CISTs: Completely Independent Spanning Trees

• CISTs are spanning trees where every pair of vertices connects via unique, pairwise edge-disjoint and internally vertex-disjoint paths, ensuring robust fault tolerance.
• They can be characterized through local conditions such as Hasunuma’s rule and Araki’s partition theorem, linking tree structure to connected dominating sets and hypergraph colorings.
• Algorithmic methods for constructing CISTs vary by graph class, with complexities ranging from polynomial solutions in bounded-treewidth graphs to NP-completeness in general and split graphs.

Completely independent spanning trees (CISTs) are spanning trees $T_1,\dots,T_k$ of a connected graph $G$ such that, for every pair of vertices $u,v\in V(G)$, the unique $u$–$v$ paths in the trees are pairwise edge-disjoint and pairwise internally vertex-disjoint. In the standard root-free formulation, this means

$$E(P_{T_i}(u,v))\cap E(P_{T_j}(u,v))=\emptyset,\qquad V(P_{T_i}(u,v))\cap V(P_{T_j}(u,v))=\{u,v\}$$

for all $i\neq j$. A fundamental equivalent characterization states that $T_1,\dots,T_k$ are CISTs if and only if they are pairwise edge-disjoint and, for every vertex $x\in V(G)$, there is at most one index $i$ with $G$0 (Lalou et al., 17 Dec 2025). This notion is stronger than rooted independent spanning trees, because it requires path independence for all vertex pairs rather than only for root-to-vertex paths (Yuan et al., 23 Dec 2025).

1. Definition, scope, and terminological variants

The standard graph-theoretic definition treats CISTs as a root-free all-pairs object: the independence condition is imposed simultaneously for every pair of vertices of the graph. Hasunuma’s characterization makes the condition local: edge-disjointness plus the rule that any vertex can be internal in at most one tree is equivalent to full complete independence (Lalou et al., 17 Dec 2025).

This all-pairs requirement is strictly stronger than the rooted independent spanning tree condition. In the rooted setting, one fixes a root $G$1 and asks only that the $G$2–$G$3 paths in distinct trees be internally disjoint for each $G$4. By contrast, CISTs require internal disjointness of the $G$5–$G$6 paths for every pair $G$7, so the rooted condition does not imply the root-free one (Yuan et al., 23 Dec 2025).

The literature is not fully uniform in terminology. In random and pseudorandom graph work, “completely independent spanning trees” may refer to rooted vertex-independence, and edge-disjointness is not always built into the definition; the same paper explicitly distinguishes this from stronger edge-disjoint variants (Draganić et al., 30 Sep 2025). Related network papers also separate CISTs from weaker notions such as completely edge-independent spanning trees (CEISTs), which require only pairwise edge-disjoint spanning trees (Li et al., 2024). This suggests that precise statement of the adopted definition is essential in any technical comparison.

2. Structural characterizations

A second major characterization is Araki’s partition theorem. A partition $G$8 is a $G$9-CIST-partition if each induced subgraph $u,v\in V(G)$0 is connected and, for every $u,v\in V(G)$1, every connected component of the bipartite subgraph $u,v\in V(G)$2 satisfies $u,v\in V(G)$3, equivalently, no such component is a tree. A connected graph admits $u,v\in V(G)$4 CISTs if and only if it admits a $u,v\in V(G)$5-CIST-partition (Lalou et al., 17 Dec 2025).

Another characterization uses connected dominating sets. A graph has $u,v\in V(G)$6 CISTs if and only if there exist $u,v\in V(G)$7 disjoint connected dominating sets $u,v\in V(G)$8 such that every component of the bipartite subgraph $u,v\in V(G)$9 has a cycle for all $u$0 (Hasunuma, 2022). This formulation is particularly useful in line graphs and in constructive existence proofs, because it converts path-independence into a combination of domination, connectivity, and cycle conditions across pairwise interfaces.

For split graphs, the structure becomes hypergraph-theoretic. If $u$1 is a split graph, its associated hypergraph is $u$2. If $u$3 has $u$4 CISTs, then $u$5 is panchromatically $u$6-colorable; if $u$7 is bipanchromatically $u$8-colorable, then $u$9 has $v$0 CISTs. If $v$1 is the maximum number of CISTs in $v$2, then

$v$3

so the CIST number of a split graph lies in a sharp window controlled by bipanchromatic colorings of its associated hypergraph (Lalou et al., 17 Dec 2025).

