+
| strength | |of the emission
strength)|
| fluctuates) | +-------------------------+
+--------------+
6.) Fictitious example: Evidence of the anti gravity (It will not certainly become implemented. It should clarify only that maybe new things would be possible)
+-------------------------------------------------------+
|
TEACHER (electrical circuit, computer ...)
|
|
He realizes the following:
|
|
The less the current weight is,
|
|
energy supply to the solid
|
|
is adjusted the less.
|
+-----|--------------------------------------|----------+
V +--------------+
^
V |
| ^
+-------|-------+
| solid | +-----|--------------------+
|
|->>-|
|->>-| Piezoelectric
|
| energy supply
| | (weight |
| crystals (for the |
+---------------+
| fluctuates) | |measurement of the weight)|
|
| +--------------------------+
+--------------+
This was a
short summary of possible APPLICATIONS. Who understood the above examples
already has already registered the essential of the following text. One
could now already begin with experimentation. In the following text, method
is only made plausible and one thinks how to optimize the method. One could
attempt for example to build up interaction concatenation within the solid
state which has causal connections with wanted property. One could achieve
this by mixing different materials.
2 The natural evolution
2.1 A simple model of the natural evolution
At first a
simple principle of the natural evolution is handled. This principle is
imitated later during the Artificial Evolution.
Random modifications occurred during the evolution of nature permanently.
New properties of the natural systems resulted permanently by these random
fluctuations. These new properties had to defend their existence against
the environment. The properties that were not successful were exposed to
a DESTROYING influencing by the environment and such things which were
successful were SUPPORTED. The evolution is controlled by this one principle
not alone, of course. But it is sufficient for our considerations.
A simple model of the natural evolution:
+------------------------------------------------+
| ENVIRONMENT
|
|
|
| It
realizes the following:
|
| The
better the new properties are
|
| successful
in the environment, the
|
| destroying
influencing onto the system |
| is
the less and the supporting
|
| influencing
is the more largely.
|
+-----|--------------------------------------|---+
v
^
v
^
+-------v-------+
+--------------+
^
| EFFECT
on | | NATURAL
| ^
| the system:
| | SYSTEM |
+-----|------+
| destroying
|->>-| (its |->>-|
EFFECT |
| or supporting
| | properties | |
on the |
| INFLUENCING
| | fluctuate) | | ENVIRONMENT|
+---------------+
+--------------+ +------------+
2.2 An example: The origin of the species
New species resulted by mutations (= fluctuations of nature). These species had to maintain themselves in the former environment, in particular against the other species. If the new species were successful in the available environment i.e. if they did not destroy their own existence conditions (e.g. food) or if they did not become an easy booty of the other species through their new properties, their existence was supported. Otherwise, their existence was influenced in destroying manner.
3 The Artificial Evolution
3.1 A general model of the Artificial Evolution
Analogous to the model the natural evolution (compare with above) a model of the Artificial Evolution is proposed now. The essential difference consists in the fact that the ENVIRONMENT was replaced by a TEACHER. Therefore, the human being now controls indirectly the destroying resp. supporting influencing. He employs a TEACHER for it. The TEACHER can be an electrical circuit, a computer, a mechanical device, a natural feedback or even a manual control. The TEACHER observes the SYSTEM with his MEASURING INSTRUMENTS. In such moments, in which the SYSTEM fluctuates into a wanted direction, the system is INFLUENCED by the TEACHER in supporting manner. Otherwise, the system is INFLUENCED in destroying manner.
Diagram of an artificial evolution process that has the destination to influence the fluctuating property PROP of the system in desired manner:
+------------------------------------------------+
| TEACHER
(electrical circuit, computer ...) |
|
|
|
He realizes the following:
|
|
The less the difference to the desired
|
|
state of PROP is, the destroying influencing |
|
onto the system is the less
|
|
and the supporting influencing
|
|
is the more largely.
|
+-----|--------------------------------------|---+
V
^
V +--------------+
^
+-------|-------+
| SYSTEM | +-----|-----------------+
| destroying
|->>-|(Its property |->>-| MEASURING INSTRUMENTS |
|resp.supporting|
| PROP |
| (for the measurement |
| INFLUENCING
| | fluctuates) | |
of PROP) |
+---------------+
+--------------+ +-----------------------+
This general
model can be applied to different systems. The imagination has no boundaries.
At first we want to find out whether it is possible that the Artificial
Evolution functions in the described manner. For this purpose, we look
at a special case: Artificial Evolution of network systems, in particular
of solids.
3.2 A more specific model of the Artificial Evolution: network systems, e.g. solids
What is a network system? A network system is a system which consists of many single elements which interact. These elements are designated as cells here. Examples of network systems are: a solid with its interacting atoms and molecules, every living being with its cells, the mankind with the individuals, a neural network with its neurons and much more.
At the following, solids are mainly handled as a special case for network systems.
Diagram of
an Artificial Evolution process that has the destination to shift the physical
property PROP of the solid into a desired direction:
+------------------------------------------------+
|
TEACHER (electrical circuit, computer ...) |
|
|
|
He realizes the following:
|
|
The less the difference to desired
|
|
state of PROP is, the energy supply
|
|
to the solid is the less.
|
+-----|--------------------------------------|---+
V
^
V +--------------+
^
+-------|-------+
| solid | +-----|----------------+
| energy supply
|->>-|(its property |->>-| MEASURING INSTRUMENTS|
|
E | |
PROP | | (for the measurement
|
+---------------+
| fluctuates) | | of PROP)
|
+--------------+ +----------------------+
THE ENERGY SUPPLY
· Energy supply is used here for destroying influencing (see general model). A high energy supply causes a great destruction strength to the structure of the solid. A small energy supply supports the existing structure.
· It is necessary that energy supply (e.g. penetrating irradiation, application of heat, (alternating-) fields) occurs evenly onto the entire volume of the solid. A possibility is also handled later with which energy supply occurs in a differentiated manner.
· It
is necessary to carry off the energy from the solid permanently again,
otherwise, the solid would heat itself up and an evolution would not be
possible. The natural energy emission would often be sufficient, otherwise,
the solid would be had to be cooled.
THE SOLID
· The property PROP of the solid must fluctuate as described above, otherwise, this evolution process does not function. Fluctuations occur in the case of every physical property (noise). This sometimes becomes only clear if one increases the sensitivity of the measuring instruments. One could put the solid into states with which fluctuations occur in a increased manner, e.g. phase transitions. The chaotic random fluctuations must have their cause in the inner structure of the solid state.
THE TEACHER
· The energy supply in dependence to the current value of PROP must be controlled by a "teacher". One could define a function dE/dt = f (PROPDesired - PROP) which the teacher must realize.
· The
above evolution principle is imitated here:
a) PROP fluctuates into the undesired direction a time. -->
Parts of the inner structure of the solid are DESTROYED
or restructured by energy increase.
b) PROP fluctuates into the desired direction another time -->
The existing inner structure of the solid is SUPPORTED
by decrease in the destroying energy supply.
3.3 A computer simulation of the model
It's not the point here to verify that an artificial evolution process always functions at solids. It should only be shown here that it COULD function and that it is useful to research this topic. Therefore, only a simple two-dimensional model is considered be here.
MODEL OF A FERRO MAGNET
1 1 0 0 0 1
0 0 1 1 1 1
1 1 1 0 0 0
0 0 1 0 0 1
Spin set upwards means 1. Spin set down means 0.
· Neighboring
spins have the tendency to adjust itself of the same kind. Each spin has
a specific value (1 or 0). A spin is selected randomly new. P is the probability
that its neighbor spin selected randomly has the same value within a defined
UNIT OF TIME. P is a measure of the link-up of two Spins.
· The
distribution of the spins fluctuates by this link-up.
(Secondary remark: The value P is dependent on the temperature: The larger the temperature of the ferro magnet, the less is the link-up P because of thermal influencing. The probability P is not the same one which is employed in literature at the Renormalization Groups.)
THE EVOLUTION DESTINATION (FERRO MAGNET)
The teacher should realize the following simple evolution destination:
1 1 1 0 0 0
1 1 1 0 0 0
1 1 1 0 0 0
1 1 1 0 0 0
Therefore,
order should be made. This destination will not be achieved completely
certainly. However, it should at least be available more 1-spins on the
left side than on the right side at the end. I.e. the property PROP : =
NumberLeft - NumberRight should be as great as possible.
This concrete destination certainly has no practical uses. However, it
should be proved that an artificial evolution is perhaps possible. In addition,
this example should represent similar destinations (shift of balances,
e.g. charge division).
THE CONTROL OF THE EVOLUTION PROCESS
Suppose, one
wanted to carry out this experiment in practice. A teacher (e.g. in the
form of an electronic switching circuit) would have to be able to measure
the fluctuating spin distribution PROP of the ferro magnet by means of
measuring instruments. In addition, he would have to be able to influence
the ferro magnet by means of energy supply. The smaller the difference
of PROP to the destination is, the teacher sets the energy supply the less.
