Qplex Theories in Quantum Foundations

• Qplex theories are probabilistic frameworks defined within a constrained subset of the full probability simplex using QBist overlap bounds and the Urgleichung.
• They decouple a core probabilistic-geometric structure from the full Hilbert-space formalism, enabling reconstruction of quantum state spaces via SICs and generalized POVMs.
• The framework offers practical insights into quantum correlations, reproducing bounds like Tsirelson's limit in bipartite tests and exposing superquantum effects in multi-outcome scenarios.

Searching arXiv for papers on qplexes and related quantum-foundational work. Qplex theories are a family of probabilistic theories motivated by QBism, in which the admissible states are not the full probability simplex but a geometrically constrained subset determined by QBist overlap bounds and the Urgleichung. In this framework, quantum state space in symmetric informationally complete coordinates is a distinguished example—a Hilbert qplex—but not the only one. Qplex theories therefore separate a probabilistic-geometric core from specifically Hilbert-space structure, making them a reconstruction program for finite-dimensional quantum theory and, more recently, a testbed for studying which quantum-correlation constraints follow from qplex geometry alone (Appleby et al., 2016, Gupta et al., 5 Jun 2026).

1. QBist origin and the Urgleichung

Qplex theory originates in QBism’s reformulation of quantum theory as a normative constraint on an agent’s probability assignments rather than a theory of ontic microstates. The starting point is an informationally complete reference measurement, ideally a symmetric informationally complete POVM (SIC). In dimension $d$, a SIC has $d^2$ outcomes and projectors $\{\Pi_i\}_{i=1}^{d^2}$ satisfying

$$\operatorname{tr}(\Pi_i\Pi_j)=\frac{d\delta_{ij}+1}{d+1}.$$

A density operator $\rho$ is represented by the probability vector

$p(i)=\frac{1}{d}\operatorname{Tr}(\rho\Pi_i),$

so quantum states become points in the simplex

$$\Delta_{d^2}=\left\{p\in \mathbb{R}^{d^2}: p(i)\ge 0,\ \sum_i p(i)=1\right\}.$$

In SIC coordinates, the Born rule becomes the Urgleichung,

$$q(j)=\sum_i \left[(d+1)p(i)-\frac{1}{d}\right] r(j|i),$$

with $r(j|i)=\operatorname{Tr}(E_j\Pi_i)$ for an actual measurement $\{E_j\}$. QBism interprets this not as a dynamical law for hidden variables, but as a consistency condition relating probabilities for hypothetical and actual experiments. The qplex program asks what state-space structure is already forced by this probabilistic rewriting, prior to imposing the full Hilbert-space formalism (Appleby et al., 2016).

This shift is conceptually decisive. Probability theory remains the background calculus, but the admissible probability assignments are no longer arbitrary. The simplex is “cropped” by nonclassical consistency conditions. In this sense, qplex theories are not noncommutative probability theories in disguise; they are ordinary probability theory supplemented by constraints motivated by the QBist reading of the Born rule (Appleby et al., 2016).

2. Geometric definition of a qplex

A qplex is defined inside the $d^2$0-outcome probability simplex by the fundamental inequalities

$d^2$1

A set of probability vectors is called consistent if every pair satisfies these inequalities. A germ is any pairwise consistent subset of the simplex, and a maximal germ is one to which no further point can be added without violating consistency. In the qplex formalism, maximal germs are exactly the self-polar sets with respect to the induced polarity; equivalently, a qplex is a self-polar subset of the out-ball in the probability simplex, with the generalized-urgleichung parameters fixed to $d^2$2 and $d^2$3 (Appleby et al., 2016).

This gives qplexes a rigid convex geometry. Every qplex is closed and convex, contains the center $d^2$4, contains the basis simplex and the in-ball, and is contained in the out-ball. Self-polarity implies

$d^2$5

so the qplex is determined by the half-spaces induced by its own points. The in-ball and out-ball are mutually polar, and the qplex sits between them as a self-dual convex body in SIC probability coordinates (Appleby et al., 2016).

Several geometric notions parallel standard quantum concepts. A point is pure when

$d^2$6

so pure points lie on the out-sphere. Two pure points are maximally distant when they saturate the lower bound,

$d^2$7

Sets of pairwise maximally distant pure points have cardinality at most $d^2$8; these are the mutually maximally distant sets. The framework also yields nontrivial combinatorial constraints: no component of any qplex state can exceed $d^2$9, and the number of zero entries of any qplex state is at most $\{\Pi_i\}_{i=1}^{d^2}$0 (Appleby et al., 2016).

