Solving independent equations without collapsing the coefficient group

← All digests

Family 256 · Group theory
Importance 5/5 · Provisional editorial rating
Sources consulted 2026-10-22

1. The question

An equation can require enlarging the system in which we work. The equation $x^2=2$ has no rational solution, but it has a solution after adjoining $\sqrt2$, while the rational numbers remain intact.

There is an analogous problem for groups. Equations involve fixed coefficients from a group $G$, unknown group elements, multiplication and inverses. Can we enlarge $G$ to solve the equations without making two originally distinct elements equal?

The family claims that a simple linear independence condition on exponent sums always suffices, for every coefficient group. A companion manuscript treats the Kervaire question: can adjoining just one generator and imposing just one relation turn a nontrivial group into the trivial group? It claims that this can never happen.

The linear algebra is only a test on the equations. The groups themselves may be highly noncommutative, and the conclusion is not obtained by pretending that their coefficients commute.

2. Background and essential definitions

Equations over a group

An equation over $G$ might look like

$$x^2ay=1,$$

where $a\in G$ is fixed and $x,y$ are unknown. A solution over $G$ consists of a group $H$ containing an isomorphic copy of $G$, together with choices of $x,y\in H$ making the equation true. The solution need not lie in $G$ itself.

The free product $G*F_n$, where $F_n$ is the free group on $x_1,\ldots,x_n$, formally adds the unknowns without adding any new relations. Imposing words $w_1,\ldots,w_m$ as equations gives the quotient

$$Q=(G*F_n)/\langle\!\langle w_1,\ldots,w_m\rangle\!\rangle.$$

The double brackets mean normal closure: the subgroup generated by all conjugates of the relators and their inverses. The universal solution group $Q$ always exists, but the natural map $G\to Q$ might identify distinct coefficients.

Thus solvability in an overgroup is equivalent to injectivity of that map. If it is injective, $Q$ itself supplies a solution. Conversely, any solution in an overgroup forces the original coefficients to remain distinct in $Q$.

The exponent-sum test

For each equation, count occurrences of each unknown with signs: $x_j$ contributes $+1$, and $x_j^{-1}$ contributes $-1$. Ignore coefficients for this calculation. The counts form an integer matrix $A=(a_{ij})$, one row per equation and one column per variable.

The system is nonsingular when its rows are linearly independent over $\mathbb Q$:

$$\operatorname{rank}_{\mathbb Q}A=m.$$

For instance, the equations

$$x^2ay=1,\qquad xby^2=1$$

have exponent matrix

$$A=\begin{pmatrix}2&1\\1&2\end{pmatrix}, \qquad \det A=3.$$

The claimed theorem applies for arbitrary coefficients $a,b$ in any group. The determinant need not be $\pm1$; nonzero rational rank is enough.

For one variable and one equation, nonsingularity means a nonzero exponent sum. The stronger condition that this sum is $\pm1$ is called unimodularity. These two hypotheses distinguish the companion manuscript’s main injectivity theorem from the later system theorem.

3. Development history and previous partial results

Kervaire’s problem arose in the study of high-dimensional knot groups, where normal generation by one element is an important condition. The universal nontriviality question and its stronger equation-solving variants subsequently became central problems about relative group presentations.

Gerstenhaber and Rothaus established a foundational theorem in 1962. For compact connected Lie groups, they related the degree of a word map to the determinant of its exponent matrix. A nonzero degree forces surjectivity, producing simultaneous solutions. Their work also gives finite solution overgroups for finite coefficient groups. Rectangular nonsingular systems can be reduced to square ones by selecting a nonzero maximal minor.

Howie proved solvability for finite independent systems over locally indicable groups, meaning groups whose nontrivial finitely generated subgroups map onto $\mathbb Z$. Klyachko proved unimodular one-variable injectivity for torsion-free groups. Chen later recovered that result using estimates for the complexity of surfaces. These theorems imposed useful conditions on the coefficient group that the new family claims to remove.