The larger the difference of PROP to the destination is, energy supply
is adjusted the greater. The kind of energy must be able to change the
direction of the spins in random manner (e.g. a external magnetic alternating
field with high frequency or heat energy or a energy form, which one converts
itself into heat energy). Main thing it is destroying in random manner.
Presumable the reflex speed of the teacher would have to be very high (maybe
1000000 per second?) so that he can react rapidly to the fluctuating spin
distribution PROP.
But at the following, a simulation should only be carried out.
A SIMULATION PROGRAM FOR THE EVOLUTION PROCESS (FERRO MAGNET)
(The complete program is in the appendix)
Start
1.) (Simulation
of the sensors)
Determine
the number of the 1-spin on the left side and the number of the 1-spin
on the right side! Form the difference
PROP := NumberLeft
- NumberRight!
It is simulated
that the physical property PROP is measured. PROP is the difference of
the 1-magnetization between left and right side. The larger PROP is, the
closer we are distant by the destination set up above.
2.) (Simulation
of the teacher)
Determine
the energy that should be provided to the ferro magnet with the following
formula:
E := k1 *
(PROPDesired - PROP). k1 and PROPDesired are constant factors.
The more closely
we are distant by the destination (PROPDesired), i.e. the more largely
PROP, the supplied energy should be the less. This corresponds to the above
principle of an evolution process.
3.) (Simulation
of the energy supply:)
Determine
the number N of spins that change their direction randomly with the following
formula:
N := k2 *
E (k2 is constant)
Select N times
a spin randomly and change its direction! (The larger the supplied energy
is, the more spins change their direction randomly i.e. the destruction
of structure is the larger.)
4.) (Simulation
of the natural energy supply:)
Select Nnat
times a spin randomly and change its direction! The ferro magnet includes
the heat energy of the environment. I.e. the spins also change permanently
their direction independent of the energy supply of the TEACHER.
5.) (Simulation
of the interactions:)
The interactions
with its neighbors are realized for the individual spins: With the probability
P, a neighbor spin is put onto the same value as the initial spin.
Go to the beginning
RESULTS (FERRO MAGNET)
· An
effect can be proved.
· The
greater the interaction with the neighboring particle (the larger W) is,
the results are the better. (P=0.99--> PROP=12, P=0.7-->PROP=11, P=0.3-->PROP=6,
P=0.1-->PROP=3, P=0-->PROP=2)
· If
the middle or the maximum energy supply which the TEACHER determines on
the basis of PROP is too high, one can hardly prove a result.
· If
PROPDesired was put too high (that is if the demands were too great), the
results became more badly.
Therefore, one should further research the topic and make practical experiments. This is the ONLY conclusion from the results.
Similar effects
occur certainly too at spin glasses. The model of the ferro magnet represents
SIMILAR and ANALOGOUS models in which ones interactions occur between the
cells in a such manner that the state of a cell is endeavored to be transmitted
to the neighboring cells.
3.4 When is a system capable for an artificial evolution?
Which conditions in the interior of the system are necessary for an evolution process?
Natural evolution:
During every
evolution, the properties RESERVATION STRIVING and CHANGEABILITY had a
big role to play in the inner conditions of the system. If PRESERVATION
STRIVING would not be available, every evolution progress would immediately
collapse again. Not even parts of that what existed would remain. If CHANGEABILITY
would be not available, evolution progress and corrections would be impossible.
During the origin of the species for example, PRESERVATION STRIVING manifested
in striving for preservation of species in particular by cooperation and
reproduction (Positive feedback: the more living beings existed, the more
living beings existed). If PRESERVATION STRIVING would not be available,
the least noxious environmental condition would have the result, that the
new species become extinct immediately again. Not even few specimens of
a species would have received the chance to continue the phylogenetic tree
in their direction. PRESERVATION STRIVING alone however would only have
the result, that the world would have been flooded with this one species.
In this way alone, an evolution would not have been possible. The CHANGEABILITY
had to be added which expressed itself in the form of destruction ability
by environmental conditions and mutations. The CHANGEABILITY carried on
to this that too other species got at least the possibilities to continue
the phylogenetic tree in their direction.
PRESERVATION STRIVING and CHANGEABILITY must have a balanced proportion.
No one of both ones may predominate.
Application to the Artificial Evolution:
It might be PROP the property which was modified by an artificial evolution process. The following conditions are necessary in the interior of the system for an artificial evolution process:
1. Within the
system there is a PRESERVATION STRIVING of those conditions which cause
the property PROP.
2. A relative
CHANGEABILITY of these conditions must be simultaneously available.
We look at
the example of the ferro magnet again: The property PROP := NumberLeft
- NumberRight depends on direction of the individual spins. Therefore,
the direction of the spins is a condition for materialization of property
PROP.
And this condition would have to have a PRESERVATION STRIVING (see 1.).
PRESERVATION STRIVING means the following in our case: If a cell has spin
1 for example, probability is great that this cell retains this state in
the nearest moment. This is really so because the state of the cell is
transmitted to the neighboring cells. If the neighboring cells have occupied
the same state now however in the nearest moment, that increases probability
that the initial cell occupies this state also again. A cell so to speak
infects the neighboring cell and this cell affects the initial cell in
the same manner again back. One also speaks here of positive feedback.
This also matches with known theories of self-organization after what positive
feedback favors self-organization. Computer simulation showed too: The
larger the positive interaction is (the larger P), the result is the better.
This is only valid in a specific range. The interaction may become not
too big. On the other hand, one can already observe a small effect in the
case of P = 0 (no interaction with the neighboring cells). I.e. the pure
"inertia" of the spins (to retain its state for a short moment) is already
sufficient to fulfill the feature of PRESERVATION STRIVING.
The CHANGEABILITY of the conditions (see 2.) is also guaranteed at the
ferro magnet. Energy supply can influence the direction of the spins in
random manner.
The FLUCTUATIONS of the system which are necessary for the evolution process could have their cause in PRESERVATION STRIVING and CHANGEABILITY. PRESERVATION STRIVING, this time in the form of positive interactions, leads to building up of the system into the one direction at first. But then it collapses because of the feature of CHANGEABILITY again. After this everything begins from the beginning.
PRESERVATION STRIVING can manifest itself in very different ways, for example: simple inertia of a property in the case of change attempts or positive interactions or the striving for spatial transmission of these properties.
Therefore, the qualities PRESERVATION STRIVING and CHANGEABILITY are necessary for an evolution process. Every existing system shows these qualities probably so that one should carry out experiments with every kind of systems. However, there are systems in which ones these qualities especially protrude and which one better would be suitable for an artificial evolution. In addition, an artificial synthesis of systems with these qualities would be possible. But this is handled in a later section.
3.5 What happens during an Artificial Evolution?
It is difficult to understand generalizingly and vividly the complicated interactions which occur in the interior of a network system. However, the attempt should be carried out here at least to clarify how functions an artificial evolution process.
1. Explanation attempt:
In the case
of wanted fluctuation directions, the teacher influences the system in
supporting manner. That has the result, that the current state of the system
is extended and consolidated (especially in case of PRESERVATION STRIVING
in the form of positive interactions). But this state is not retained into
every eternity because of quality of CHANGEABILITY. It can be corrected
later by a destroying influence of the teacher.
In the case
of unwanted fluctuation directions, the teacher influences the system in
a destroying way, what is effective because of CHANGEABILITY. However because
of the quality of the PRESERVATION STRIVING, this does not lead to a complete
relapse into the old state, but parts of the old structure, those were
effective to now, survive. Therefore, the next evolution phases can occur
already with a higher level, with regard to the wishes of the teacher.
The next evolution phases with a more higher level etc. Therefore, the
Artificial Evolution strides forwards.
2. Explanation
attempt:
Every system
is endeavored to hold a state of least energy. This state is achieved if
the property PROP of the system satisfies best the wishes of the teacher,
because the destroying energy supply is then least.
3. Explanation
attempt:
These evolution
processes also function in nature. One should imitate this simply. Imitating
of natural phenomena often produced benefit. One needs not always understand
everything completely.
3.6 Synthesis of evolution capable systems by consideration
How can one produce systems which are especially suitable for an Artificial Evolution?
One could begin
with the wishes which one pursues. Suppose, we have the destination to
develop a specific markedness of property PROP. Now one would have to think
about the following question: How can one make in the interior of the system
conditions those cause the property PROP over a causal effect chain or
have some connection with PROP and those are characterized simultaneously
by a PRESERVATION STRIVING. These conditions must also be at least in part
destructible again (CHANGEABILITY), for example by energy supply.
At net systems, in particular at solid states, the cause of an external,
global property PROP is presumably at the smallest cells of the system,
that is at the atoms, molecules or particles and its interactions. PRESERVATION
STRIVING and CHANGEABILITY would have to be available at the cells. Further
the global property of magnetization at the ferro magnet (see above) also
had its roots in the microcosm at the spins. PRESERVATION STRIVING and
CHANGEABILITY were also available at the spins.