A crucial structural fact is that qplexes are much broader than quantum state spaces. Quantum state space in SIC coordinates is a qplex, but the fundamental inequalities and maximality do not determine it uniquely. The theory proves that there are uncountably many non-isomorphic qplexes. Consequently, qplex theory isolates a substantial probabilistic-geometric core of quantum theory while leaving open what additional principle singles out the Hilbert case (Appleby et al., 2016).

3. Hilbert qplexes and the reconstruction of quantum theory

A Hilbert qplex is a qplex isomorphic to the density-matrix state space of a $\{\Pi_i\}_{i=1}^{d^2}$1-dimensional quantum system. The decisive reconstruction theorem states that a qplex $\{\Pi_i\}_{i=1}^{d^2}$2 is isomorphic to quantum state space iff there exists an $\{\Pi_i\}_{i=1}^{d^2}$3-linear bijection $\{\Pi_i\}_{i=1}^{d^2}$4 such that

$\{\Pi_i\}_{i=1}^{d^2}$5

and

$\{\Pi_i\}_{i=1}^{d^2}$6

Moreover, such an isomorphism exists iff there is a SIC $\{\Pi_i\}_{i=1}^{d^2}$7 with

$\{\Pi_i\}_{i=1}^{d^2}$8

Thus Hilbert qplexes are exactly SIC representations of quantum state space (Appleby et al., 2016).

The reconstruction then turns on symmetry. The preservation group $\{\Pi_i\}_{i=1}^{d^2}$9 of a qplex is the group of qplex-preserving measurement maps, and the theory proves that this is exactly the full symmetry group of the qplex. For Hilbert qplexes,

$$\operatorname{tr}(\Pi_i\Pi_j)=\frac{d\delta_{ij}+1}{d+1}.$$0

and, more sharply, a qplex is Hilbert iff its preservation group contains a subgroup isomorphic to $$\operatorname{tr}(\Pi_i\Pi_j)=\frac{d\delta_{ij}+1}{d+1}.$$1. The qplex program therefore splits quantum reconstruction into two layers: the probabilistic-geometric layer yields the qplex, while a quantum-specific symmetry principle selects the Hilbert qplex (Appleby et al., 2016).

This symmetry criterion also reframes the SIC existence problem. The theory introduces quasi-SICs—operator families satisfying the SIC trace relations without assuming positivity—and shows that if a qplex has $$\operatorname{tr}(\Pi_i\Pi_j)=\frac{d\delta_{ij}+1}{d+1}.$$2 symmetry, the quasi-state space collapses to genuine quantum state space and the quasi-SIC becomes a true SIC. One corollary is that a SIC exists in dimension $$\operatorname{tr}(\Pi_i\Pi_j)=\frac{d\delta_{ij}+1}{d+1}.$$3 iff $$\operatorname{tr}(\Pi_i\Pi_j)=\frac{d\delta_{ij}+1}{d+1}.$$4 is isomorphic to a stochastic subgroup of $$\operatorname{tr}(\Pi_i\Pi_j)=\frac{d\delta_{ij}+1}{d+1}.$$5 (Appleby et al., 2016).

A common misconception is that qplexes are merely a reparametrization of ordinary quantum states. The formalism is broader. Hilbert qplexes recover finite-dimensional quantum theory, but general qplexes retain only part of the structure: SIC-based coordinates, convexity, pairwise overlap bounds, self-polarity, and measurement representations derived from the Urgleichung. They relax Hilbert-space realization, projective-unitary symmetry, and the operator/tensor-product framework ordinarily used for composites (Appleby et al., 2016, Gupta et al., 5 Jun 2026).

4. Generalised qplexes and morphophoric POVMs

The original qplex formalism is SIC-centric, but it admits a substantial generalisation. A POVM $$\operatorname{tr}(\Pi_i\Pi_j)=\frac{d\delta_{ij}+1}{d+1}.$$6 is called morphophoric when the image of quantum state space under the measurement map

$$\operatorname{tr}(\Pi_i\Pi_j)=\frac{d\delta_{ij}+1}{d+1}.$$7

is similar to $$\operatorname{tr}(\Pi_i\Pi_j)=\frac{d\delta_{ij}+1}{d+1}.$$8. Concretely, there must exist $$\operatorname{tr}(\Pi_i\Pi_j)=\frac{d\delta_{ij}+1}{d+1}.$$9 such that

$\rho$0

for all states $\rho$1. The resulting probability image $\rho$2 is a generalised qplex (Słomczyński et al., 2019).

