Minimal complex and minimal fibration

Intuition [Beginner]

A simplicial set can contain redundant simplices -- degenerate copies that carry no new information about the shape. A minimal complex is a simplicial set that has been stripped down to its essential simplices, with every simplex uniquely determined by its geometric position.

Think of a triangle made of wire. You could describe it with just three edges and three vertices. But you could also add extra edges that run alongside the real ones -- degenerate copies that do not change the shape. A minimal complex removes all such redundancy. Every non-degenerate simplex is geometrically essential.

A minimal fibration is a Kan fibration that is "as efficient as possible" -- the lifting of horns to fillers is unique up to homotopy. In a general Kan fibration, there may be many different ways to fill a horn, but in a minimal fibration, there is only one essential filler.

The main theorem is that every Kan complex contains a minimal subcomplex as a deformation retract, and every Kan fibration can be factored through a minimal fibration. Minimal complexes are unique up to isomorphism for a given homotopy type.

Visual [Beginner]

On the left, a simplicial circle drawn with multiple redundant edges (three edges connecting the same two vertices, with two being degenerate copies). An arrow points to a minimal simplicial circle on the right, with exactly one vertex and one non-degenerate edge. The redundant simplices have been stripped away, leaving the geometric core.

The minimal complex captures the homotopy type with the fewest possible simplices.

Worked example [Beginner]

Minimal model of $S^1$. The standard simplicial circle has one vertex $\ast$ and one non-degenerate edge $e$ with $d_0(e)=d_1(e)=\ast$. This is already minimal: you cannot remove $e$ without changing the homotopy type, and there are no other non-degenerate simplices. The degenerate simplices (like $s_0(\ast )$, $s_0(e)$, $s_1(e)$, etc.) do not contribute to the minimal complex.

Minimal model of the 2-sphere. A minimal simplicial model of $S^2$ has one vertex $\ast$, one non-degenerate 2-simplex $\sigma$, and no non-degenerate edges or 1-simplices. The boundary of $\sigma$ satisfies $d_0(\sigma )=d_1(\sigma )=d_2(\sigma )=s_0(\ast )$ (all faces are the degenerate edge at $\ast$). This has the homotopy type of $S^2$ with just one non-degenerate simplex.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Definition (Minimal simplicial set). A simplicial set $X$ is minimal if for any two simplices $x,y\in X_n$ that satisfy $d_{i}x=d_{i}y$ for all $i$ and $\gamma (x)=\gamma (y)$ for some homotopy relation induced by the simplicial operators, then $x=y$. Equivalently, $X$ is minimal if it has no pair of distinct non-degenerate simplices sharing the same boundary in the same homotopy class.

Definition (Minimal fibration). A Kan fibration $p:E\to B$ is minimal if for any two fillers of the same horn -- i.e., two simplices $x,y\in E_n$ with $d_{i}x=d_{i}y$ for all $i\ne k$ and $p(x)=p(y)$ -- the simplices $x$ and $y$ are equal. Equivalently, the fibre-wise filling is unique.

Theorem (Existence of minimal subcomplexes). Every Kan complex $X$ contains a minimal simplicial subset $X^{\min }$ that is a deformation retract of $X$. Moreover, $X^{\min }$ is unique up to isomorphism.

Key theorem with proof [Intermediate+]

Theorem (Factorisation via minimal fibrations). Every Kan fibration $p:E\to B$ factors as $p=q\circ r$ where $r:E\to E^{\prime }$ is a simplicial homotopy equivalence and $q:E^{\prime }\to B$ is a minimal fibration.

Proof sketch. Define $E^{\prime }$ as the quotient of $E$ that identifies simplices related by fibre-wise homotopy. Specifically, two simplices $x,y\in E_n$ with $p(x)=p(y)$ are equivalent if they share the same boundary $d_{i}x=d_{i}y$ for all $i$ and are connected by a fibre-wise homotopy (a homotopy that projects to the constant homotopy in $B$). The quotient map $r:E\to E^{\prime }$ identifies equivalent simplices, and $q:E^{\prime }\to B$ is induced by $p$.

The map $q$ is a minimal fibration because the equivalence relation has identified all fibre-wise homotopic fillers, leaving at most one filler per horn. The map $r$ is a homotopy equivalence because the equivalence relation preserves the homotopy type (each fibre is collapsed to its minimal representative without changing the connectivity). $\square$

Bridge. The minimal-complex construction refines the skeletal decomposition of 03.12.24 by removing degenerate redundancies while preserving homotopy type, paralleling how the CW-approximation of 03.12.26 strips away point-set pathology while preserving homotopy groups. The uniqueness of the minimal model echoes the uniqueness of the Thom class in 03.12.29 -- in both cases, a canonical representative is extracted from a homotopy class.