How can one
make the quality of PRESERVATION STRIVING at the particles? Or more specific:
How can one make a positive interaction between the particles? At first
we look at the example of the ferro magnet again. One could symbolize the
interactions at the spins in simplified manner as follows:
s --> s
|
| (Effect chain between two neighboring
s)
<--
Therefore, there is an effect chain of an atom to the neighbor atom and again back. This effect chain is relatively short and, in addition, only consists of homogeneous elements. Now one could attempt to also make longer effect chains with different elements:
a -->
b --> c --> d --> e --> a
|
| (Effect chain between two
<-- e <-- d <-- c <-- b <--
neighboring a)
The interaction to the initial element a occurs in the same way as the link-up to the neighbor element. Therefore, we shorteningly write only the one direction in future:
The same like above:
a -->
b --> c --> d --> e --> a (Effect chain between two
neighboring a)
Such chains
are also possible with which the effect chain is actual no chain. The effects
could run cross and cross between a,b,c,d...
so that in this case a positive interaction comes out only in the final
effect. The a,b,c,d... could be atoms, molecules,
particles or specific states within these particles. It could be even time
consecutive states of a specific particle.
In other words: One could attempt to make evolution capable solid states
by MIXING of several substances, whose atoms, molecules or particles interact.
One should not see the term MIXING too narrowly. Perhaps one would be able
to put the chain a,b,c,d... within a single
molecule.
In the end this effect chain must somehow also achieve the property PROP,
which should be modified according to our wishes - symbolic:
|--> PROP
|
|
a -->
b --> c --> d --> e --> a
Of course, such interaction processes can not be held at operation without continuous energy supply EI (interaction energy). Some element in this chain must be susceptible to this energy form. Further, it could be possible that at specific elements a,b,c,d... an effect on neighbor elements occurs only at specific external physical conditions COND. The destroying energy supply ED of the teacher which one must intervene somewhere into the chain is very important, of course (CHANGEABILITY).
ED -->| |--> PROP
EI
-->| | |
| <-- COND
| | |
|
a -->
b --> c --> d --> e --> a
This is only
a simplified model, of course. The effects could run cross and cross between
the elements a,b,c,d,e... ED,EI,COND
and PROP could not only apply to a single element in each case, as in the
illustration, but could stand in connection with several elements simultaneously.
Further ED and EI could be the same energy form, i.e. the energy which
the teacher provides could be also the energy supplier for inner interactions.
One could describe the last illustration as follows with less place expenditure
and less paint work:
a --> b
--> c --> d --> e --> f --> a, EI --> b, ED --> c, d --> PROP, COND -->e
SUMMARIZING REMARKS ABOUT THE NETWORK SYSTEM
Among other
things, it should be attempted to make a conceptual clarity before we go
on.
Because we now speak of mixed systems, there are various cells also in
the net system now. A mixture of three materials is symbolized here. Every
cell of a system is designated with a lower case letter. There are cells
a
and different cells b and c in our example:
a -
b - a - c - a - b
|
| | | | |
b -
a - c - a - b - c
|
| | | | |
c -
b - a - c - a - b
|
| | | | |
b -
a - c - a - b - c
Every cell
of a net system (it are the cells a,b and c with us)
has a changeable property which has a specific markedness at a specific
time. Every cell kind so to speak represents another changeable property
kind. Therefore, it could be that two different cells represent only two
different property kinds of the same atom. In the case of our simple mixed
systems, a specific cell corresponds to a single particle, a molecule or
another object separated spatially, which has changeable qualities.
The properties of a cell are dependent on the properties of the other cells,
therefore, an effect chain can exist between neighboring cells, e.g.:
a --> b --> c --> a
The neighboring
cell (right
a) effects back in the same way as the initial cell
(left a) (feedback). The expression a --> b should mean
that the markedness of the property of the cell b is somehow causally
dependent on the markedness of the property of the cell
a. Further,
an arrow --> contains no information about
the kind of dependence. Agreement: A closed circuit i.e.
a --> b --> c --> a (the same is
at the end as at the beginning) should always be a link-up with PRESERVATION
STRIVING between initial and final element (if it is not defined differently).
Further, the final element can effect back on the initial element in the
same way (what however is left out in symbolism). A such kind of the positive
feedback e.g. would be the following: The final-a is forced to occupy
the same markedness of its property as the initial-a with a certain
probability, so that the final-a effects back in the same way (spin
grid, see above). Another possibility of the feedback: Initial-a
and final-a build up themselves into the same direction to a stabilization
point (Even less properties of a result finally from small properties by
building up and even bigger properties of a result finally from big properties).
There are perhaps also other possibilities for such feedbacks.
By means of the following feedback chain, the symbols ED, EI, COND, PROP should once again be explained:
a --> b
--> c --> d --> e --> f --> a, EI --> b, ED --> c, COND --> e, d --> PROP
·
EI
--> b meant: Energy EI is necessary, so that b can transmit
the effect on within the chain,.
·
ED
--> c meant: The destroying energy supply ED has its working
point at c. ED and EI can also be identical.
·
COND
--> e meant:
e can only maintain its function within
the chain with the condition COND which must be made.
·
d
--> PROP meant: There is a connection between
d and
the property PROP, which one wants to modify by the Artificial Evolution.
PROP could be e.g. a global external property of the solid state, which
consists of a specific spatial or temporal property distribution of much
small d- particles.
3.6.1 Cooperative preservation striving by inertia
The explanations made just now should now be applied by means of a problem. This example should also provide an insight into way of thinking.
Feedback in the ferro magnet:
s --> s
The following
proves to be bad at our ferro magnet: It is indeed wanted that a spin is
endeavored to extend onto the neighboring cells because that means PRESERVATION
STRIVING. But the preservation striving of the one spin destroys the attempts
of other spins those perhaps adjusted differently itself. Therefore, it
is NON-COOPERATIVE form of preservation striving. Every spin only "thinks"
of itself. Therefore, we want to consider, how we can give the spins a
certain INERTIA so that the existing direction of a spin can not immediately
be destroyed again.
At first one could increase the link-up P between spins by temperature
reduction, what the fluctuations of the spins would reduce. Then the spins
are endeavored to occupy a relatively homogeneous direction i.e. they have
a certain inertia to change their direction now. ALL spins "want" to show
into the same direction. That is now too good something. Perhaps one shall
distribute the spins into small GROUPS, better said onto individual particle.
At first the ferro magnet is reduced to small pieces and then these particles
are compressed again. The spins of an individual particle are relatively
stable then. All its spins are endeavored to show into the same direction.
However, the spins of the other particles have the freedom to adjust itself
into other directions now (because of the smaller link-up because of the
greater distance between the particles). The temperature must not be too
low, otherwise, the spins couldn't be tipped over by energy supply (the
feature of the CHANGEABILITY would be no more available). If required,
the distances between the particles could be enlarged by mixing with a
neutral or magnet-shading material. As a result, the link-up between the
particles could be limited to a measure that an individual neighbor wouldn't
be able to tip the spins of a particle over. Because this particle became
so inert now that SEVERAL neighbors must point into the same direction
to tip over the spins.
But one could mix also a very specific material x which intervenes
active into the feedback process (see above). Material x should
be searched now:
s --> x --> s
Therefore, a particle of material x was inserted into the feedback chain by mixing. s symbolizes a particle of the ferro magnet. In order to make the reaction possible, energy is necessary:
EI -->|
|
s -->
x --> s
Therefore,
x
should support s and vice versa. Therefore, we must clarify the
question, which processes in x could have influence on magnetization
of s. It is the most obvious that x has also again magnetic
properties. How can one make magnetic properties of x by energy
supply EI ? One could to bring the electrons contained in x in motion
by energy EI. The electrons have a magnetic field automatically then. I.e.
x
could be a semiconductor, from which one can knock out detached electrons
and other charge carrier by energy supply in most different form as is
generally known. Because of natural laws the courses of the charge carriers
organize themselves in the following way: The magnetic fields of s
and the magnetic fields of the moved charge carriers in x adjust
itself OF THE SAME KIND (because energy is then least). I.e. magnetic fields
of s and x influence themselves reciprocally in supporting
manner - positive interaction occurs. Therefore, the energy EI is used
to maintain the interaction (i.e. is used for PRESERVATION STRIVING). However,
it can also be used for the CHANGEABILITY. If the energy supply EI is increased
to a specific extent, so many charge carriers result, that these can not
organize by the magnetic field of s. In x then disarranged
eddy currents result. The magnetic fields of eddy currents can restructure
the spins of s. I.e. the greater EI is, the influence of EI is more
destructive. The lower EI is, the influence of EI is the more supporting.
That is precise that what we want (see above: Model of the Artificial Evolution).
The average size of every particle sort, the mean interconnecting number
of both materials, the middle energy supply etc. must be balanced of course
precisely by series of measurements - but to this, later more.
The individual x must be isolated from each other electrically probably
because otherwise the link-up would be too great between all x of
the network. One would then have non-cooperative preservation striving
again. One can achieve this e.g., when one mix non-conductor particles
to this or use former material for s (Fe and non-conductor mixed
finely).
