🔢 mathematics

How characterisations of traces and central elements fail: sharp violation constants via michelson contrast and noncentrality

This paper quantifies the sharp violation constants for various characterisations of traces and central elements in matrix algebras and finite von Neumann algebras, demonstrating that these constants are governed by Michelson contrast and noncentrality parameters respectively, with exact values derived for numerous inequalities including Gardner's inequality and the Novikov-Tikhonov characterisations.

Original authors: Sami Abdullah Abed, Irina Nikolaeva, Andrej Novikov

Published 2026-10-27

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Sami Abdullah Abed, Irina Nikolaeva, Andrej Novikov

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the vast landscape of mathematics, there is a special class of rules called inequalities. These are not the rigid equations that demand two sides be exactly equal, but rather flexible boundaries that say one quantity must be less than or greater than another. For decades, mathematicians have known that if a certain mathematical function behaves like a "trace"—a specific kind of averaging process used to summarize the properties of complex systems—then it must obey a long list of these inequalities. If the function is a trace, the rules hold perfectly. But what happens when the function is not a trace? For a long time, the answer was simply "no." The rules would break, but the extent of the breakage was a mystery. Scientists knew the rules failed, but they did not know how badly they failed, nor did they understand if the size of the failure told them anything about how far the function was from being a trace.

This question matters because these functions are the tools physicists and engineers use to describe everything from the energy of atoms to the flow of information in quantum computers. When a system is perfectly balanced, it follows the trace rules. When it is out of balance, it does not. The new research by Sami Abdullah Abed, Irina Nikolaeva, and Andrej Novikov sets out to measure exactly how much the rules break when the balance is lost. They treat the failure not as a binary error, but as a measurable quantity. They ask: if we know exactly how "unbalanced" a system is, can we predict exactly how much the mathematical rules will stretch or snap?

The researchers focused on specific measures of unbalance called "contrasts." Imagine a system where some parts are very heavy and others are very light. The contrast is a way of quantifying how extreme that difference is. If everything is the same weight, the contrast is zero, and the system is perfectly balanced. If one part is infinitely heavier than another, the contrast is at its maximum. The team wanted to see if these numbers could predict the size of the violation for thirty-eight different mathematical rules. Crucially, they found that for many rules, the failure depends on two distinct contrasts: the contrast of the system itself (the density) and the contrast of the specific operators being tested. For instance, for one of the most famous rules, the size of the failure is determined entirely by the product of these two contrasts. They tested this on matrices, which are grids of numbers used to represent these systems, and on more abstract structures called von Neumann algebras, which generalize the idea of matrices to infinite dimensions.

What they found was a story of surprising precision and unexpected limits. For many of the rules, the answer was yes: the size of the failure is determined entirely by the contrast (or the product of contrasts). They discovered a precise formula that links the two. For example, for one of the most famous rules, the violation grows in a specific, predictable way as the contrast increases. The researchers calculated the exact maximum amount the rule could be broken for any given level of unbalance. In some cases, the failure is tiny, growing very slowly as the system becomes unbalanced. In others, the failure is dramatic, exploding rapidly even with a small amount of unbalance. They found that for some rules, the failure is so severe that it is impossible to bound it with a simple number unless the system is perfectly balanced.

However, the story is not the same for every rule. The researchers discovered that for some inequalities, the contrast alone is not enough to tell the whole story. In these cases, the failure depends on a different, more subtle property called "noncentrality." This property measures how far the system is from its center of symmetry, rather than just how extreme its edges are. The team proved that for certain complex systems, you cannot predict the failure just by looking at the contrast; you must also look at this deeper structural distance. They showed that in some scenarios, a system could have a high contrast but still obey the rules perfectly, while another system with the same contrast could break them badly, simply because of how its internal parts were arranged.

The paper also tackled the question of how these failures behave when the system gets larger. They found that for some rules, the failure vanishes quickly as the system approaches a balanced state, while for others, it lingers. They identified specific thresholds: points where the contrast is so low that the rules can never be broken, no matter how the system is arranged. For instance, they proved that if the contrast of the operators involved is below a certain value, a specific inequality regarding the square root of a sum can never be violated. This provides a safety zone for engineers and physicists, telling them exactly when they can rely on these rules without fear.

To reach these conclusions, the team used a combination of rigorous mathematical proof and careful numerical simulation. For some of the more complex rules, they could not find a simple formula, so they ran thousands of computer experiments to map out the boundaries of the failure. These simulations acted as a rigorous lower bound, showing that the failure must be at least a certain size, even if the exact maximum remained a conjecture. They found that for many rules, the worst-case scenarios happen in very specific, small configurations, often involving just two dimensions, which allowed them to generalize the results to much larger systems.

The work reveals a deep hierarchy among these mathematical rules. While they are all equivalent when the system is perfectly balanced, they react very differently when the balance is disturbed. Some are fragile, breaking at the slightest hint of unbalance, while others are robust, holding firm until the system is wildly out of equilibrium. The researchers mapped out these sensitivities, showing that two rules that seem similar can fail at completely different rates. This distinction is crucial for understanding the limits of mathematical models in physics and engineering.

Ultimately, the paper transforms a vague understanding of mathematical failure into a precise science. It moves beyond asking "does it break?" to asking "how much does it break, and why?" By linking the size of the violation to specific, measurable properties of the system, the researchers have provided a toolkit for predicting the behavior of complex, unbalanced systems. They have shown that while the rules of the trace are strict, the ways in which they fail are governed by their own elegant and predictable laws.