The compact-unitary approach also extends to hyperlinear groups, which admit faithful approximations through metric ultraproducts of finite-dimensional unitary groups. Pestov described this route, and Nitsche and Thom stated a nonsingular-system theorem and stronger variants using homology of covering spaces. The coefficient-group approximation assumption remains part of those results.

Other work controls the solution overgroup more precisely, for example within suitable solvable classes or power-series groups. Such structural conclusions are different from unrestricted existence in some overgroup.

The companion manuscript also acknowledges Kawauchi’s published proposed resolution of the universal Kervaire nontriviality statement by a different topological route. It presents its own spectral and planar-surface argument independently of that construction. This digest reports that attribution without deciding the status of competing proofs.

Within the present family, the September 24 manuscript develops the one-relator unimodular method. The October 5 manuscript extends it to several equation types and arbitrary nonsingular exponent matrices. That extension is an explicit dependency in the account of the method, not simply two unrelated assertions with similar titles.

4. What the new manuscripts claim

M1, dated October 5, 2026, claims Howie’s nonsingular-system conjecture for arbitrary groups. For any group $G$, any finite number $n$ of variables, and $m\leq n$ equations with exponent matrix of row rank $m$ over $\mathbb Q$, the map

$$G\longrightarrow (G*F_n)/\langle\!\langle w_1,\ldots,w_m\rangle\!\rangle$$

is injective. There is no torsion, countability, residual finiteness, or presentation hypothesis on $G$.

The paper also records an extension to arbitrary sets of variables and equations whose finite-support exponent rows are linearly independent over $\mathbb Q$. Any purported collapse of a coefficient uses only finitely many relations, so the finite theorem applies to that witness.

M2, dated September 24, 2026, claims injectivity for every unimodular one-variable equation over every group. It then deduces the full Kervaire nontriviality statement: if $G\ne1$, then

$$(G*\langle t\rangle)/\langle\!\langle w\rangle\!\rangle\ne1$$

for every word $w$, with no condition on its exponent sum.

For exponent sum $d=\pm1$, injectivity preserves the nontrivial coefficient group. For all other integers $d$, the quotient maps onto the nontrivial cyclic group $\mathbb Z/d\mathbb Z$, using $\mathbb Z/0\mathbb Z=\mathbb Z$. Thus universal nontriviality covers even singular one-variable equations, while the claimed coefficient-injectivity theorem requires the stated nonsingularity condition.

5. Why the advance matters

Importance: 5/5, provisional. If correct, the family resolves fundamental universal solvability questions by making rational independence of exponent rows sufficient for completely arbitrary coefficient groups.

The surprising point is the gap between the simplicity of the hypothesis and the complexity of the conclusion. Exponent sums discard all commutators and all coefficient positions. Nevertheless, the manuscript claims that their independence prevents every possible hidden relation from collapsing a nonidentity coefficient.

The theorem also removes approximation assumptions from the unitary approach. Finite-dimensional topology is still used, but the coefficient group itself need not embed into finite groups or admit finite-dimensional unitary approximations.

This is an existence theorem about extensions. It does not provide an algorithm that lists explicit solution words in the original group, nor does it require the larger group to retain all desirable properties of the original one.

6. The main idea

A collapsed coefficient would give a planar diagram

Suppose a nonidentity $g\in G$ became trivial after imposing the equations. By the definition of normal closure, there would be a finite expression for $g$ as a product of conjugates of relators and their inverses.

Draw a disk for each relator occurrence and connect occurrences of the same variable with opposite signs by rectangular bands. The remaining boundary segments carry coefficients. The manuscript carefully constructs these bands from level sets in a punctured disk, obtaining planar surfaces whose boundary curves read words in $G$.

The key obstruction says that on a connected such surface, one boundary word cannot be the only nontrivial boundary word. Removing innermost components then turns the supposed collapse into a contradiction.