3.6.2 Neuronal Nets
In above symbolism, we always considered only neighboring cells of the net. We now consider a greater section:
s -
x - s - x - s - x
|
| | | | |
x -
s - x - s - x - s
|
| | | | |
s -
x - s - x - s - x
|
| | | | |
x -
s - x - s - x - s
|
| | | | |
s -
x - s - x - s - x
A line -
should symbolize a interaction between DIFFERENT particle kinds. A interaction
kind which is also available was not yet drawn in here: that between the
x
particles. The magnetic fields of the conduction currents of neighboring
x
particle also influence themselves and are endeavored to adjust itself
of the same kind. The state of a x is transmitted by induction to
the neighbor x, and that is dependent on direction of the s
which are in between.
Our net would represent a classical neural net, where the x play
the role of the neural cells and the s play the role of the connections
between cells. Interesting common characteristics at least exist with a
" traditional" neural net. The following comparison shows that:
| General neural net | Neural net with s and x |
| The state of a neural cell is relatively changeable i.e. it fluctuates between different stimulation states. | X plays the role of a neural cell. Its state is relatively changeable i.e. it can fluctuate between different states (current directions and current intensity). |
| The neural cells are connected with each other. | s plays the role of a connection between the neural cells. |
| The kind of a connection is relatively unchangeable. | The direction of s is relatively unchangeable. |
| Depending on kind of the respective connections, the state of the one cell determines the state of the neighboring cells. | Depending on kind (direction) of the respective magnetic particle s between the cells, the electrical state of the one cell is transferred onto the other cell. |
| The connections can change. This change occurs e.g. according to rule of Hebb (learning rule): A frequent simultaneous stimulation of neighboring cells supports its positive connection, otherwise, a negative connection is supported. | There is a regularity which is similar to the rule of Hebb: Homogeneous stimulation (=current direction) of two neighboring cells x support another kind (=direction) of the connection s than opposite stimulation. |
In our example,
s
an
x were ferromagnetic material and semiconductor. Other material
combinations would also be certainly possible in order to make neural network
structures. A candidate for it would be every material with a slightly
changeable property p1 (for the neural cells) mixed with a material with
a relatively unchangeable property p2 (for the connections). p1 and p2
must influence reciprocally themselves.
One could already designate our simple spin grid (see above, computer simulation)
as a neural net and in literature is also made this. But what we have here
(s and x mixed) is a net in the classical sense, which could
also have learning ability in the usual sense because of the inertia of
the connections s. The already learned things would be relatively
stable and new things, which should be learned, could based on it without
the fact that the old things cracks immediately. That is favorable conditions
for an evolution. I.e. one could pursue considerably more complicated evolution
destinations with such structures than only to modify spin distribution.
One could perhaps achieve complicated spatial-temporal spin distributions
which can cause a specific technical destination or operate data processing
(for this later more).
It could be, that it don't function just as it was described with ingredients
s
and x. But perhaps it functions in similar way. However, it was
the destination to clarify for the reader at least which considerations
one could make during synthesis of evolution capable systems.
One could also certainly recognize parallels to natural neural nets here,
e.g. our brain. During unwanted behavior of a child for example, the TEACHER
causes that the available brain structure loosens itself and could partial
be restructured. But during wanted behavior of a child available brain
structure is supported. This "loosening" of brain structure could quite
have global character as in the case of our model of evolution. During
punishment for example, certain emotions are caused which specific global
chemical modifications cause in the brain. The same is valid during reward.
3.6.3 Mixing
Mixing is not like Mixing. Suppose, one has found out the materials a,b,c... by consideration which could be applied for an evolution capable system. One could certainly make theoretical considerations for kind and manner of the mixing of these materials. But one should also set up series of measurements by means of practical experiments with which the individual parameters of the mixing are varied. One could optimize the evolution ability of the mixing even if one has no or little success at the beginning in this way.
Which parameters of the mixing one could vary?
· The
mean interconnecting number
· The
extensions of the particle
· Addition
of neutral particles for the purpose of enlargement of the distance
· Another
particle kind is perhaps missing? (There are slightly changeable particle,
particle for maintaining the interaction, to the absorption of the destroying
energy supply, to the realization of conditions, for the realization of
inertia...)
· Mixing
in the transferred sense: Uniting of the individual particles onto a molecule
· Varieties
of mixing: In boundary layers where two materials clash a mixing can also
occur. Many shifts could placed on top of each other possibly.
· Arranged
mixing: All particles of a specific sort are completely equal, e.g. flatly
ashlar shaped. The particles to be mixed are arranged onto clearly defined
places in regular manner and same distance on an array. (One could use
the same procedures as with chip production.) One could now integrate possibilities
on this chip in order to determine and influence the states of the individual
particles. One would now have the possibility to observe the evolution
process in detail. Further, one could produce copies of structure at the
ready chips which already ended the evolution. One could transfer these
copies onto other chips in a simple manner without the fact that these
would have to go through an evolution process itself.
· Hierarchical
construction of the particle: Hierarchies exist in nature, e.g. Living
beings --> organ --> tissue --> cell --> cell component --> molecule -->
atom. One should perhaps take an example to it and do experiments to build
a mixed system exactly in the same way: The overall system consists of
1000 particles. Each of this particle in turn consists of 1000 less particles.
Each of this less particle in turn consists of 1000 even less particles
etc. Every particle represents a relatively closed evolution capable system
with its sub-particles on its hierarchy level. Jumps now perhaps occur
also in the evolution? The TEACHER must then be capable of more differentiated
procedure (later more). That should only be an idea which was copied from
nature.
· There
are certainly more other possibilities. Please, think itself!
3.6.4 Form
One should
also make for itself thought about the form and the measurements of the
evolution capable system. In order to achieve a regular distribution of
energy supply that is controlled by the TEACHER, a flat form, the form
of a ribbon or a lamellar form would be proper in some cases. If the energy
then strikes the big surface, it crosses the entire thin layer and is not
absorbed as in the case of a thick layer by the top layers of the system.
One could also manufacture thin layers by deposition by evaporation. A
flat, thin-layered form would also have the advantage that the energy in
the form of low heat can be carried off simply again. The thinking part
of our brain is incidentally also flat - the brain periderm. Why ?
In the experimentation phase it is perhaps advisably to determine the size
of the system small (microscopic). One could perhaps achieve so faster
solutions.
3.6.5 Growth
In nature,
it is also noticeable that the organisms grow during their individual evolution.
A tiny cell becomes a big organism. That is certainly useful for its evolution
capability. One should perhaps attempt to imitate that: Begin with a small
size of the system. If the evolution destination is achieved during this
small size, increase the system a little (by enlargement of the surface
or deposition by evaporation etc.). The evolution process then begins again.
If the destination is achieved, enlarge the system again etc.
3.6.6 Indirect
Suppose, one constructed a specific evolution capable system, perhaps one is possibly automatically able to produce diverse other evolution capable systems in this way. Our mixed system with s and x should serve again as a example. This system consists as is generally known of a net of many feedback circuits of the following kind:
s --> x --> s
In order to build as many as desired other systems from this (theoretically), one must only connect one or several effect chain to this circuit:
|--> x1 --> x2 --> x3 --> x4 --> PROP
|
s -->
x --> s
Depending on
which x1...xn one connects at the end of the chain, it could
come out most different kinds of properties PROP. I.e. an single evolution
capable system could be used for most different applications. The connected
chain must not run linear by the way. Chain links could exist which only
perform their role by additional external energy supply of a specific kind
or only during specific external conditions which one must then ensure.
Within the chain, further different feedback circuits could occur. Further,
the chain not necessarily must apply to the x but also apply to
the s. A chain could branch off both from the x and from
the s in each case. From every link of chain could in turn branch
off chains etc.
3.6.7 Combined moves
That is somehow
reminiscent of a combined moves between some chain links a,b,c...
and ED, EI, COND and PROP where a component in this network can also occur
several times and can occur in different states of matter (e.g. COND as
COND1,COND2,COND3 ...).
Actual one would register all possible chain links and components in a
database including their possible inputs and outputs. Depending on setting
of a task, a computer or too a human being then builds the chain combinations
together. There are two fundamental settings of a task:
· Suppose,
one would like to influence a specific property PROP. Which effect chains
with which components a,b,c ..., which conditions
COND and energy EI and ED one could put together so that the property PROP
is at the end of the chain?
· Turned
around question: Suppose, one has a specific number of available chain
elements a,b,c..., COND, EI and ED. Which different
properties PROP one could make by its combination?
Suppose, one
found a chain configuration of that one means that it leads to a specific
PROP. That must not necessarily function in practice, of course. One would
be allowed to designate this chain as a candidate only which must be checked
by experiments. One could attempt to project more precise models of dependences
b = f(a) between the chain links, because a --> b represents
actually only a qualitative connection. This represents a challenge certainly
but one should consider whether one is more rapid with qualitative considerations
combined with experimentation because systems with diverse reactions can
be computed heavily.
3.7 Artificial Evolution by trying out
During the
synthesis of evolution capable systems, one could come to the destination
by an exact modeling attempt of the chain links including exact calculations
etc. perhaps. That would be the one method. The method of trying out including
simple qualitative considerations is the other method.
