Reliable Computing

About Reliable Computing

Reliable Computing — Journal Profile

Reliable Computing is a specialist journal in numerical mathematics and computer science devoted to computations whose results can be trusted in a strict mathematical sense. Its own description, carried on the journal home page hosted at reliable-computing.org, defines the field in one sentence: it is an open electronic journal devoted to mathematical computations with guaranteed accuracy, to the bounding of ranges, to mathematical proofs based on floating point arithmetic, and to other theory and applications of interval arithmetic and directed rounding. That short statement is the most reliable guide to the journal, because it comes from the journal itself rather than from a secondary ranking site, and almost everything else in this profile — its history, its article types, its community and its guidance for authors — follows from it. A computation is reliable in this sense when the published result is accompanied by a proof that the true answer lies inside stated bounds, even after rounding errors, measurement uncertainty and modelling approximations are taken into account.

The journal record held on Journals World lists the print ISSN 1385-3139 and the electronic ISSN 1573-1340, a publication history beginning in 1995, and a publisher given in garbled form as a number. Resolved against official and bibliographic sources, the publisher is Springer, through its Netherlands mathematics programme: OpenAlex resolves both ISSNs to a single source record, names Springer Nature as host organisation, gives the Netherlands as country, and points readers to the Springer volumes-and-issues page for journal code 11155. Secondary summaries of the SCImago record likewise give Springer Netherlands as publisher. Authors should therefore treat Springer as the publisher of the archival journal and the Louisiana-hosted electronic site as the journal community home page, and they should check both places before relying on any third-party description.

Identity and identifiers

The core identifiers are stable across every source consulted for this profile. The print ISSN is 1385-3139 and the electronic ISSN is 1573-1340. OpenAlex lists 1385-3139 as the linking ISSN and returns the same source, S64721560, whichever of the two ISSNs is used to resolve it, which is a strong practical test that the two numbers belong to one journal rather than to two different titles. The standard abbreviation, as given in secondary SCImago summaries, is Reliab. Comput. The journal title itself is deliberately plain: it names an approach to computation rather than a discipline, a university or a country, which is why it is sometimes confused in casual searching with the much broader idea of dependable or fault-tolerant computer systems. Those are different fields. Fault tolerance asks whether a system keeps working when components fail; reliable computing in the sense of this journal asks whether a numerical result is mathematically guaranteed, with explicit enclosures for error and uncertainty, when the computation finishes.

Understanding that distinction matters for anyone using this profile to decide where to submit, where to cite, or how to catalogue the journal. A paper on redundant hardware, failover design or system availability, however strong, does not belong here unless its contribution is a guaranteed numerical method. Conversely, a paper that looks like pure numerical analysis — a tighter enclosure for the solution of a linear system, a verified solver for an ordinary differential equation, a proof that a global optimum lies in a stated box — is exactly the centre of the journal, even when its application area is engineering, physics, chemistry, economics or control.

History and development

The journal began under a different name. Community histories of interval computation record that a new international journal called Interval Computations was launched in 1991, and that from 1995 it was issued under the new title Reliable Computing. The year 1995 therefore appears in database coverage as the start of the present title, not as the birth of the field or of the journal line. The renaming was significant: it signalled that the scope was broader than interval arithmetic alone and included every method that delivers results with a mathematical guarantee — directed rounding, validated numerics, rigorous global optimisation, computer-assisted proof and the careful treatment of uncertainty in data.

The Springer period supplies the archival volumes that libraries and indexing services know best. Database coverage summaries describe coverage running from 1995 to 2008, resuming from 2010 to 2018, with further isolated coverage noted in 2021, 2022 and 2024. The journal community site, hosted with support from the University of Louisiana at Lafayette mathematics community, preserves tables of contents from 2010 onward in an open electronic form: volume 14 in 2010, volume 15 as a special volume devoted to material presented at SCAN 2008, volume 16 in 2011–2012, volume 17 as a special issue on the use of Bernstein polynomials, volume 18 in 2013, volume 19 devoted to SCAN 2012, and continuing volumes through volume 26 in 2018, alongside a supplement containing proceedings of the 2002 SIAM Workshop on Validated Computing. Individual special volumes honour central figures in the field, including a volume in honour of Ray Moore, the Stanford researcher whose 1962 dissertation and 1966 monograph founded modern interval analysis.

