Any new direction in the field of digital computer technology is a system of knowledge, which can be viewed in terms of its basic criteria.

1. It should always have a theoretical basis, which differs from the previous knowledge and the opportunity to create a new and effective step forward.

2. It must have a distinctive language in which we can discuss the objects of its investigation, the rules by which the observation can be performed to bring in a coherent system aiming to show their efficiency in comparison with known science or known trends in science.

3. It must have a method to create new objects with new properties and qualitatively assess their results compared with known similar objects in the field of information systems.

4. It is desirable that this new direction in the field of information systems would be interdisciplinary and would cover such typical areas as: automata theory, methods of construction of the elementary memory circuits, methods of construction of typical devices and new approaches and methods in building the software of not only deterministic devices but also probabilistic ones being developed.

The system of existing knowledge of digital computer technology

The system of existing knowledge of computer technology has hierarchical interdisciplinary connections described by V.M. Glushkov [1].

Performance limitations of modern computer systems and networks are due immutable features of elementary memory circuits (flip-flops), which affect the principles and methods of devices design theory of computers and computer systems.

“The transition from the deductive method that satisfies the existing deterministic systems to inductive methods associated with probability systems requires a different way of thinking. The inertia of the old way of thinking has not yet been overcome. The real obstacle to the development of a new direction is not the complexity of the problems, but conservative people”, wrote St. Beer [2].

These interdisciplinary developments of theoretical knowledge in the field of modern computers and neuro-computers have been and are now based on the element basis of integrated circuits. This basis implements functions of combinational circuits based on Boolean algebra, and multi-state triggers and memristors are used as memory, which imposes the fundamental limitations on the developed computer systems and those being developed now [3–4].

The author believes that all the above reviewed work on reconfigurable memory circuit was conducted by reconfigurable excitation and outputs functions, based on RS-trigger. They have fundamental limitations with significant impact on the building of computers and systems, namely [5]:

1. They all operate in the automaton discrete time ti (i = 1, 2. ..., n, ...).

2. The basic memory circuit (RS-trigger) does not allow rebuilding of stores states work.

3. All these devices are described by Mealy and Moore automata that define the consecutive operation of the devices.

4. The transition in the memory circuits occurs using one variable x(t).

5. The used principle of program management, proposed C. Babbage, does not allow simultaneous processing of general and local information.

Presentation of hierarchical information

Modern computer and neuro-computing systems built on the modern basis use the consistent information as input data signals x(t) and use this information in the discrete automaton time [1–4].

In fact, the information is hierarchical being the third element of the universe alongside with matter and motion [6].

The author could hardly disagree with this, since he has been involved in creation of a trend in the field of digital computer technology, which deals with the storage and processing of hierarchical information, for almost forty years [7].

Instead of the input signal x(t), supplied to the memory circuit (trigger) of the computing devices the author used the input word p(T) = x(t), e(Δ), consisting of two consecutive signals x(t) and e(Δ), supplied to the multi-level memory in a single machine cycle T [47–48]. That difference was enough to have the opportunity to simultaneously process the general and local information in a single machine T cycle, which could not be principally implemented in devices with memory on triggers and memristors [8].

New system of knowledge in computer engineering

So what is the new system of knowledge that determines the new scientific direction in the field of computer technology?

Multifunctional and hierarchical abstract machines capable to process hierarchical information (of general and local type) simultaneously in a single machine T cycle and 4-th kind machines, which control operation of the memory elementary circuits, were originally developed by the author [7].

1. There were developed multifunctional machines of the 1st, 2nd and 3rd kind that work in the automaton uninterrupted time Ti = ti + Δi (i = 1, 2, ..., n, ...). Monofunctional Mealy and Moore automata, which operate in a discrete time, are a special case of multifunctional machines of the 1st and 2nd kind.

2. Multifunction machines can be considered as deterministic, having two deterministic transitions: simple and enlarged, which expands the functional properties of deterministic Mealy and Moore automata as probabilistic and fuzzy. Multifunction machines are able to make the transition in a matrix structure of states of two variables: x(t) and e(Δ) in one T cycle of machine time.

3. A mathematical model of hierarchical abstract machine with multi-function organizational system memory was designed.

4. There were developed 4th kind machines controlling operation of memory elementary circuits.

The theory of microsynthesis and analysis of multi-functional and multi-level memory elementary circuits were originally developed by the author [7].

1. The theory of microsynthesis and analysis of two classes of multifunctional memory circuits (MFMC) was developed. The MFMC have a matrix structure of storing information and two input variables: x(t) and e(Δ). The MFMC have flexibility, selectivity, improved reliability, reduced need in logic elements per one state and are open structure and thus positively differentiating from an asynchronous RS-trigger. The techniques incorporated in this theory, allow the developer according to the criteria and the number of states or the number of reconfigurable subsets of states to easily design the structure and functional memory circuit.

2. The theory and analysis microsynthesis two classes of multi-level memory circuits (MLMC). The MLMC have the half-closed structure. The MLMC allow simultaneous storing of general and local information, and have such features as flexibility, selectivity, improved reliability, reduced need in logic elements per state and thus positively differentiating from an asynchronous RS-trigger. The techniques incorporated in this theory, allow the developer based on the criteria and the number of states or the number of reconfigurable subsets of states to easily design the structure and functional memory circuit.