For the realization
of trying method, one could construct a machine which the mixing and the
inspection of the evolution ability automatizes. Suppose, one would like
to influence the markedness of a specific property PROP. One makes all
those materials available to the mixed machine in different receptacles
which could be chain links in the widest sense. The mixing machine now
makes the most different combinations of the materials - in different concentrations,
during different conditions, in the case of different kinds and sizes of
the energy supply EI or ED and tests in the case of every combination the
evolution ability. The trying out occurs as follows: The closer the fluctuating
PROP to the wanted markedness of PROP is, the TEACHER which is integrated
into the machine determines the destroying energy the less. If during a
specific time the property PROP is statistically more often near his wanted
markedness an evolution ability would then be available. If a combination
shows least signs of an evolution ability, the machine further researches
stand-alone by small modifications of this combination and attempts to
optimize this.
Suppose, one has found out evolution capable combinations by trying out,
one could investigate the cause now and could start other trying experiments
on this basis - etc.
It is actual a double evolution here. The one time the TEACHER carries
out an evolution attempt at every combination. On the other hand, the mixed
machine is controlled by a genetic algorithm and selects the most favorable
combinations out.
There are certainly materials which would be suitable especially for evolution
processes. This could be materials which already have a tendency for self-organization,
e.g. specific organic materials or materials which are planned for Biochips.
Further, a general evolution tester would be useful while experimentation. This device synthesizes no evolution capable systems unlike the mixed machine but it checks available technical systems or systems set up manually for their evolution ability. Such a tester is basically a measuring instrument and perhaps also looks as. Such a device (with its integrated control logic: the TEACHER) would certainly specialized in specific destroying energy form ED (e.g. electro-magnetic alternating field) and in a specific property kind PROP (e.g. electrical field strength) to be measured. Modules for different kinds of ED and PROP would also be conceivable, which one could connect to the TEACHER as required. One could define a measurement variable for the evolution ability at a specific property PROP.
Presumably,
many available materials and technical systems are evolution capable with
regard to specific property kinds PROP from the beginning. Our ferro magnet
(spin grid) e.g. iron was also evolution capable (at least theoretical)
without the fact that we had to produce it first. We would not even have
to think about theory. We could take some material simply e.g. iron, we
could apply a TEACHER to it and we would find evolution ability without
having the least knowledge of which kind the interactions are in its interior.
I.e. independent of it whether we know it or whether we want it or whether
it is right with us, some interaction chain a --> b --> c --> ... -->
a which favor an evolution ability could exist in customary systems.
The microcosm is very manifold. At an evolution process, interactions could
have a role to play (without the fact that the experimenter knows this)
which science already knows. It could however be also constellations of
known interactions which science does not yet know. Or it could be completely
unknown interactions which have a role to play, independent of it, whether
it suits us into the theoretical concept. It at least exists a reasonable
presumption that evolution ability is often available and one should always
investigate this in manifold manner by trying out.
Therefore, we should take available systems (e.g. a solar cell) and should
attempt to improve its function by Artificial Evolution. For this purpose,
we must only determine actual an wanted destination for a property PROP
(e.g., increase the voltage at the solar cell) and we need a TEACHER (e.g.
electric switching) with its measuring instruments and its destroying energy
supply ED. If that does not help, we could vary more or less precisely
the material composition of the system with the aid of above trying-mixing-method.
3.8 The control of the Artificial Evolution: The teacher, the energy supply and the measuring instruments
Said as already,
the TEACHER observes with his instruments the fluctuating property PROP.
The closer PROP of wanted markedness of PROP is, the TEACHER determines
the destroying energy supply ED the less (and the greater the distance
of PROP to the desired value is, ED is adjusted the larger). If after a
specific time, the property PROP is statistically more often near its wanted
markedness, an evolution ability would then be available.
Therefore, there is a middle energy supply EDMeans. The TEACHER modifies
the energy ED inside an area EDMeans - EDDelta <ED <EDMeans + EDDelta.
One can see easily, that the artificial evolution process only functions
at specific values for EDMeans and EDDelta because in the case of too high
ED, every evolution is destroyed immediately again and nothing modifies
itself in the case of too low ED. Therefore, it would be expedient to equip
the TEACHER with a more intelligent control logic at least in the experimentation
phase. Then the TEACHER would be able to modify stand-alone EDMeans and
EDDelta in order to find out optimal values.
Further it could be that the optimal values for EDMeans and EDDelta change
itself during the evolution process. Perhaps it could be necessary at the
beginning, that the TEACHER must react quite generously (big EDMeans) in
order to achieve changes at all. At the end a less EDMeans would then be
sufficient. A continuous advance of the evolution would be conceivable.
However, there are perhaps jumps in some cases in the evolution so that
the TEACHER must change his control logic suddenly. In addition, it would
be possible that the evolution destination (a specific value of PROP) is
achieved after a certain time. The teacher could then determine a new,
a little higher destination and the evolution process would then begin
from the beginning on higher level. Therefore, it could be that one must
form the control logic of the TEACHER flexible. In the simplest case, this
control logic would implement a simple mathematical function:
ED per time = f(PROPDesired - PROP)
e.g. ED per
time = k1/ (PROPDesired - PROP + k2) + k3
or
ED per time = k1 * exp (PROPDesired - PROP) + k2.
One could certainly
find out optimal functions theoretically. But if required it could be also
a complicated algorithm. Secondary remark: One can also discover common
characteristics with the human evolution process. At the beginning every
small progress must be rewarded generously. If a evolution phase is then
concluded, new tasks can be set up on higher level. Its solution is rewarded
generously at the beginning etc.
A further important parameter of the control of the artificial evolution
process is the reaction time. The reaction time consists of three phases:
1. The measurement
of the current value of fluctuating property PROP
2. The reaction
of the TEACHER then (the calculation of the necessary energy supply ED)
3. The realization
of the energy supply ED
The reaction
time must not be too large, because ED should either support or hinder
the CURRENT inner system causes of a specific CURRENT value of PROP. If
the reaction time would be too long, it could happen that a big destroying
ED is provided in a such moment in which a support (small ED) would have
been actual necessary because PROP currently approaches the wanted value.
Presumably, the reaction time must be less than the middle half periodic
time of fluctuation of PROP (= middle time interval of two consecutive
crossover points of the fluctuation). One should adjust an as short as
possible reaction time in case of doubt. One could find out the maximum
reaction time by experiments. There are systems which fluctuate very slowly
so that the necessary reaction time could be great (some seconds). The
human being himself would be able to undertake the role of the TEACHER
when he controls the evolution process "by the hand". One could control
faster systems which fluctuate million times per second with an electric
switching as TEACHER. On the other hand, there are systems which fluctuate
so fast that a TEACHER who is outside of the system would have to react
with a speed faster than light. One would be able to solve this problem
when one attempts to integrate the TEACHER into the system. For this later
more.
PROP could also consist of several properties p1,p2,p3 ... It must not
be different property kinds, but the p could be from same kind which one
were measured at different places. One would designate that as a property
distribution. One could understand the put PROP as a single value which
can be computed from the individual p: PROP = f(p1,p2,p3...), or one could
understand PROP as multidimensional quantity. In the case of multidimensional
properties, one would have to define the distance to the wanted PROPDesired,
e.g. as a Euclidean distance sqrt (sqr(p1-p1Desired) + sqr(p2-p2Desired)
+ sqr(p3-p3Desired) +...), in order to compute the energy supply from this
necessary in the respective moment. Therefore, the property could be theoretically
very complex, it could e.g. also reflect complicated models while information
processing.
It could happen, that one indeed achieves the evolution destination that
however an unwanted or noxious property pu increased simultaneously and
unintentionally during the evolution process. One could operate cause investigation
and could carry out modification of the system now. One could however also
start a further evolution attempt. However, the property pu is added this
time to PROP - however in negative manner, e.g. as a reciprocal. If for
example a large pu is unwanted, the new PROP could defined e.g. as f(p1,p2,p3,1/pu
... ).
Because energy is provided to the system permanently, it must also be guaranteed
that this energy is also carried off again in the form of low-order energy
(heat), otherwise, the system would heat itself up and would destroy evolution
progress. The natural energy radiation which one can support by a flat
external form (see above) would mostly be certainly sufficient. A cooling
will perhaps also be required in some cases.
If the evolution destination is then achieved, it could be that these new
markedness of PROP further exists also after the TEACHER stopped his work.
However, it will often be like this that the evolution process must be
maintained permanently as long as one would like that PROP retains its
wanted markedness. The evolution process should perhaps occur in some cases
during a cooling process, in order to achieve that the result is stable
at the end without teacher.
If the evolution process ran successfully in the experimentation phase,
one should make counter experiments. The experiment is now repeated with
the same experimental setup - however with inverse control logic of the
TEACHER: The CLOSER the distance of PROP from the wanted markedness of
PROP is, the TEACHER adjusts the destroying energy supply ED the more LARGELY
(and the wider PROP is away from desired value, ED is adjusted the LESS).
