Bolzano: Foundations of Analysis and Infinity

• Bolzano is a foundational topic in mathematical analysis defined by Bernard Bolzano’s pioneering work on the arithmetization of the continuum, measurable numbers, and completeness.
• His methodology introduced rigorous constructions of infinite number expressions and precise order relations, laying the groundwork for modern theories of continuity and the real-number system.
• The legacy of Bolzano endures in named theorems like the intermediate value and Bolzano–Weierstrass principles, as well as in contemporary AI-assisted multi-agent research systems.

Searching arXiv for relevant papers on Bolzano to ground the article in published work. Bernard Bolzano (1781–1848) figures in modern scholarship as a major precursor of the arithmetization of analysis and of later formal work on infinity, completeness, and continuity. In the early 1830s, in the unpublished manuscript Grössenlehre (“Theory of Values”), he undertook one of the first genuinely arithmetical constructions of the continuous number-line; in Reine Zahlenlehre he introduced “infinite number expressions” and “measurable numbers”; and in Paradoxes of the Infinite (1847–48) he defended a continuum understood as an infinite class of points with no isolated points. Subsequent mathematics attached his name to the intermediate-value tradition, the Bolzano–Weierstraß principle, and the Bolzano–Poincaré–Miranda theorem, while contemporary computer-science literature has reused the name for an open-source multi-agent system for mathematical research (Sinkevich, 2015, Trlifajová, 9 Aug 2025, Kreuzer, 2010, Grebík et al., 18 Apr 2026).

1. Historical corpus and conceptual program

From his first geometrical writings of 1804–1817, Bolzano conceived a continuum as an infinite collection of points whose “neighbourhood structure” admits no gaps. In the account reconstructed by Trlifajov, this early topological orientation remained stable even as the technical framework changed: originally the line or surface was treated as a set of points with a “neighbour” relation at given distances, and by 1817 Bolzano already required an analytic notion of “point-on-the-line” in order to prove the Intermediate-Value Theorem. This led him to an $\varepsilon$–$\delta$ style definition of continuity and to the convergence criterion later called the Bolzano–Cauchy criterion, but still without a fully articulated real-number system (Trlifajová, 9 Aug 2025).

The arithmetical turn occurs in the early 1830s. In Grössenlehre and Reine Zahlenlehre, Bolzano replaced older geometric intuition with measurable quantities or measurable numbers, to which he assigned precise order-relations, density properties, infinitesimal and infinite magnitudes, and completeness results. These manuscripts remained unpublished until the twentieth century; the historiographical consequence emphasized in the literature is that they anticipated later work by Méray, Dedekind, Cantor, Heine, and Weierstrass without exerting direct contemporary influence (Sinkevich, 2015, Russ et al., 2018).

2. Measurable numbers and the arithmetization of the continuum

Bolzano’s arithmetization begins with “infinite number expressions,” namely expressions involving infinitely many occurrences of $+$, $-$, $\times$, or $\div$. Among these he isolates measurable numbers. In the standard reconstruction, an expression $S$ is measurable if, for each $q\in\mathbb N$, there exists an integer $p$ and positive expressions $P_1,P_2$ such that

$\delta$0

The rational $\delta$1 is the measuring fraction at precision $\delta$2. This is an approximation scheme by rational brackets of width $\delta$3, and Trlifajov presents it as the arithmetic counterpart of Bolzano’s topological “no isolated points” conception of continuum (Trlifajová, 9 Aug 2025).

Bolzano revised the equality criterion after observing that distinct infinite expressions can differ by a quantity so small that all measuring intervals collapse to the same rational mark as $\delta$4 grows. In the revised criterion, two measurable numbers $\delta$5 are equal iff their difference is infinitely small, equivalently iff for every $\delta$6 the difference admits measuring fraction $\delta$7. Modern commentary in the literature treats this as the discovery of equivalence classes of infinite expressions. With that revision in place, Bolzano proves order and algebraic laws: trichotomy, transitivity, Archimedean behavior, density, compatibility with addition, closure under $\delta$8 and $\delta$9, and the field laws in modern formulation. One paper summarizes the resulting structure as a commutative ordered ring, in fact a field (Russ et al., 2018).

