Atiyah's Minkowski Space Conjecture Fails for Every $n\ge3$

Abstract: Atiyah's Minkowski-space version of the configuration-of-points construction assigns to an admissible marked configuration of $n$ worldlines a collection of $n$ binary forms of degree $n-1$, whose roots are the ordered retarded celestial directions. He conjectured that these forms are always linearly independent. We disprove this conjecture for every $n\ge3$. For $n=3$, an explicit planar one-parameter family yields a real coefficient determinant with exactly one simple zero in a specified interval. At this parameter, all six ordered celestial roots are distinct and the coefficient matrix has rank exactly two. A null-translation construction then multiplies the first three forms by a common factor and produces counterexamples for every $n>3$. Consequently, within the class of complete pairwise disjoint timelike affine lines, universal independence holds at $n=2$ and fails for every $n\ge3$; for each of the counterexamples, the normalized Atiyah--Sutcliffe determinant is defined and vanishes.

• The paper constructs an explicit three-worldline counterexample using complete, disjoint, future-directed timelike affine lines with strictly subluminal velocities, where retarded binary quadratics become linearly dependent.
• Exact interval estimates identify a unique, simple, transverse rank-two determinant crossing while all six celestial roots remain distinct, proving the failure results from retarded timing rather than collisions or degeneracies.
• A null-translation stabilization extends the counterexample to every particle number n≥3, establishing an exact dichotomy: independence holds for n=2 but fails for all larger n, while stronger distinct-root versions remain open.

Atiyah's Minkowski space conjecture asserts that the binary forms built from retarded celestial directions of $n$ nonintersecting worldlines are always linearly independent over $\mathbb{R}$. In "Atiyah's Minkowski Space Conjecture Fails for Every $n\ge 3$" (2608.16693), Ziran Liu constructs explicit counterexamples showing that this universal assertion is false, and moreover that the failure occurs already within the most regular subclass Atiyah singled out: complete, pairwise disjoint, future-directed timelike affine lines, i.e., inertial observers with strictly subluminal velocities. The paper establishes a sharp dichotomy — independence holds for $n=2$ and fails for every $n\ge 3$ — and shows that the normalized Atiyah–Sutcliffe determinant vanishes on every configuration constructed.

The conjecture and its context

The construction assigns to each label $i$ the binary form $$\beta_i \in W_n = \mathrm{Sym}^{n-1}((\mathbb{C}^2)^*)$$ whose roots are the ordered directions from the marked event $x_i$ to the other worldlines, obtained where the past light cone of $x_i$ meets $\xi_j$ (the retarded points). Since $\mathbb{R}$0, linear independence of the $\mathbb{R}$1 forms is a nondegeneracy condition turning the projective classes into an ordered projective frame; via unitary polar decomposition it yields a point of $\mathbb{R}$2, connecting to Berry–Robbins transported spin and Atiyah's equivariant flag-map problem.

The literature reviewed in the paper splits cleanly. The Euclidean branch has extensive positive results: elementary proofs for $\mathbb{R}$3 and collinear configurations [Atiyah], Eastwood–Norbury's proof for four points [(Mizushima et al., 2010)-style results cited as EastwoodNorbury], computer-assisted verification of Conjectures II and III for all four-point configurations by Bou Khuzam–Johnson, and Malkoun's Gram-matrix proof of four-point independence. The hyperbolic branch likewise has positive results for four-point configurations due to Malkoun. Crucially, these branches impose reciprocity relations on oppositely ordered roots that generic retarded Minkowski data lack. Atiyah's own cyclic construction produced dependent forms but required nonuniform velocities, and he explicitly asked whether uniform straight-line motion could yield dependence or whether independence might hold universally — a prove-or-disprove challenge he restated in his 2010 lectures as Conjecture 1.6.1. Liu's paper answers this negatively.

The three-worldline counterexample

The core example lives in the plane $\mathbb{R}$4, so it is automatically a counterexample in full Minkowski space. Three complete affine worldlines are given explicitly, with direction vectors $\mathbb{R}$5, $\mathbb{R}$6, and $\mathbb{R}$7; squared norms $\mathbb{R}$8, $\mathbb{R}$9, and $n\ge 3$0; and spatial speeds $n\ge 3$1, $n\ge 3$2, $n\ge 3$3 — all strictly subluminal. The only free parameter $n\ge 3$4 shifts the temporal placement of the third line, isolating retarded timing from collisions and velocity changes. For every $n\ge 3$5 in the interval $n\ge 3$6, the lines are verified pairwise disjoint and admissible, so all six ordered retarded intersections exist uniquely.