If this counter experiment is also successful, it is not a artificial evolution
process because one hindered the wanted markedness of PROP. It has other
causes that it functioned nevertheless. Perhaps a pure energy supply is
then sufficing without control through a TEACHER.
EVEN OR DIFFERENTIATED ENERGY SUPPLY? ONE OR SEVERAL TEACHERS?
The destroying
energy supply ED has the function either to support or to hinder the causes
for current markedness of PROP as is generally known. One must consequently
provide the energy ED in a such way that it penetrates all volume areas
in which the cause is for the markedness of the property PROP. However,
a specific cell is combined with all its neighboring cells in the case
of net systems. Every cell is consequently combined indirectly with every
other cell of the system by a chain of neighboring cells. I.e. the causes
for markedness of PROP are in the entire system, not only at a specific
place. I.e. it is not wrong if the energy supply ED penetrates the entire
system evenly. It is the question about what one should think whether this
is always optimal. Because link-up is not equal link-up. Much remote cells
can be linked more strongly or more weakly with each other cell depending
on the regularities which are in the system.
We once again look at our model of spin. As many as possible spins should
be set upwards (=1) on the left side and as many as possible spins should
be set down (=0) on the right side. The higher the temperature of a ferro
magnet is, the link-up of the spins is the less so that one could attempt
to apply two TEACHERS onto this system: one for the left and one for the
right side. Therefore, each of these TEACHERS would be responsible for
a specific spatial area, would have its own measuring instruments and would
be able to control stand-alone the energy supply ED in its area of responsibility.
The two TEACHERS would also pursue different destinations: The left one
want to arrange its Spins upwards and the right one down.
Depending on setting of a task, degree of the link-up of the cells and
the spatial distribution of the causes of PROP, it would even be possible
to apply very many TEACHERS for only system. All TEACHERS could either
be independent from each other or could be linked together or could be
supervised by a superordinate TEACHER. The TEACHERS could have different
or common destinations, could use different or identical instruments, etc.
INNER TEACHERS
To hold the reaction time small, also for practical reasons, one could attempt to integrate the TEACHER into the system at specific applications. In other words: In mixed systems, one could add particles to the system which then undertake the function of the teacher. Therefore, these TEACHER-particles have the task to reduce the destroying energy supply ED during approximation to wanted markedness of PROP. During use of an external TEACHER, the feedback chain looked as follows for example: a --> b --> c --> d --> e --> f --> a, f --> PROP, ED --> c . The intensity of the energy supply ED was adjusted by an external technical construction depending on markedness of PROP. We now leave out the external TEACHER and we mix a particle kind y to the system as a INNER TEACHER. The circuit would now look like this:
a --> b --> c --> d --> e --> f --> a, f --> PROP, f --> y, ED --> y --> c
· In
former times ED influenced directly on c. Now is the particle
y
in between.
y must be able to convert the destroying energy ED into
another energy which has no influence on c.
· The
closer f is to its desired markedness,
y carries out this
transformation into the "harmless" energy form the better. (f determines
the markedness of PROP).
The EXTERNAL
TEACHER controlled the energy supply ED in former times. ED is now provided
to the system in permanent way and the INNER TEACHER undertakes this task
when he converts ED into "harmless" energy forms depending on current state
of f. It is certainly no easy task to find such a y. Y
could also consist of several particles. If need, one could construct a
molecule which meets these demands. By the way one could also provide "harmless"
energy to the system from the outside. Now the task of the INNER TEACHER
would be to convert this energy in a destroying energy form depending on
current state of f.
Each individual TEACHER-particle would strictly speaking represent a TEACHER.
Therefore, there are many INNER TEACHERS acting stand-alone in the system.
However, several TEACHERS only lead to the destination (see previous section)
in the case of specific conditions. Further, the most different variations
of INTERIORS TEACHERS would be possible again. E.g. an individual TEACHER-particle
could be responsible for several cells. Or perhaps all TEACHER-particles
could have interaction with each other and synchronize their behavior so
that there is in fact one only INNER TEACHER. In order to make an wanted
imbalance within a system, one could attempt to use different kinds of
TEACHER-particles in different areas of the system.
4 Applications of the Artificial Evolution
4.1 A hypothetical application: The increase of the efficacy of a solar cell
How could one increase the efficacy of solar cells with the aid of method of the Artificial Evolution? The way of thinking and the procedure should only be illustrated here by means of this problem. It is NOT said that this functions in such a way. We put ourselves merely short-time into the situation of a constructor of solar cells and look at us which considerations he would have to make.
The property which should be modified must fluctuate. This condition is available at a solar cell in actual fact. The voltage of a solar cell noises. One could now attempt to apply the different methods which we derived.
1. Method: TRY OUT. No structural modifications are carried out on the solar cell. Different control logics and - parameters of the TEACHER and different forms of energy supply are tried out only. The teacher (e.g. in the form of an electrical circuit) measures permanently the moment value of the voltage of the solar cell and adjust accordingly the current energy supply to the solar cell (the higher the voltage, the less energy supply). One should experiment with different forms of this energy supply: Electro-magnetic alternating fields or irradiation which penetrate into the entire surface or short alternation current impulses those are put on in short intervals on the terminals of the voltage. It is not discussed here why the method of trying out could possibly have success. This topic was handled further up.
2. Method: TRY OUT. Different control programs of the TEACHER and energy supply kinds are also tried out again precisely as in the case of method 1, however in addition one experiments with material modifications of the solar cell. A kind of mixing machine could be used (see above).
3. Method:
CONSIDERATION possible in combination with methods 1 and 2. Suppose, one
already knows that property PROP (= charge separation) depends on the condition
COND and the states of the cell a: COND --> PROP, a --> PROP.
One must now consider which other conditions and states of cells which
one must implement and/or mix influence COND and/or the states of the cell
a in some kind. Then one gets effect chain, e.g. COND1
--> COND2 --> COND --> PROP, d --> f --> g --> a --> PROP
and
other. A cell (e.g. the cell d) can here get different states. Depending
on the state of cell d the property PROP is influenced in more desired
or in more unwanted manner. Therefore, one must somehow attempt to hold
a favorable state of d unchangeable (i.e. PRESERVATION STRIVING).
Therefore, one must now think whether one can position interaction chains
at the cell d or at other places of these effect chain for the realization
of PRESERVATION STRIVING or whether one can form these chain to feedback
chains, e.g. one could connect the feedback chain f --> i --> k -->
m --> f with cell f in above chain d --> f --> g --> a
--> PROP. One have to add the corresponding cells i,k,m. Further one
would have to think about energy EI for maintaining the feedback in this
interaction chain and about the supply of the destructive energy ED. Possibly
one must add cells to the system which can absorb specific energy forms.
In addition, it would be beautiful if the energy EI be able to be converted
from solar energy. One should still make in the end for itself thoughts
over it whether one could install an additional effect chain which serves
as a INNER TEACHER. If this can convert its destroying energy ED from solar
energy, our construction would be perfect. Perhaps one can also achieve
that the Artificial Evolution must be carried out only during the production
process, otherwise, it would have to run permanently.
4.2 General considerations about possible application destinations
One could distinguish
the possible applications with regard to the degree of complexity of the
evolution destination.
· Applications
are conceivable which have only the destination to shift balances or to
change the distributions of properties or similar simple things, similar
as at our ferro magnet. Although the degree of complexity is small, this
could use for example during energy transformations (e.g. change of charge
distribution). Before one risks higher degree of complexity, one should
explore these simple destinations.
· Applications
with something higher degree of complexity, e.g.: The system should carry
out a specific reflex in the form of a specific property modification each
time when a specific external physical condition occurs. Or one could have
the destination to generate cyclical or time changeable properties, etc.
· High
degree of complexity, e.g. information processing
Applications are conceivable which could be inserted in actual fact in practice but too such ones which merely serve the basic research. In nuclear research, one is happy about every individual atom nowadays even if it only exists a millionth second because it proves something just now. It could be now exactly the same way that a basis researcher is happy about the shift of a balance by a millionth percent into a direction realized not up to now, which he achieved by means of Artificial Evolution. He would not be able to apply that now, however, it verify that the basic possibility exists and that he should also research with traditional methods further in order to enlarge the effect.
The area of
application can be very diverse. One could attempt to change physical,
technical, chemical properties PROP according to its wishes. Even in the
medicine, applications are conceivable because the human organism represents
an ideal net system.
How own enquiries showed, the specialists of all departments have responsibility
problems. They normally say that they might not be responsible for this
theory, one should leave this to the chaos researchers. This attitude could
be a cause for it why the application of the Artificial Evolution is not
yet disseminated. These specialists of all departments shall but recognize
that the Artificial Evolution is only a new method and/or a new tool which
would be able to help to achieve their specific professional destinations
better.
