Virtual Institute of Continuity in Computations

Library


CONSTRUCTION AREA - ENTER AT YOUR OWN RISK !!!


Library - temporary closed

This is the most important area of our Institute. Here you will find what we believe is known and reasonably well understood about the continuous models and continuous nature of computations. The relevant long-term research goals are listed. We will also gradually accumulate selected pointers and references to the relevant literature here.


Sections


Continuous nature of computations

When a software engineer or a computer scientist first hears about attempts to view digital programs as continuous objects, the typical first reactions is: "Your approach is absolute and total nonsense. We know that all digital computer systems are totally discrete. These systems can be viewed as transformers of bit sequences at the lowest level of granularity. In principle, we know everything about these systems, because their behavior is discrete and precise".

Yet, upon careful look at the practice and theory of computations, one observes a number of continuity phenomena. One of the most important continuity phenomena comes from the everyday contemporary practice of creation and use of software products. Virtually all software products used today contain a number of bugs - leading to refusal to work, crashes, and incorrect processing of data. Yet for a usable software product these unpleasant effects are observed relatively infrequently - most of the time the product works satisfactory. What we should say here is that the real product is sufficiently close to the ideal product, close enough to be used in lieu of the unavailable ideal product. This is a phenomenon of continuity.

This continuity results from the process of testing and amending the software and is crucial for our ability to use any software at all. The ability to introduce relatively small improvements into software is also a phenomenon of continuity. Moreover, it is quite possible that the ideal product cannot exist at all (e.g. if the specifications are for an undecidable problem), yet we still can consider models which contain representatives for such ideal products and for their real-life approximations.

Another evidence of continuity in computations comes from the research in the effective versions of mathematical analysis, e.g. constructive analysis. It turns out that all constructive functions from constructive real numbers to constructive real numbers must be continuous. It is not difficult to understand why computations at the points of discontinuity cannot be performed correctly. If one has to deal with non-continuous functions, they have to be partially defined - points of discontinuity cannot belong to their domain.

On the more abstract level one should point out the close connections between intuitionistic logic and computability together with the topological nature of the models for intuitionistic logical calculi. See also our Results section for more examples.


Some of the reasons to study continuous models of computations


Some results in continuous models of computations

Construction in progress...

Selected open problems

Construction in progress...

References to literature and resources

Construction in progress...
Back to the lobby of the Institute