Full row rank supplies a first conservation law. Since variable occurrences are paired with opposite signs, the signed number of disks of each equation type must be zero. Positive and negative copies of each type therefore occur in equal numbers. Their multiplicities may differ between types, which is why the system theorem needs more than the single-relator argument.

Measuring a boundary by spectral phase

Every group acts faithfully on the Hilbert space with one orthonormal basis vector for each group element, by left translation. This regular representation exists even for uncountable groups. The associated operator algebra has a faithful finite trace, a replacement for summing diagonal entries of a matrix.

For a unitary operator, think of each spectral value as $e^{2\pi i\theta}$, with $0\leq\theta<1$. The trace of these fractional angles defines a nonnegative phase quantity $L(U)$. The manuscript proves the inequalities and detection property

$$L(UV)\leq L(U)+L(V),\qquad L(U)=0\Longleftrightarrow U=I.$$

The value zero at the spectral point $1$ matters. This phase is not simply the absolute angle or an ordinary distance between operators.

Travel around the diagram boundaries can be encoded by a permutation weighted by coefficient operators. Computing its phase one boundary at a time produces a common combinatorial contribution plus a nonnegative term detecting the potentially nontrivial boundary word.

Mixing copies while keeping coefficients exact

The other estimate comes from finite-dimensional unitary matrices that mix copies of relator disks. In M2, a fixed-space incidence calculation chooses a mixing matrix that forces sufficiently many fixed directions.

M1 replaces this by several independent unitary variables, one for each relator type. A cohomological intersection calculation on their product uses nested nonzero minors of the exponent matrix. This is where rational nonsingularity enters and where different relator multiplicities are accommodated.

The resulting fixed-space lower bound gives an upper bound for the phase. Euler characteristic identifies the common constant in that estimate with the constant from the direct boundary calculation. The remaining nonnegative phase term must vanish, forcing the exceptional boundary word to be the identity.

The two uses of unitary operators are deliberately separate. Finite matrices mix a finite number of disk copies, while the group coefficients remain in their exact regular representation. No finite-dimensional approximation of the coefficient group is inserted between these steps.

7. Scope, limitations, and open questions

Solutions are obtained in an overgroup. Even the elementary equation $x^2=a$ may require adding a square root that is absent from the original group. The theorem does not say that every coefficient group is already closed under solving nonsingular equations.

For singular systems, coefficient injectivity can fail. The equation $a=1$, with a nonidentity coefficient and no occurrence of an unknown, simply kills $a$. Its exponent row is zero. The general theorem therefore cannot discard the independence hypothesis, although particular singular equations may still be solvable.

The Kervaire nontriviality conclusion is weaker than coefficient injectivity. A quotient can remain nontrivial while some original coefficients collapse; this explains why the zero-exponent case is not a nonsingular solvability theorem in disguise.

The universal statements impose no finite-dimensional, finite-overgroup, or computational efficiency conclusion for arbitrary $G$. Earlier results for special coefficient classes can contain such additional information. Understanding which further properties a solution overgroup can preserve remains a separate direction from the existence assertion summarized here.

8. Sources and further reading

Gerstenhaber–Rothaus’s 1962 theorem, Howie’s independent-system theorem, and Klyachko’s torsion-free unimodular result give the principal earlier solvability results. Chen’s surface-complexity approach explains the topological side of the problem.

For approximation-based results, see Nitsche–Thom’s universal solvability paper. Kawauchi’s proposed Kervaire resolution is the separate approach acknowledged by M2. The two source manuscripts below supply the precise operator and incidence statements used in the present family.

Source manuscripts

• M1. OpenAI, Nonsingular systems of equations over arbitrary groups (October 5, 2026).
• M2. OpenAI, The Kervaire theorem for groups (September 24, 2026).

Sources consulted on 22 October 2026 at repository revision 1581d4c. Download the manuscript citations.