4.3 Unexpected new applications by Artificial Evolution?
We take a purely fictitious example again, which also represents other things to be realized impossibly (?).One will never be able to realize it certainly, but we must now take some example. If you want, you could imagine an other thing for it. The complete or partial overcoming of earth attraction has a great practical importance as is generally known. However, the flight ability of a rocket or of an airplane requires a high technical expenditure unfortunately. There are other possibilities, except for aerodynamics and rocket propulsion, perhaps to overcome earth attraction? Therefore, the gravitational effect of a subject should be cancelled partially or completely or even be reversed. Now we want to invent the anti gravity so to speak. That is certainly Science - Fiction, but we need some example. The carrying out of a verification experiment would not be very erroneous. There are hypotheses after what the gravity of two solids does not only depend on the mass. It perhaps depends too on some conditions within a solid? Perhaps one could change the condition structure in such a way with the aid of the Artificial Evolution that gravitational strength changes?
The experiment
could be carried out as follows:
A solid state
might be given. Its weight, i.e. its attraction to earth, one should reduce.
Therefore, we need a quickly reacting and highly sensible instrument for
the measurement of the weight. Perhaps one should apply the same instrument
in the experimentation phase which is also used for the verification of
gravitational waves. Piezoelectric crystals could make provable possible
weight modifications electrically fast. The TEACHER would have to realize
that in such moments in which the weight of the solid state increases the
energy supply to the solid state is enlarged. Otherwise, energy supply
is reduced. A condition is, of course, that one can verify the fluctuation
of the weight with the aid of the instruments. Therefore, trying method
is applied here. Therefore, one should carry out the experiment at the
most different conditions and with the most different kinds of solid states.
Even if the
effect would be so small that a practical application would not be possible
in such a way, one would know now that one should research on this field
and find out the causes. That it was on topic anti gravity.
It was already
established why one would be able to have success with trying method possibly
further up (see "Artificial Evolution by trying out"). Some feedback chains
a
--> b --> c --> ... --> a could indeed exist in variety of the microcosm
which science do not yet know qualitatively or in a specific combination
and those could play a role during an Artificial Evolution. These chains
could also contain quantum mechanical link-ups. The history of science
shows that there were again and again new discoveries although one believed
to have found everything on a specific field already. It is to be supposed,
that our knowledge about micro physical interactions is at the beginning
only. These unknown interactions could nevertheless have a role to play
at artificial evolution processes independent of it whether we know it
or intended it. From variety of the microcosm undreamt-of new possible
applications could result in this way.
Perhaps one could also interpret the fluctuations and uncertainties in
the field of quantum physics as trying out of the nature which one can
control into wanted paths by a evolution process at least in part?
One verified,
that if only one of the three elementary fundamental physical constants
- namely speed of light, gravitational constant and Planck's quantum of
energy - another value would have, would collapse our universe. It is certainly
no coincidence that these constants have the favorable values precisely.
The cause for it is certainly that our universe (including its laws and
fundamental physical constants) went through an evolution and it resulted
with time a stable balance of the fundamental physical constants. Science
does not yet understand these things. From where one should know that this
balance is not connected with some condition structures within the microcosm?
Then this balance would quite be a suitable candidate for a (local!) artificial
evolution process. One would indeed have the possibility theoretically
by the Artificial Evolution to displace balances without the fact that
one must take care of the inner processes and without the fact that one
must understand these processes. One could attempt to achieve specific
destinations alone by controlled external energy supply. The fundamental
physical constants should be modified here not necessarily locally. It
should only be suggested here which great things could become possible
by the method of Artificial Evolution.
4.4 The importance of verification experiments
Now the theory
of the Artificial Evolution is based mainly on theoretical considerations,
analogy considerations for the natural evolution and computer simulations.
The general interest is small therefore after experience of the author.
Only a verification experiment would be able to cause that the necessary
attention is brought to the theory of the Artificial Evolution. Every reader
might be requested to carry out these verification experiments if he has
the means for this purpose.
For such a experiment, one should perhaps select a solid state with a property
PROP fluctuating strongly and slowly. A radioactive material which emits
a corpuscle at irregular time intervals would possibly be such a candidate.
The destination should be to increase the emission strength. I.e. in such
moments in which a corpuscle is emitted the TEACHER reduce the energy supply
for a short space of time. In moments in which no corpuscle is emitted
the energy supply is increased.
Perhaps one should at first attempt to control known self-organization
processes into wanted paths by controlled energy supply by a TEACHER.
4.5 Further hypothetical applications
Each specialist
must check for its own field which possible applications he could create.
He must not be an expert for artificial evolution. He must only have understood
this theory in the fundamentals. The artificial evolution is only a simple
method or an instrument,which a specialist on a special field could apply
in precisely his special field. There is probably nowadays nobody who is
responsible for this theory. Therefore, this fact is stressed here because
specialists of all departments had argued during own enquiries again and
again that they are not be responsible for this theory. That is of course
wrong: All of them are equally responsible because the possible applications
are very manifold.
In the case of the following examples, it is attempted to concentrate on
the essential. I.e. every time is not described at length that an artificial
evolution process is controlled by a TEACHER which observes a fluctuating
property PROP of the system and which one, if PROP fluctuates into an wanted
direction, the energy supply decreases and otherwise increases. Further,
one does not mention each time that one could modify the effect chain in
the system by adding of variously materials, possibly by trying method.
These variation possibilities of an evolution process which were already
described at length above are now normally mentioned no more. The reader
can apply this himself easily to the concrete example.
4.5.1 Transformation of energy
One should
then examine at first all effects with which energy conversions have a
role to play whether an influencing of efficacy is possible by artificial
evolution. These could be energy conversions as they occur for example
at solar cells, galvanic elements, thermocouples, piezoelectricity, nuclear
transformations, nuclear fusion, emissions and many other processes. Therefore,
the TEACHER would have to measure directly or indirectly the current power
of the system (= PROP) permanently and would react with the corresponding
energy supply.
Therefore, energy would also be "used up" permanently by the TEACHER. The
thing would be of course only practice suitable if the "won" energy is
greater than the "used up" one.
Energy conversions occur actual within all physical systems permanently, not only in such ones which are planned for the practical use. One could measure these energy conversions (= PROP) and influence possibly by an Artificial Evolution. The energy conversion chain between the molecules of a solid state for example could be known or unknown so that it could be possible to discover unknown possibilities of energy conversion from external or inner energy forms.
In practice,
it often causes difficulties to convert a energy form into a specific other
form, e.g. mechanical energy. One should perhaps carry out some experiments
for the direct transformation of a specific energy form into mechanical
energy. On the one hand, one requires a material which has variable magnetic
properties (e.g. ferro magnet, spin glass) and which one would be able
to cause a vibration by change of its magnetic properties for example or
to start a rotor. On the other hand, one needs a material which the original
energy form absorbs well and which would be able to influence the variable
magnetic property of the other material through some effect chain. The
two materials are mixed with each other or only joined together. Nothing
happens in this way. One would have to "teach" the spins in the magnetic
material an wanted arranged behavior by an Artificial Evolution. Therefore,
the TEACHER would observe the movement state of the rotor (= PROP) permanently
and also during least micro movements he would act in accordance with his
task.
Even if the original energy form is low heat energy, the laws of thermodynamics
would not be hurt her, of course, because the teacher is forced to provide
his destroying energy ED permanently.
One can perhaps also develop specific heat pumps. The temperature difference
between the one side of a solid state and the other side would be the property
PROP.
4.5.2 Physical processes
All physical properties of a solid state one could attempt to modify by an artificial evolution process. Therefore, a simple example should at this point be described, independent of it whether one can prove an effect. A material might be given (e.g. radioactive), which emits particles. Our wish should be to increase the emission strength. Therefore, the emission strength (= number of particles / time) is the property PROP. If the observation time interval of the TEACHER is chosen small enough at every measurement, one will find that PROP fluctuates strongly what is useful for the realization of an artificial evolution process.
4.5.3 Medicine
The human organism
is a complex net system with numerous effect chains. Therefore, it could
represent an ideal field for artificial evolution processes. Certainly
one will apply it (if at all) only in the case of specific illness kinds.
A condition for a medical application would be that one would be able to
measure the current activity or the state of an illness or of the healing
process (= PROP) because the TEACHER would have to provide energy depending
on markedness of PROP. The author is no specialist on the field of the
medicine, therefore, fictitious examples should only be stated here. We
take an illness limited locally, e.g. a ulcer as a example. Now one would
have to set up the question for itself which measurable property does distinguish
the sick from the healthy tissue? There are certainly many possibilities.
We suppose - independent from that whether it is right - it might be the
electric resistance. We would then have a simple property PROP to be measured
easily where the TEACHER would have the task to bring the electric resistance
to level of healthy tissue now by means of controlled energy supply. Whether
that functions at all and whether a cure can be achieved as a result in
actual fact, numerous tests must first prove of course. If one can not
employ the electric resistance as a PROP, one should then perhaps attempt
it with the measurement of some other electrical signals which distinguish
from healthy tissue somehow or other (frequency, incidence, strength ,
form...). If need, one could also position a matrix of measurement points
on the tissue and include values into PROP. If that does not function,
one should attempt it with chemical sensors or with some active measurements
with which ones one gets a measured value as a reaction to some external
stimulus. Therefore, the measurement and determination of property PROP
is already a theory by itself which the best expert for artificial evolution
can not set up alone. The specialists of the respective special field must
achieve main work therefore always.