That mixture — a commercial archival publisher and an open community electronic home — explains several confusing features a researcher may meet. The same article may be reachable through a Springer page with a digital object identifier and through a free PDF on the journal site. Citation databases may show gaps that reflect changes in production and indexing rather than years in which no research appeared. And the journal may be described in one place as a Springer journal and in another as an open electronic journal; both descriptions capture part of its real history. For tenure, reporting or library purposes, cite the version of record for the individual article and give both ISSNs where a form asks for them.

Scope and subject coverage

The scope follows directly from the journal motto quoted above. Core topics include interval arithmetic in its theoretical and implemented forms; directed rounding and floating point arithmetic designed so that rounding works in a known, provable direction; enclosures and bounds for the ranges of functions; validated and verified numerical methods for linear and nonlinear equations, optimisation, integration and differential equations; and mathematical proofs that use floating point computation as a rigorous step rather than as an illustration. OpenAlex topic mapping for the source places its largest body of work in numerical methods and algorithms, followed by polynomial and algebraic computation, mathematical and computational methods, formal methods in verification, and matrix theory and algorithms — a distribution that matches the journal description closely.

Applications are welcome when the guaranteed method is the contribution. The tables of contents preserved on the journal site show the breadth plainly: verified solutions for eigenproblems and linear systems, robust control and state feedback for interval systems, guaranteed parameter estimation for dynamical systems, interval methods in global optimisation and constraint programming, verified simulation of ordinary and differential-algebraic equations, applications in finance under interval and fuzzy uncertainty, physiological model validation, and computer arithmetic libraries such as C-XSC that implement extended precision and interval data types. A recurring pattern is a hard applied problem — one where a wrong sign, a missed root or an underestimated error bound would matter — solved by a method that certifies its own answer.

Equally important is what falls outside the scope. Heuristic numerical work that reports only that a method appeared to converge on test examples, without enclosures, proofs or rigorous error bounds, is a weak fit. General software engineering, general artificial intelligence and general applied mathematics papers belong here only when guaranteed accuracy is central to the claim. Authors unsure about fit should read two or three recent tables of contents in full: the titles alone teach the house style, in which words such as verified, guaranteed, enclosure, rigorous, validated and bounding recur constantly because they name the standard a submission must meet.

Publisher, place and editorial community

OpenAlex attributes the source to Springer Nature, with the Netherlands as the country of the publishing programme, and the Springer link for journal 11155 is the archival home. At the same time the living journal community is strongly associated with the interval analysis research network centred for many years at the University of Texas at El Paso and the University of Louisiana at Lafayette. Vladik Kreinovich appears across the journal record as a guest editor, preface author and community organiser — for example as guest editor, with Martine Ceberio, of the SCAN 2008 special volume — and R. Baker Kearfott and colleagues at Louisiana host and maintain the journal web presence. A current named editor-in-chief and a full current editorial board list could not be verified from an official page retrieved for this profile, so no board roster is printed here; authors who need board information for a submission decision should take it from the journal site or the Springer journal page at the time they submit, not from an older secondary listing.

The journal is closely tied to its conference community. Special volumes devoted to SCAN meetings — the International Symposium on Scientific Computing, Computer Arithmetic and Validated Numerics — and to SWIM workshops show a deliberate pattern: work is presented, debated and refined inside the community, and the strongest material is then published in refereed journal form. The supplement preserving the 2002 SIAM Workshop on Validated Computing proceedings points in the same direction. For a new author, this means the journal is not an anonymous submission portal; it is the publication arm of a identifiable, long-running research community with shared standards, shared software and a shared memory of what counts as a convincing verified result.

Indexing and bibliometric record

All bibliometric figures in this section are attributed to the source that produced them, and the sources do not agree exactly, which is normal for a small specialist journal. OpenAlex, resolved in the foreground for this profile through its ISSN endpoint to source S64721560, records 532 works, 7,601 citations, an h-index of 39, an i10-index of 196, a first publication year of 1995 and a last publication year of 2018 in its own corpus, with only 14 works counted as open access in that corpus. The stored Journals World record gives an h-index of 36, and secondary summaries of the SCImago record also give an h-index of 36, a SCImago Journal Rank of 0.129, an overall rank of 26,479 and a best quartile of Q4 in each of applied mathematics, computational mathematics and software. The difference between 39 and 36 is a database difference, not an error to be averaged away: OpenAlex and Scopus count different document sets over different windows.