3. Generalizations and related notions

CISTs are the spanning case of a broader Steiner theory. For a terminal set $v$4 with $v$5, an $v$6-Steiner tree is a subtree spanning $v$7 whose leaves all belong to $v$8. A family of $v$9-Steiner trees is completely independent if, for every pair of terminals $$E(P_{T_i}(u,v))\cap E(P_{T_j}(u,v))=\emptyset,\qquad V(P_{T_i}(u,v))\cap V(P_{T_j}(u,v))=\{u,v\}$$0, the $$E(P_{T_i}(u,v))\cap E(P_{T_j}(u,v))=\emptyset,\qquad V(P_{T_i}(u,v))\cap V(P_{T_j}(u,v))=\{u,v\}$$1–$$E(P_{T_i}(u,v))\cap E(P_{T_j}(u,v))=\emptyset,\qquad V(P_{T_i}(u,v))\cap V(P_{T_j}(u,v))=\{u,v\}$$2 paths in different trees are pairwise edge-disjoint and internally vertex-disjoint. When $$E(P_{T_i}(u,v))\cap E(P_{T_j}(u,v))=\emptyset,\qquad V(P_{T_i}(u,v))\cap V(P_{T_j}(u,v))=\{u,v\}$$3, this definition specializes exactly to CISTs (Yuan et al., 23 Dec 2025). The associated packing number is $$E(P_{T_i}(u,v))\cap E(P_{T_j}(u,v))=\emptyset,\qquad V(P_{T_i}(u,v))\cap V(P_{T_j}(u,v))=\{u,v\}$$4, and the generalized $$E(P_{T_i}(u,v))\cap E(P_{T_j}(u,v))=\emptyset,\qquad V(P_{T_i}(u,v))\cap V(P_{T_j}(u,v))=\{u,v\}$$5-connectivity is

$$E(P_{T_i}(u,v))\cap E(P_{T_j}(u,v))=\emptyset,\qquad V(P_{T_i}(u,v))\cap V(P_{T_j}(u,v))=\{u,v\}$$6

The Steiner formulation preserves the same structural core. A family of $$E(P_{T_i}(u,v))\cap E(P_{T_j}(u,v))=\emptyset,\qquad V(P_{T_i}(u,v))\cap V(P_{T_j}(u,v))=\{u,v\}$$7-Steiner trees is completely independent if and only if the trees are pairwise edge-disjoint and their sets of internal vertices are pairwise disjoint (Maheshwari et al., 21 Apr 2026). It also admits a directed analogue: completely independent spanning arborescences (CISA) in a directed $$E(P_{T_i}(u,v))\cap E(P_{T_j}(u,v))=\emptyset,\qquad V(P_{T_i}(u,v))\cap V(P_{T_j}(u,v))=\{u,v\}$$8-minor are equivalent, in the appropriate sense, to completely independent Steiner trees in the undirected host graph (Maheshwari et al., 21 Apr 2026). This directed correspondence extends the CIST framework beyond undirected spanning trees.

Several relaxations have been studied. The $$E(P_{T_i}(u,v))\cap E(P_{T_j}(u,v))=\emptyset,\qquad V(P_{T_i}(u,v))\cap V(P_{T_j}(u,v))=\{u,v\}$$9-disjoint spanning tree framework defines $i\neq j$0 to be $i\neq j$1-disjoint when at most $i\neq j$2 vertices are internal in more than one tree and at most $i\neq j$3 edges are shared by more than one tree. In that language, CISTs are exactly the $i\neq j$4-disjoint case; $i\neq j$5 captures internally vertex-disjoint spanning trees, and $i\neq j$6 captures edge-disjoint spanning trees (Darties et al., 2017). On the other side, CEISTs impose only pairwise edge-disjointness and are strictly weaker than CISTs (Li et al., 2024).

4. Existence theorems, exact values, and graph classes

Research on CISTs combines exact enumeration in highly structured graphs with upper bounds in sparse or topologically restricted families.

Graph family	Main CIST result	Source
Complete graph $i\neq j$7	$i\neq j$8	(Yuan et al., 23 Dec 2025)
Line graph $i\neq j$9	$T_1,\dots,T_k$0 for $T_1,\dots,T_k$1	(Hasunuma, 2022)
Hypercube $T_1,\dots,T_k$2	$T_1,\dots,T_k$3 CISTs, each of diameter $T_1,\dots,T_k$4	(Shaw, 2024)
Planar graphs	Any planar graph has at most $T_1,\dots,T_k$5 CISTs	(Maheshwari et al., 21 Apr 2026)
Treewidth-$T_1,\dots,T_k$6 graphs	Any graph of treewidth at most $T_1,\dots,T_k$7 has at most $T_1,\dots,T_k$8 CISTs	(Maheshwari et al., 21 Apr 2026)
3-connected 2-outerplanar triangulated discs	Two CISTs always exist	(Araki, 11 Jun 2026)

For complete graphs, the Steiner generalization yields the exact formula

$T_1,\dots,T_k$9

for $x\in V(G)$0 and $x\in V(G)$1, so the spanning case gives

$x\in V(G)$2

This agrees with the classical CIST packing number in $x\in V(G)$3 (Yuan et al., 23 Dec 2025). In line graphs, the exact number can increase: for every $x\in V(G)$4,

$x\in V(G)$5

and this value is optimal (Hasunuma, 2022).

Hypercubes provide a second major benchmark. For each fixed $x\in V(G)$6, sufficiently high-dimensional hypercubes $x\in V(G)$7 contain $x\in V(G)$8 CISTs, and more strongly there exist

$x\in V(G)$9

pairwise completely independent spanning trees in $i$0, each with diameter

$i$1

Since every spanning tree of $i$2 has diameter at least $i$3, this diameter is asymptotically optimal (Shaw, 2024). A complementary construction shows that $i$4 has three CISTs for every $i$5, with diameter bounds at most $i$6, $i$7, and $i$8 for the three trees (Barabde et al., 2024).