Exercises [Intermediate+]

Advanced results [Master]

Uniqueness of minimal models. If $X$ and $Y$ are minimal Kan complexes and $f:X\to Y$ is a weak homotopy equivalence, then $f$ is an isomorphism of simplicial sets. This is the simplicial analogue of the Whitehead theorem: a weak equivalence between minimal models must be a genuine isomorphism, not merely a homotopy equivalence.

Minimal fibrations and fibre-wise structure. A minimal fibration $p:E\to B$ has fibres that are minimal Kan complexes, and the fibre-wise homotopy type is locally constant on $B$. Over each simplex of $B$, the fibre is uniquely determined up to isomorphism, giving a "rigid" fibration theory.

Synthesis. Minimal complexes and fibrations provide the canonical forms for simplicial homotopy types; the existence theorem refines the CW-approximation machinery of 03.12.26 by producing a unique combinatorial representative for each homotopy type within the simplicial framework of 03.12.24, the factorisation through a minimal fibration dualises the Thom-class uniqueness of 03.12.29 in the sense that both extract canonical representatives from homotopy classes, and the rigidity of minimal models ensures that computations performed on them (homology, homotopy groups) give unambiguous answers. The theory connects to bisimplicial sets 03.12.36 where the diagonal of a bisimplicial set admits a minimal model computed from the simplicial direction.

Full proof set [Master]

Proposition (Deformation retraction onto minimal subcomplex). Let $X$ be a Kan complex. Then $X$ contains a minimal subcomplex $X^{\min }$ and a deformation retraction $r:X\to X^{\min }$ that is a simplicial homotopy equivalence.

Proof. Construct $X^{\min }$ skeleton by skeleton. In dimension 0, choose one vertex from each path component. Given $X^{\min }$ through dimension $n-1$, for each $(n-1)$-simplex $\sigma$ in $X_{n-1}^{\min }$, consider all non-degenerate $n$-simplices of $X$ with boundary $\sigma$. These simplices fall into homotopy classes (two such $n$-simplices are equivalent if they differ by a simplicial homotopy relative to their boundary). Choose exactly one representative from each class and adjoin it to $X_n^{\min }$. The resulting $X^{\min }$ is a simplicial subset of $X$.

The retraction $r:X\to X^{\min }$ is defined by sending each simplex to its minimal representative: in dimension 0, send each vertex to the chosen basepoint of its component. Inductively, for $x\in X_n$, the faces $d_{i}x$ already have minimal images by induction, and $x$ is sent to the unique simplex in $X_n^{\min }$ in the same homotopy class with the same boundary. The Kan filling condition ensures this is well-defined. A simplicial homotopy $H:X\times {\Delta }^1\to X$ from $\mathrm{id}$ to $r\circ i$ (where $i:X^{\min }\hookrightarrow X$) is constructed using the Kan property to fill horn-shaped homotopies at each dimension. $\square$

Connections [Master]

Simplicial sets 03.12.24 provide the ambient category in which minimality is defined; the Eilenberg-Zilber decomposition into degenerate and non-degenerate simplices is the structural input for the construction.

CW approximation 03.12.26 produces a combinatorial model for any space, and the minimal complex produces the canonical such model -- the unique representative with no redundancy.

Bisimplicial sets 03.12.36 and the realisation lemma use minimal models to control the homotopy type of the diagonal; the Bousfield-Kan tower of a bisimplicial set is built from minimal fibrations at each stage.

Bibliography [Master]

@article{barrett-moore1966,
  author = {Barrett, Charles S. and Moore, John C.},
  title = {Minimal complexes and minimal fibrations},
  journal = {Illinois J. Math.},
  volume = {10},
  pages = {431--447},
  year = {1966}
}

@book{goerss-jardine1999,
  author = {Goerss, Paul G. and Jardine, John F.},
  title = {Simplicial Homotopy Theory},
  publisher = {Birkh{\"a}user},
  year = {1999}
}

@book{may-simplicial,
  author = {May, J. Peter},
  title = {Simplicial Objects in Algebraic Topology},
  publisher = {University of Chicago Press},
  year = {1967}
}

@book{gabriel-zisman1967,
  author = {Gabriel, Peter and Zisman, Michel},
  title = {Calculus of Fractions and Homotopy Theory},
  publisher = {Springer},
  year = {1967}
}

Historical & philosophical context [Master]

The theory of minimal complexes and minimal fibrations was developed by Barrett and Moore in the mid-1960s, building on earlier work by Eilenberg and Zilber on the decomposition of simplicial sets. The 1966 paper by Barrett and Moore ^{[Barrett-Moore 1966]} established the existence and uniqueness theorems.

The philosophical significance is that minimal complexes provide a "canonical form" for homotopy types. Just as every integer has a unique prime factorisation, every homotopy type has a unique minimal simplicial model (up to isomorphism). This canonical form makes it possible to compare homotopy types by comparing their minimal models as simplicial sets, reducing homotopy-theoretic questions to combinatorial ones. The minimal-complex theory is the simplicial-set precursor to Sullivan's minimal models in rational homotopy theory, which apply the same philosophy to differential-graded algebras.