The central characterization theorem states that $\rho$3 is morphophoric iff the traceless parts of its effects form a tight operator frame. In the rank-1 equal-trace case, this becomes especially transparent: a rank-1 equal-trace POVM is morphophoric iff it is a complex projective $\rho$4-design. SIC qplexes are then the minimal case $\rho$5, while complete sets of MUBs and other $\rho$6-design constructions generate larger generalised qplexes (Słomczyński et al., 2019).

The intrinsic geometry of a generalised qplex is the same as that of quantum state space, but its external geometry changes because the image now sits in a $\rho$7-dimensional affine section $\rho$8 of a larger simplex $\rho$9. Two associated polytopes organize this geometry: the basis polytope

$p(i)=\frac{1}{d}\operatorname{Tr}(\rho\Pi_i),$0

where $p(i)=\frac{1}{d}\operatorname{Tr}(\rho\Pi_i),$1 are the basis distributions, and the primal polytope

$p(i)=\frac{1}{d}\operatorname{Tr}(\rho\Pi_i),$2

They satisfy

$p(i)=\frac{1}{d}\operatorname{Tr}(\rho\Pi_i),$3

The generalised qplex remains self-dual, and the basis and primal polytopes are dual to each other with respect to the appropriate centered sphere (Słomczyński et al., 2019).

The QBist primal equation also extends. For a rank-1 equal-trace morphophoric POVM, one has

$p(i)=\frac{1}{d}\operatorname{Tr}(\rho\Pi_i),$4

which reduces to the SIC formula when $p(i)=\frac{1}{d}\operatorname{Tr}(\rho\Pi_i),$5. For arbitrary morphophoric POVMs, the formulation shifts from conditional probabilities to centered deviations from the maximally mixed state,

$p(i)=\frac{1}{d}\operatorname{Tr}(\rho\Pi_i),$6

This preserves the idea that probabilities for one measurement are constrained by probabilities relative to a reference measurement, but it no longer takes the SIC-specific law-of-total-probability form (Słomczyński et al., 2019).

The generalised theory thus broadens qplex geometry without abandoning its core invariants. Self-duality, inner and outer bounding structures, and the central role of measurement-induced probability bodies all persist, but SICs cease to be the only natural reference measurements. This widens the qplex program from a SIC reconstruction to a more general theory of quantum-state-space embeddings in probability space (Słomczyński et al., 2019).

5. Bipartite correlations and Bell scenarios

For a long time, qplex theory was primarily a single-system framework. Recent work extends it to bipartite Bell scenarios by expressing joint expectation values as inner products of suitably defined $p(i)=\frac{1}{d}\operatorname{Tr}(\rho\Pi_i),$7-vectors. The analysis is carried out entirely in qplex language: Bob updates his local SIC-coordinate state after learning Alice’s setting and outcome, and self-polarity is used to identify effects with states up to scaling. This reduces the correlation problem to the geometry of overlaps among qplex-compatible state vectors (Gupta et al., 5 Jun 2026).

In the CHSH scenario, the shared inner-product structure of the $p(i)=\frac{1}{d}\operatorname{Tr}(\rho\Pi_i),$8-vectors constrains the Bell expression

$p(i)=\frac{1}{d}\operatorname{Tr}(\rho\Pi_i),$9

Under the additional assumptions of unbiased marginals, maximally mixed local states, repeatable sharp measurements, and maximal norm saturation $$\Delta_{d^2}=\left\{p\in \mathbb{R}^{d^2}: p(i)\ge 0,\ \sum_i p(i)=1\right\}.$$0, the maximal value is

$$\Delta_{d^2}=\left\{p\in \mathbb{R}^{d^2}: p(i)\ge 0,\ \sum_i p(i)=1\right\}.$$1

This reproduces the Tsirelson bound without invoking Hilbert-space tensor products, operator commutation relations, or standard composite-system formalism. The paper is explicit, however, that this result does not follow from the bare pairwise overlap inequalities alone; it follows from the common inner-product representation of all four correlators in one vector space (Gupta et al., 5 Jun 2026).