Constructing methods of standard computing devices on MFMC and MLMC were originally developed by the author [7]. These are:

1. Reconfigurable registers on MFMC and MLMC.

2. Reconfigurable counters MLMC.

3. Reconfigurable control devices.

4. Computer reconfigurable structure.

The human brain has several advantages over all the techno-cybernetic devices on such important, in the author’s opinion, properties.

Firstly, the input signals from the external environment, affect eyes, ears, body, taste of food and have multifunctional, matrix structure.

Secondly, the information coming from the environment is summarized. This is illustrated by the eye. There are 18–20 million receptors in the eye. The cones that summarize the visible information through the eyes receptors amount to about 72 thousand. So the data compression in approximately 256 times occurs at the second level. It is important to understand and technically solve the issue of information compression.

Thirdly, one must consider the natural growing links between the neurons of the human brain. The child gradually establishes connections necessary to generalize the received (expandable) information, construction of relevant templates and models, reflecting the real world of the individual.

Fourthly, the human brain has between 14 and 20 billion neurons. This is a fairly large structure in terms of the number of neurons, which is difficult to physically set up at the present stage of technological development, and moreover to manage it.

The talented mathematician Frank Plumpton Ramsey proved that complete disorder is impossible in such large structures as human brain, the universe, etc. Thus, each reasonable large set of numbers, points, or objects necessarily has the ordered structure. Researches in this field confirmed this important result [9]. However, the problem of creating ordered structures in models of the human brain remains.

Fifthly, the brain is not a computer; it does not have logical theories, positional number systems, but only its own logic of getting the information, data compression, the choice of communication path with other cells, to summarize this information. The calculations, reasoning, number systems, and any other algorithms are derived from those models that have generalized and represent human interest, according to the interesting work by A.V. Nikitin “Cell control logic” [65].

The sixth point is about the neuron structure, which has two sets of input signals: excitatory and inhibitory. Triggers and memristors do not have this feature, and they use only setting (excitatory) input signals x(t).

Multilevel memory circuits that have two sets of input signal: setting (excitation) x(t) and stored (selecting) e(Δ) have been firstly proposed as neurons and neural networks. In the scope of neurons, neural connections and architectural ensembles of neural models, the author offers the following results, which follow from the proposed new direction [10–12].

Let’s consider construction of neurons based on MLMC and its characteristics:

1. The brain neurons have different structures that can be created in analog form in MLMC. In addition, these structures correspond to the laws of nature, which is expressed as the golden ratio. This law manifests itself in many structures, such as: human being, shells, fish, etc. For example, MLMC, which stores 18 states and consists of 10 elements, is characterized by 1,8, which is close to the number of Φ = 1,618.

2. The neuron on MLMC has two sets of input signals: setting (excitatory) x(t) and storing (selective) e(Δ).

3. The functioning of such neuron in the automaton uninterrupted time is described.

4. The efficiency of the neuron can be tested by the controlling machine of the 4th kind.

5. In case of failure or non-use in the process, it can be turned off, similar to the biological neuron.

6. Most importantly, such neuron can selectively store information in its matrix structure of states memorizing.

Based on the properties of the three-level memory circuits, the axon register, which can selectively connect the output of the neuron to one or more neurons in deterministic, probabilistic or fuzzy modes, can be built [12]. This allows building models of neural networks, both deterministic and probabilistic and fuzzy.

In the field of computer software and neuro-computers it also allows to use reconfigured microprocessors that are able to change the structure of commands without loss of performance due to the introduction of common code in the address system of commands based on the hierarchical principle of programmed control [12].

In due time, the described achievements have been applied in specific devices under the author’s guidance. Theoretical and practical results of these studies are presented in the author’s doctoral thesis. At the moment the work on the development of this new direction is continued by author’s postgraduates under his guidance.

This new computer technology knowledge led to the following conclusions:

1. All devices on MFMC and MLMC work in the automaton uninterrupted time Ti (i = 1, 2, ..., n, ...).

2. The basic memory circuits MFMC and MLMC to reconfigure the work of the storable states.

3. There is a description of all devices with memory on MFMC and MLMC by Maraсhovsky automata (multifunction automata of the 1st, 2nd and 3rd kind), which define the nature of the devices reconfigured.

4. The transition occurs in the memory circuits in the two variables x (t) and e (Δ).

5. The used principle of hierarchical programming control, proposed by L.F. Marakhovsky allows simultaneous processing of general and local information.

Usually scientific paradigm change is among the most dramatic events in the history of science. When discipline is changing one paradigm to another, it is called “scientific revolution” or “paradigm shift”. The decision to abandon one paradigm is always at the same time a decision to accept another paradigm, and the verdict leading to this decision includes both a comparison of both paradigms with nature and comparing paradigms with each other.

Conclusion

References to knowledge in interdisciplinary areas of knowledge of the new directions in the field of computer systems allow raising the processing level of hierarchical information to a higher level, are described. The favorable efforts to implement these developments, unfortunately, are made only by leading companies such as Intel, IBM etc., or government programs, because the development should start from a scratch.

Companies will be allowed to be ahead of their competitors, and countries – to raise their prestige and economy.

Also, it will make a step forward in the field of computers, and neuro-computers, and will create competitive devices compared to present ones on the existing integrated circuit technology.