A notable feature of the reconstruction is methodological. Although Bolzano admitted infinitely small and infinitely large quantity-expressions, he did not use them to found analysis. Trlifajov argues that Bolzano regarded the Archimedean, measurable subset as the secure basis for continuity, derivative, and integral, because infinitesimals as such lacked the extra logical structure later identified with a Transfer Principle (Trlifajová, 9 Aug 2025).

3. Completeness, cuts, and the number-line

A central late formulation is what Trlifajov calls Bolzano completeness: if $+$0 is a class of points with positive distance-function $+$1, then $+$2 is a continuum in Bolzano’s sense iff for every $+$3 and every real $+$4 there is another point $+$5, $+$6, such that

$+$7

In words, no point of the continuum is isolated at arbitrarily small scale. Bolzano restates this in Paradoxes of the Infinite as the condition that every point has at least one neighbour “at every distance however small” (Trlifajová, 9 Aug 2025).

The same literature attributes to Bolzano three main completeness theorems for measurable numbers. The first is a Bolzano–Cauchy completeness theorem: every Cauchy sequence of measurable numbers has a measurable limit. The second is a Supremum Theorem: every nonempty set of measurable numbers bounded above has a least upper bound. The third is a cut-separation theorem: if variable measurable numbers $+$8 and $+$9 satisfy $-$0 always, with neither $-$1 attaining a maximum nor $-$2 a minimum, then there exists at least one measurable $-$3 with $-$4; if $-$5, exactly one such $-$6 exists; and if $-$7 is bounded below by some $-$8, there are infinitely many such $-$9. Trlifajov treats an important corollary of the third theorem as a proof of Bolzano completeness in the arithmetical setting (Trlifajová, 9 Aug 2025).

In the history-of-analysis literature, this same complex is described as an early “cut-like” theorem. Bolzano bisects intervals, generates nested intervals whose lengths shrink to zero, and uses his convergence criterion to isolate the unique common point in the vanishing-width case. He also asserts a “pantachisch” density property: between any two unequal measurable quantities there exists at least one other measurable quantity. The resulting picture is that of a complete dense linear order, and one reconstruction explicitly states that the measurables are thereby isomorphic to the modern real line (Sinkevich, 2015).

4. Infinite quantities and the part-whole principle

Bolzano’s treatment of infinity is structurally different from Cantorian cardinality. The key opposition, as reconstructed in later work, is between the Cantor Principle (two sets have the same size iff there is a bijection between them) and the Part-Whole Principle (a proper subset is strictly smaller than the whole). For infinite collections these conflict, as illustrated by Galileo’s paradox with $\times$0 and the squares. Bolzano sides with the Part-Whole Principle and bases comparison not on arbitrary bijection but on what the later literature calls the “determining ground” or “way of being formed” (Trlifajová, 2019).

In Paradoxes of the Infinite, Bolzano compares infinite series termwise. For

$\times$1

both series have the same multitude of terms, but $\times$2 is eventually positive and unbounded for every finite $\times$3. Hence $\times$4 is “infinitely greater” than $\times$5. Equality of infinite series, and thus of infinite pluralities, holds only when they share the same determining ground; mere bijection is not sufficient (Trlifajová, 2019).

Modern reconstructions extend this program constructively for countable sets. A calculable set is written as a disjoint union of finite layers $\times$6, with characteristic sequence $\times$7 and size sequence

$\times$8

Equality and strict comparison are then defined by eventual agreement and eventual strict inequality: $\times$9

$\div$0

This yields a partially ordered, non-Archimedean, commutative semiring. Under canonical labeling, $\div$1, while the evens and odds are both $\div$2 than $\div$3 but are incomparable with each other (Trlifajová, 2019).

A parallel algebraic rehabilitation appears in later work on “Bolzano’s Infinite Quantities.” There, a fragment of Bolzano’s theory is extended to either a linearly ordered integral domain $\div$4, built from polynomial-term sequences modulo eventual equality, or a partially ordered ring $\div$5 based on the Fréchet filter. The former is non-Archimedean and linearly ordered; the latter contains infinitesimals, infinitely great elements, and zero-divisors, and is related in the paper to both Robinson’s nonstandard analysis and Tao’s “cheap” nonstandard analysis (Trlifajová, 2017).