Perhaps one could stimulate growth processes at injuries or fractures in
similar way.
The TEACHER could supply the energy ED e.g. by means of irradiation, sound,
vibration, movement or sense stimulus. The question exists whether one
exposes the surrounding healthy tissue to energy supply likewise (or even
exclusively) and it with it include into the evolution process. One would
"teach" the healthy tissue to cure the sick tissue so to speak.
One should perhaps move the indirect method into consideration here also
(see section "Indirect"). The property PROP is measured at the sick tissue
still. However, the artificial evolution does not occurs in the tissue
now, but in a evolution capable material which is fixed to the tissue and
that can interact with the tissue. One "teaches" the material, so to speak,
to cure the tissue by means of its interaction.
Normal method:
TEACHER --> energy supply -->
tissue --> PROP -->
TEACHER
Indirect method:
TEACHER --> energy supply -->
Evolution capable material
--> tissue -->
PROP --> TEACHER
One would
have to find out the state of material by experiments to tissue samples,
of course.
Devices are already certainly in use nowadays those would fit well into
theory of the artificial evolution without the fact that the customer is
conscious of it. At feedback devices for example, a quality PROP(e.g. skin
resistance or electrical signals) is always measured in each case which
one the TEACHER (the feedback device) moves into a specific stimulus which
then can cause metabolism modifications and specific energy conversions
in the body.
It is interesting that the TEACHER must not be some technical device. In
some cases, one could attempt to carry out the energy supply manually in
accordance with the illness signal - or control even exclusively with own
consciousness when one concentrates on the corresponding part of the body.
4.5.4 Chemical balance
One could attempt to displace the balance of chemical reactions. The property PROP would be the concentration of a desired material. One shall experiment with the indirect method here, too (see previous section "medicine"), when one inserts an evolution capable (solid?) material into reaction. One so to speak "teaches" this material to catalyze reaction.
4.5.5 Information processing
Pattern recognition
should serve as a example here: A matrix of receptors is fixed on the surface
of a suitable evolution capable solid state. This matrix converts the light
stimulation into suitable physical stimulation forms (electric, magnetic
ones or other) those can extend within the solid state in the form of effect
chain in accordance with above theory. One should "teach" the solid state
to react to specific stimulation pattern with specific reaction pattern
on a sensor matrix that is also fixed on the surface of the solid state.
One would determine a measure of similarity between wanted reaction and
actual reaction as property PROP here. If the solid state memorized a specific
pattern including the corresponding reaction, one in the same way continues
with learning of other patterns. Repetition practices are then carried
out later with all models learned already. One should perhaps begin the
learning process with relatively simple patterns and reaction manners and
increase the difficulty slowly.
It would be favorable to form the composition of the solid state in such
a way that a neural network structure results (see section "Neuronal Nets").
Further one should attempt to apply the method of the arranged mixing and
the possibility of copy of structure (see section "Mixing"). Behavior learned
already would be able to be copied then onto other systems without the
fact that these ones must go through new evolution process.
Later one could at least undertake the attempt to "teach" the solid state
more complex and time structured behavior or even artificial intelligence
analogously to pattern recognition.
4.5.6 Communication
Any signal
source might be given which should serve as a sender. It must not be a
signal source of the traditional sense here (light, electro-magnetic waves).
In the simplest case, this signal source could be a solid state. One must
able to change a specific physical property of the solid state. This variable
physical property would then be the signal.
One "teaches" a receiver, e.g. an evolution capable solid state, to react
onto the signals of the sender with specific measurable physical property
modifications. Therefore, the receiver must carry out an evolution process.
Therefore, the property which should serve as a receiving reaction must
fluctuate randomly at the beginning as usually. And also the sender should
send out random signals permanently in the learning phase. The measure
of similarity of the actual reaction of the receiver with the desired reaction
is determined as property PROP. Therefore, the TEACHER must observe the
sender during the evolution process in order to be able to estimate the
correctness of reaction of the receiver and as usual in moments of error
behavior to increase the energy supply ED to the receiver.
May be, that the link-up between receiver and sender - if at all - would
also occur in traditional way. e.g. via electro-magnetic waves, and that
one can solve this problem better with the usual methods. However, there
are hypothetical assumptions that other modes of transfer could exist.
4.5.7 Measuring instruments and detection means
The principle
of this application is completely analog of the transmission of communications
(see above). The measuring instrument has the role of the receiver to play
here while the sample to be examined represents the sender. The instrument
must run through an evolution process before it is applied. As a result,
it gets the ability to react specifically to specific properties or structures.
5 Appendix
5.1 Source code
uses crt;
const GridSize =
14; {Gerade Zahl!}
NumberGridPoints = GridSize * GridSize;
P = 0.7;
PROPDesired = 25;
NaturalEnergy = NumberGridPoints div 100;
var Grid: array [1..GridSize,
1..GridSize] of integer;
NumberPROP,PROPSum:longint;
{================== Help Procedures =================================}
procedure RandomGridPoint(var
x,y:integer);
(*-----------------------------------------*)
{Random Point in
the Grid}
begin
x:= random (GridSize)+1;
y:= random (GridSize)+1;
end;
function GetRandom(P:real):boolean;
(*--------------------------*)
begin
if random < P
then GetRandom := true else GetRandom := false;
end;
procedure Init;
(*-----------*)
{Initialization.
Random spin values are assigned to the spin grid}
var x,y:integer;
begin
randomize;
clrscr;
for x :=1 to GridSize
do
for y := 1
to GridSize do
Grid[x,y] := random(2);
NumberPROP := 0;
PROPSum := 0;
end;
procedure Graphic;
(*----------------*)
var x,y:integer;
begin
gotoxy(45,2); write(NumberPROP:7,(PROPSum/NumberPROP):10:0);
for x :=1 to GridSize
do
for y := 1
to GridSize do begin
gotoxy(x,y);
if Grid[x,y] = 1 then write('X') else write(' ');
end;
end;
{================== Procedures of the simulation ==========================}
function PROPMeasuring:
longint;
(*---------------------------------*)
{The TEACHER measures
PROP = NumberLeft - NumberRight}
var NumberLeft,NumberRight,x,y:longint;
begin
NumberLeft := 0;
{Number left count}
for x :=1 to GridSize
div 2 do
for y := 1
to GridSize do begin
NumberLeft := NumberLeft +Grid[x,y];
end;
NumberRight := 0;
{Number right count}
for x := GridSize
div 2 + 1 to GridSize do
for y := 1
to GridSize do begin
NumberRight := NumberRight +Grid[x,y];
end;
PROPMeasuring :=
NumberLeft - NumberRight;
end;
function TeacherAssessment
(EIG:longint): longint;
(*-------------------------------------------*)
{The TEACHER calculates
the energy, which must supplied on dependent of PROP}
var Energy:longint;
begin
Energy := PROPDesired
- EIG;
TeacherAssessment
:= Energy;
NumberPROP := NumberPROP
+ 1;
PROPSum := PROPSum
+ EIG;
end;
procedure EnergySupply
(Energy:longint);
(*--------------------------------------*)
{Energy is supplied
to the spin grid}
var i:longint; x,y:integer;
begin
if Energy > 0 then
for i := 1 to Energy
do begin
RandomGridPoint(x,y);
if Grid[x,y]
> 0 then Grid[x,y] := 0 else Grid[x,y] := 1;
end;
end;
procedure NaturalEnergySupply;
(*--------------------------------*)
{Some spins change
their direction by environment temperature, too}
var i:longint; x,y:integer;
begin
if NaturalEnergy
> 0 then
for i := 1 to NaturalEnergy
do begin
RandomGridPoint(x,y);
if Grid[x,y]
> 0 then Grid[x,y] := 0 else Grid[x,y] := 1;
end;
end;
procedure Interactions;
(*-----------------------*)
{For the individual
spins is realized the interaction with the neighbors:
With the probability
P, a neighbor spin is put onto the same value as the
initial spin.}
procedure
InteractionsWithNeighbors(x,y:byte);
var dx,dy:integer;
begin
if GetRandom(1/3)
then dx := 1 else if GetRandom(1/2) then dx := -1 else dx := 0;
if GetRandom(1/3)
then dy := 1 else if GetRandom(1/2) then dy := -1 else dy := 0;
if (x+dx >0)
and (x+dx<=GridSize) and
(y+dy >0) and (y+dy<=GridSize) and
not((dx=0) and (dy=0)) then
if GetRandom(P) then Grid[x+dx,y+dy] := Grid[x,y];
end;
var x,y:integer;
i:longint;
begin
for i := 1 to GridSize
* GridSize do begin
RandomGridPoint(x,y);
InteractionsWithNeighbors(x,y);
end;
end;
(* MainProgram *)
(*---------------*)
var Energy,PROP:longint;
begin
Init;
repeat
PROP := PROPMeasuring;
Energy :=
TeacherAssessment (PROP);
EnergySupply
(Energy);
NaturalEnergySupply;
Interactions;
Graphic;
until keypressed;
end.
February 1997
Carsten Zander
E-Mail: Carsten.Zander@t-online.de