Read sensibly, the figures describe a small, durable, highly specialised journal rather than a high-volume citation magnet. A few hundred papers, cited a few thousand times, with a long tail of heavily used foundational work, is the expected shape for validated numerics, where individual methods and software papers can be used for decades. The two-year mean citedness reported by OpenAlex is 0.0 in its current summary, which should not be read as proof that recent work is uncited in every database; it reflects the age profile of the corpus OpenAlex holds for this source. Researchers evaluating the journal for a promotion or funding file should quote the database their institution actually uses, name it, and give the date of checking, rather than mixing an h-index from one service with a quartile from another.

A Directory of Open Access Journals search for both ISSNs, run through the DOAJ search API for this profile, returned no records in the output received. That result is reported here exactly as a search result; it is not treated as a definitive statement about listing status in any other directory. The journal describes itself as an open electronic journal and makes the post-2010 tables of contents and full texts freely reachable from its own site, which is the practical access fact most readers need: the community archive is open to read even though formal directory status could not be verified beyond the search described.

Fees, access and publication model

OpenAlex records no article processing charge prices for this source and gives no APC amount. No official fee schedule was retrieved from the journal site or the Springer page for this profile, so any specific charge figure would be invention and none is stated here; fees for this journal are therefore described as not published in the sources consulted. Historically the journal model combined a subscription-era Springer archive with free electronic access to later community volumes, so a reader may meet a paywall on one platform and a free PDF of the same work on the journal site. Authors with a funder open-access mandate should confirm, in writing from the journal or publisher at submission time, which route applies to their article, which licence will be applied, and whether any charge arises, and they should keep that confirmation with their grant records.

Guidance for authors

Write for a reader who will check your claim. Every central numerical statement in a Reliable Computing paper should be accompanied by the information needed to judge its guarantee: the arithmetic used, the rounding mode, the interval or enclosure method, the treatment of input uncertainty, and the software and version that produced the result. State theorems precisely, separate what is proved from what is observed experimentally, and report failures and wide enclosures as honestly as successes — in this field an honest wide bound is a contribution, while an unjustified narrow one is a defect. Where code exists, describe how a reader can obtain it; the journal community has a long tradition of shared libraries and reproducible verified computation, and papers that connect to that tradition are easier to evaluate and to build on.

Before submitting, do the ordinary checks with unusual care. Confirm that the manuscript is not a general numerical paper with the word reliable added to the title. Read the journal description again and be able to point to the sentence in your abstract that promises guaranteed accuracy, a bound, a proof or a validated method. Prepare references in the style of recent issues, give complete bibliographic details for interval and validated-numerics sources, and expect review by specialists who know the standard counterexamples in the area. Timelines, acceptance rates and a current submission portal could not be verified from official material retrieved for this profile and are therefore not stated; intending authors should use the contact and submission information on the journal home page itself and keep a dated copy of whatever instructions they follow.

How to verify this profile and cite the journal

To verify any fact here, start with the identifiers. Search a library catalogue or the ISSN record by 1385-3139 for print and 1573-1340 for electronic, resolve the same numbers through the OpenAlex source record, and compare the result with the Springer journal page for code 11155 and the tables of contents on reliable-computing.org. Cite articles with author, title, journal name Reliable Computing, volume, pages and year, adding the digital object identifier where the version of record provides one. If a database shows the predecessor title Interval Computations for work from 1991 to 1994, that is the same journal line under its earlier name, and a complete publication history should mention both titles rather than treating them as unrelated venues.

In short, Reliable Computing is the house journal of guaranteed numerical computation: founded as Interval Computations in 1991, renamed in 1995, published in its archival form by Springer in the Netherlands, sustained by an international interval-analysis community, and still the natural first reference point for work in which a computed answer must come with a proof. Its metrics are modest in the way of a true specialist journal, its archive is unusually open for its era, and its standard — no claim without an enclosure, a bound or a proof — remains the clearest test of whether a manuscript belongs in its pages.