5. Named theorem traditions: intermediate value, Banach-space zeroes, and Bolzano–Weierstraß

Bolzano’s name persists in several theorem schemata that substantially postdate his manuscripts. In one dimension, Bolzano’s intermediate-value theorem asserts that if $\div$6 is continuous and $\div$7, then there exists $\div$8 with $\div$9. A modern extension due to Ariza-Ruiz, García-Falset, and Reich replaces the interval by a bounded subset of a Banach space and the sign-change condition by a boundary-sign condition formulated through a bilinear functional $S$0 satisfying positivity and homogeneity. Under complete continuity and the existence of $S$1 such that $S$2 has constant sign on $S$3, one obtains

$S$4

and under an additional boundary condition, $S$5. The classical one-dimensional theorem is recovered by taking $S$6 and $S$7 (Ariza-Ruiz et al., 2018).

A related infinite-dimensional line of development treats abstract equations

$S$8

with $S$9 a closed densely defined linear operator in a Banach space, $q\in\mathbb N$0 a closed convex constraint set invariant under resolvents, and $q\in\mathbb N$1 a bounded H-upper semicontinuous set-valued perturbation with convex weakly compact values. Under tangency conditions expressed through the Bouligand or Clarke tangent cone, compactness of the resolvent, and, in the nonconvex case, nonzero Euler characteristic, the resulting theorems are presented as infinite-dimensional variants of Bolzano and Miranda–Poincaré. The applications include strong steady-state solutions to Dirichlet, Neumann, and periodic boundary problems for elliptic partial differential inclusions, as well as second-order ODE boundary-value problems, without monotonicity or cooperativity assumptions (Kryszewski et al., 2016).

In logic and computability, Kreuzer analyzes the Bolzano–Weierstraß principle

$q\in\mathbb N$2

and a weak variant asserting only the existence of a Cauchy subsequence. Over $q\in\mathbb N$3, full BW is instance-wise equivalent to $q\in\mathbb N$4-WKL and equivalent to $q\in\mathbb N$5; the weak variant $q\in\mathbb N$6 is instance-wise equivalent to the strong cohesive principle $q\in\mathbb N$7. A principal corollary is the strict separation

$q\in\mathbb N$8

The same analysis shows that $q\in\mathbb N$9 does not solve the halting problem and does not lead to more than primitive recursive growth, whereas BW has the higher computational strength associated with $p$0-WKL (Kreuzer, 2010).

6. Contemporary reuse of the name: the Bolzano LLM system

In 2026, “Bolzano” also designates an open-source multi-agent research assistant built on state-of-the-art LLMs such as GPT, Gemini, and Claude. Its core architecture is a sequence of iterative research rounds with three phases: a prover phase with $p$1 agents in parallel proposing proof ideas, counterexamples, special cases, or partial proofs; a verifier phase with one agent that checks for logical gaps, repairs minor errors, and is the only component permitted to write to the knowledge base; and a summarizer phase that produces a concise human-readable status report. Across rounds the system maintains three persistent files, notes.txt, proofs.txt, and output.txt, all re-read at the start of each round (Grebík et al., 18 Apr 2026).

The case-study paper reports six research problems in mathematics and theoretical computer science. These include a black-box separation showing that black-box $p$2 is not Turing-closed, structural results on multi-slope tilings in $p$3, special soundness for multi-point multi-polynomial KZG batching, a strengthened inequality for weak and strong working set properties for heaps,

$p$4

a counterexample and replacement theorem for partitioning under function preimage constraints, and NP-completeness plus an $p$5 dynamic program on forests for optimization in the KKOS cultural dynamics model. Using the significance-autonomy taxonomy of Feng et al., four of the six results are classified as “publishable research,” and three as essentially autonomous (Grebík et al., 18 Apr 2026).

The authors emphasize that every generated proof was verified by domain experts before being declared complete, and they explicitly note limitations: human choice of problems and high-level strategy remains crucial, and the relative advantage of the multi-agent architecture over single-session chatbots is not yet quantified. A plausible implication is that the name “Bolzano” now marks not only a nineteenth-century foundational program but also a contemporary experimental paradigm for AI-assisted mathematical research (Grebík et al., 18 Apr 2026).