The same geometric method behaves differently in the three-outcome CGLMP scenario. For the inequality $$\Delta_{d^2}=\left\{p\in \mathbb{R}^{d^2}: p(i)\ge 0,\ \sum_i p(i)=1\right\}.$$2, the expectation values are weighted by the complex root of unity $$\Delta_{d^2}=\left\{p\in \mathbb{R}^{d^2}: p(i)\ge 0,\ \sum_i p(i)=1\right\}.$$3, and the Bell expression can again be written in terms of $$\Delta_{d^2}=\left\{p\in \mathbb{R}^{d^2}: p(i)\ge 0,\ \sum_i p(i)=1\right\}.$$4-vectors. Yet the corresponding qplex-derived norm and inner-product constraints now permit the algebraic maximum: $$\Delta_{d^2}=\left\{p\in \mathbb{R}^{d^2}: p(i)\ge 0,\ \sum_i p(i)=1\right\}.$$5 The resulting correlations are superquantum. The paper interprets this as a structural limitation: pairwise overlap geometry, self-polarity, centeredness in the unbiased case, and unit-norm conditions are enough to recover CHSH’s quantum bound, but not enough to recover the full set of multi-outcome quantum correlation constraints (Gupta et al., 5 Jun 2026).

This split result sharpens the distinction between general qplexes and Hilbert qplexes. Two-outcome Bell inequalities may not separate them operationally, while multi-outcome inequalities can. The framework also respects no-signaling: averaging over Alice’s outcomes yields Bob’s marginal independently of Alice’s setting. Qplex theories therefore occupy an intermediate region between quantum theory and the full no-signaling set—more constrained than arbitrary no-signaling models, but not constrained enough to exclude all superquantum points (Gupta et al., 5 Jun 2026).

6. Scope, ambiguities, and open problems

The central open problem of qplex theories is the same one that motivated the framework from the beginning: what additional principle singles out Hilbert qplexes from the much larger class of qplexes? The original reconstruction identifies projective-unitary symmetry as one sufficient answer, but not as the final conceptual explanation. The recent Bell-inequality analysis adds a more operational diagnosis: higher-outcome scenarios probe cyclic and complex-phase relations not controlled by pairwise overlaps alone. This suggests that the missing ingredient is an additional geometric, algebraic, or symmetry-based constraint, especially one visible in multi-outcome composite experiments (Appleby et al., 2016, Gupta et al., 5 Jun 2026).

A second open direction concerns abstraction beyond the SIC case. Generalised qplexes generated by morphophoric POVMs preserve the intrinsic geometry of quantum state space while altering its embedding into probability space. This suggests that the essential object may be broader than SIC probability bodies, but it also raises the question of whether there is an abstract definition of generalised qplexes independent of the specific measurement construction. The graph-theoretic treatment of MUB-like $$\Delta_{d^2}=\left\{p\in \mathbb{R}^{d^2}: p(i)\ge 0,\ \sum_i p(i)=1\right\}.$$6-design POVMs points toward one possible route, since affine constraints and outer polytopes can sometimes be written purely in combinatorial terms (Słomczyński et al., 2019).

The term itself also invites ambiguity. In quantum foundations, “qplex” refers to the QBist-inspired convex-geometric framework described above. It is unrelated to the uppercase acronym “QPLEX” used for duplex dueling multi-agent Q-learning in cooperative MARL (Wang et al., 2020), for a DOcplex-based software layer integrating quantum optimization backends (Giraldo et al., 2023), for queueing-control QPLEX Decision Processes (Dieker et al., 16 May 2026), or to “quasi-theories” in equivariant cohomology (Huan, 2018). The shared orthography is terminological rather than conceptual.

Within quantum foundations, the most persistent misconception is the assumption that qplex theory already is quantum theory. It is not. Qplex geometry explains a substantial part of finite-dimensional quantum structure and, in restricted Bell scenarios, even reproduces a major quantum bound. But general qplexes form a strictly larger family, and recent correlation results show that this difference has operational consequences. Qplex theories are therefore best understood as a reconstruction program whose achievements and limitations are now both sharply delineated (Gupta et al., 5 Jun 2026).