The algebraic engine is a six-bracket identity reducing linear dependence of the three binary quadratics to the vanishing of

$n\ge 3$7

where $n\ge 3$8 and $n\ge 3$9 are explicit nonzero real lifts of the six celestial roots in a globally defined planar Cayley coordinate (with an accompanying lemma confirming that dependence in this coordinate is equivalent to dependence in the standard stereographic coordinate). The lifts reduce to rational functions of the variable radicals $n=2$0 and fixed radicals $n=2$1.

The main theorem then rests on exact rational interval estimates, proved in an appendix:

$n=2$2

with $n=2$3. Hence there is a unique parameter $n=2$4 at which the coefficient determinant vanishes, simply. The same interval analysis establishes fifteen nonvanishing bracket inequalities, so all six roots remain pairwise distinct at the crossing; since no two quadratic rows are proportional, the singular coefficient matrix has rank exactly two, and the path meets the smooth rank-two stratum of the determinantal hypersurface transversely. The failure therefore cannot be attributed to a collision, lightlike limit, repeated root, or higher-order contact — it is a genuine rank defect produced solely by retarded timing inside the strictly timelike inertial region.

Stabilization to all particle numbers

A single counterexample at $n=2$5 refutes the universal conjecture but not the statement at each fixed larger $n=2$6, since adjoining worldlines changes both form count and degree. The paper supplies a separate geometric mechanism at every $n=2$7: a null-translation construction. One builds a Lorentzian two-plane $n=2$8 with $n=2$9 future timelike and $n\ge 3$0 past null, containing the three marked events but none of the three velocity vectors, and adjoins lines $n\ge 3$1 with distinct offsets $n\ge 3$2. For each old observer and new emitter, the retarded displacement is a positive multiple of the same past-null vector $n\ge 3$3, so every new emitter contributes one common factor $n\ge 3$4 to each of the first three forms:

$n\ge 3$5

Multiplying the original three-term relation by $n\ge 3$6 yields a nontrivial relation among all $n\ge 3$7 forms. The result is an admissible, pairwise disjoint, timelike $n\ge 3$8-worldline configuration with dependent forms for every $n\ge 3$9. A general stabilization principle is proved abstractly before being applied to the numerical core, so the mechanism is independent of the particular example.

Geometry of the rank defect and the normalized determinant

The crossing admits a clean projective description. The multiratio-type quantity $i$0 formed from the six brackets is invariant under independent lift rescaling and under $i$1; at $i$2 it equals $i$3. The relation space among the three quadratics is one-dimensional, and every nonzero relation has all coefficients nonzero. Via the Veronese embedding of $i$4, the rank-two condition says precisely that the three secant lines associated with the three observers are concurrent.

For each constructed configuration, the normalized determinant

$i$5

is well defined (reciprocal roots are distinct by a general lemma proving $i$6), independent of lift choices, projectively invariant, and equal to zero. Consequently the dichotomy within the class of complete pairwise disjoint timelike affine lines is exact: universal independence holds at $i$7 (the two linear factors have distinct roots) and fails for every $i$8.

Limitations and open questions

The paper is careful about scope. For $i$9, the stabilization intentionally introduces repeated common roots; while such coincidences are permitted by Atiyah's printed Conjecture 1.6.1, whether counterexamples exist at every $$\beta_i \in W_n = \mathrm{Sym}^{n-1}((\mathbb{C}^2)^*)$$0 with all ordered celestial roots pairwise distinct remains open. Likewise, the marked events in the construction cannot be made simultaneous in any inertial frame (their differences include timelike separations); imposing synchronized observation would define a different, stronger problem not addressed here. Finally, since the examples are neither static nor formed by inertial rays from a common event, the Euclidean and hyperbolic branches of the conjecture — where positive results abound — are untouched.

Conclusion

This paper resolves Atiyah's uniformly-moving Minkowski challenge negatively, by an explicit one-parameter family of three subluminally moving observers whose retarded binary quadratics become dependent at a unique, simple, transverse rank-two crossing, together with a null-translation mechanism propagating the failure to every particle number. Within the class of complete pairwise disjoint timelike affine lines, the resulting dichotomy — independence exactly at $$\beta_i \in W_n = \mathrm{Sym}^{n-1}((\mathbb{C}^2)^*)$$1, failure for every $$\beta_i \in W_n = \mathrm{Sym}^{n-1}((\mathbb{C}^2)^*)$$2 — is definitive. The remaining questions the paper leaves open concern stronger genericity hypotheses (all ordered roots distinct at arbitrary $$\beta_i \in W_n = \mathrm{Sym}^{n-1}((\mathbb{C}^2)^*)$$3) and synchronization constraints, both of which would constitute different problems under Atiyah's printed formulation.