Topological restrictions yield sharp thresholds. In planar graphs, any CIST packing has size at most $i$9; in graphs of bounded treewidth $G$00, the CIST number is at most $G$01 (Maheshwari et al., 21 Apr 2026). For 3-connected 2-outerplanar triangulated discs, two CISTs always exist, while there is a 3-connected 4-outerplanar triangulated disc with no two CISTs (Araki, 11 Jun 2026). At the level of degree-sum conditions, if $G$02 is a connected graph of order $G$03 and

$G$04

then $G$05 has two CISTs (Ma et al., 17 Feb 2025).

5. Complexity and algorithmic methods

The general decision problem is computationally hard. Determining whether a graph contains $G$06 CISTs is NP-complete, even for $G$07 (Ma et al., 17 Feb 2025). This hardness persists in highly structured classes: deciding whether a split graph has two CISTs is NP-complete, and the related hypergraph problem of bipanchromatic $G$08-colorability is also NP-complete (Lalou et al., 17 Dec 2025).

The Steiner generalization strengthens these hardness results. For any fixed number of terminals $G$09 and any fixed $G$10, the decision problem asking whether a graph contains $G$11 completely independent Steiner trees is NP-complete. The corresponding maximization problem is NP-hard to approximate within $G$12 and is APX-hard already for $G$13 (Maheshwari et al., 21 Apr 2026). These results reinforce the view that CIST packing is substantially harder than ordinary spanning-tree packing.

Despite this, several algorithmic regimes are tractable. For fixed terminal number $G$14 and fixed packing size $G$15, completely independent Steiner trees can be decided in polynomial time by enumerating constant-size templates and reducing feasibility to a constant-size vertex-disjoint paths problem (Maheshwari et al., 21 Apr 2026). For bounded-treewidth graphs and fixed $G$16, existence is expressible in monadic second-order logic, so Courcelle’s theorem yields linear-time algorithms (Maheshwari et al., 21 Apr 2026). In split graphs, integer linear programming formulations are given for $G$17, $G$18, and $G$19, providing exact computation for small-to-medium instances (Lalou et al., 17 Dec 2025).

On the constructive side, several existence proofs are algorithmic. The complete-graph and complete-bipartite CISST constructions are polynomial-time and therefore yield polynomial constructions of optimal CIST families in the spanning case (Yuan et al., 23 Dec 2025). In hypercubes, the explicit deterministic construction producing $G$20 CISTs can be implemented in total time $G$21 to output all trees (Shaw, 2024).

6. Network applications, resilience, and open directions

The principal motivation for CISTs is fault-tolerant multipath communication. Because for every pair $G$22 the $G$23–$G$24 paths across the trees are pairwise edge-disjoint and internally vertex-disjoint, up to $G$25 arbitrary edge failures or up to $G$26 internal-vertex failures, excluding the endpoints, still leave at least one intact tree-path between $G$27 and $G$28 (Maheshwari et al., 21 Apr 2026). In the Steiner setting, the same mechanism focuses redundancy on a specified terminal set, which is relevant to multicast, backbone, and service-node architectures (Yuan et al., 23 Dec 2025).

The Steiner papers also make the design contrast explicit. Completely independent $G$29-Steiner trees provide stronger fault tolerance than internally disjoint $G$30-Steiner trees, because complete independence forbids shared internal bottlenecks on terminal-to-terminal routes. The emergency communication example with terminals $G$31 (Command Center), $G$32 (Hospital), $G$33 (Fire Station)$G$34 shows that an internally disjoint design may still route $G$35–$G$36 communication through $G$37, whereas a CISST design can preserve $G$38–$G$39 connectivity even if $G$40 fails (Yuan et al., 23 Dec 2025).

Implementation-oriented work in dense Gaussian networks illustrates the performance angle. A recent construction of two CISTs partitions the dense Gaussian network, builds the first tree, obtains the second by rotation, and reports an improvement of at least $G$41 in the average maximum number of steps required to deliver a message from the root node to all other nodes, relative to existing approaches (Hussain et al., 22 Jun 2026). This suggests that CISTs are not only a resilience certificate but also a latency-sensitive routing primitive in symmetric interconnection networks.

Several open directions remain central. For hypercubes, the exact asymptotic constant $G$42 such that $G$43 has $G$44 CISTs is unknown; the best uniform lower bound is $G$45, while the trivial edge-count upper bound gives $G$46 (Shaw, 2024). The question of whether $G$47 has three CISTs was left open by the explicit tri-CIST construction for $G$48, $G$49 (Barabde et al., 2024). In locally twisted cubes, the stronger CIST problem remains open even though the optimal number of CEISTs is known (Li et al., 2024). More broadly, tightening the sufficient connectivity bound

$G$50

for guaranteeing $G$51 CISTs, and obtaining sharper characterizations in planar, bounded-treewidth, and network-specific graph classes, remain active themes (Maheshwari et al., 21 Apr 2026).