Technical Summary: Violation Constants, Contrast, and Noncentrality in Trace Characterisations

Problem Statement
The paper addresses the quantitative failure of inequalities that characterise traces and central elements in $C^*$ -algebras and von Neumann algebras. While qualitative characterisation theorems (e.g., Gardner's inequality $|\phi(X)| \le \phi(|X|)$ ) establish that a positive linear functional $\phi$ is a trace if and only if specific inequalities hold, they do not quantify how strongly these inequalities fail when $\phi$ is not a trace. The authors ask: given a functional $\phi$ that is not a trace, what is the sharp constant $c(\phi)$ such that the inequality holds with a factor $c(\phi) \ge 1$ ? Furthermore, they investigate whether this violation constant depends solely on a measure of how far $\phi$ is from being a trace, and if so, what that measure is.

Methodology
The authors employ a combination of operator theory, convex analysis, and numerical optimisation to determine these sharp constants.

Parameters of Deviation:
- Contrast ( $\delta$ ): For matrix algebras $M_n(\mathbb{C})$ , the deviation of a density $\rho$ from a scalar is measured by the Michelson contrast $\delta = \text{OMC}(\rho) = \frac{\lambda_{\max} - \lambda_{\min}}{\lambda_{\max} + \lambda_{\min}}$ .
- Noncentrality ( $\kappa_Z$ ): For general finite von Neumann algebras, the contrast is insufficient because a central density (even with high contrast) yields a trace. The governing parameter is the noncentrality, defined via the central envelopes $\lceil \rho \rceil$ and $\lfloor \rho \rfloor$ (the least central element $\ge \rho$ and greatest central element $\le \rho$ ). The central condition number is $\kappa_Z(\rho) = \|\lceil \rho \rceil \lfloor \rho \rfloor^{-1}\|$ . The logarithmic noncentrality is $LNC(\rho) = \frac{1}{2} \ln \kappa_Z(\rho)$ .
- Operator Contrast: The authors also prescribe the contrast $\delta'$ of the operator variables (e.g., $|X|$ ) to study how the violation depends on both the functional and the operators.
Transfer Principle:
A key methodological tool is a "transfer principle" (Theorem 4.15). It states that if an inequality is compatible with central rescaling and an upper bound is proved in terms of the condition number in an arbitrary finite von Neumann algebra, and a matching example exists in $M_2(\mathbb{C})$ , then the sharp violation constant in every finite von Neumann algebra is the same function of $\kappa_Z(\rho)$ . This allows the authors to lift results from $2 \times 2$ matrices to general algebras.
Geometric and Spectral Analysis:
The proofs utilise operator angles (Krein and Gustafson), the Kantorovich constant $G = \sec \psi$ , and the Specht ratio. The authors reduce many problems to two-dimensional blocks where the operators and densities act as $1 + \delta S$ and $1 + \delta' S'$ , reducing the violation to a geometric problem involving the secant of an angle determined by the product of contrasts.
Numerical Verification:
For cases where closed-form solutions are elusive, the authors perform constrained optimisation on $M_2, M_3,$ and $M_4$ using 40-digit arithmetic to provide rigorous lower bounds for the suprema, supporting conjectured values.

Key Contributions and Results

Sharp Constants for 38 Characterisations: The paper determines the exact violation constants for 38 different characterisations of traces and central elements.
- Exact Results: For 18 cases (including Gardner's inequality, subadditivity of the modulus, and various projection inequalities), the constant is an explicit function of $\kappa_Z(\rho)$ valid in every finite von Neumann algebra. For example, for Gardner's inequality, the constant is $G_Z = \cosh(LNC(\rho)) = (1 - \delta_Z^2)^{-1/2}$ .
- Matrix-Specific Results: For 5 cases, exact envelopes are found only for matrix algebras (e.g., certain monotonicity inequalities).
- Bounds: For 13 cases, proved upper and lower bounds are provided.
- Numerical/Conjectured: For 2 cases, only numerical values are available.
Dependence on Contrast Product: For Gardner's inequality, the authors prove that the sharp constant depends on the density contrast $\delta$ and the operator contrast $\delta'$ only through their product $\delta \delta'$ . The constant is $(1 - \delta^2 \delta'^2)^{-1/2}$ .
Orders of Violation: The paper reveals that different characterisations fail at different rates as the functional approaches a trace (i.e., as $\delta \to 0$ ).
- Some fail linearly ( $O(\delta)$ ), e.g., the triangle inequality for projections.
- Some fail quadratically ( $O(\delta^2)$ ), e.g., Gardner's inequality.
- Some fail quartically ( $O(\delta^4)$ ), e.g., the monotonicity of $A \mapsto \phi(A^2)$ .
- One case (weighted monotonicity) is shown to have a lower bound of order $O(\delta^8)$ .
Central Elements: The paper solves the dual problem: characterising central elements. For a positive invertible element $a$ , the best constant in $p a p \le \lambda a$ (for projections $p$ ) is $\cosh^2(LNC(a))$ , and for $|\phi(a)| \le C |\phi|(a)$ , it is $\cosh(LNC(a))$ .
Thresholds: The authors establish sharp thresholds for operator contrasts below which certain inequalities cannot fail, regardless of the functional. For instance, the subadditivity of the square root $\phi((A+B)^{1/2}) \le \phi(A^{1/2}) + \phi(B^{1/2})$ cannot fail if the operator contrasts satisfy a specific condition related to $2\sqrt{2}/3$ .

Significance and Claims
The paper claims to provide the first quantitative versions of these qualitative characterisation theorems. It identifies noncentrality (specifically the central condition number $\kappa_Z$ ) as the correct governing parameter for violation constants in general von Neumann algebras, replacing the simple Michelson contrast which is only valid in factors.

The authors emphasise that the "violation envelope" is not a universal function of the contrast; in non-factor algebras, the constant depends on the noncentrality, and in direct sums, it depends on the maximum contrast within the blocks. The work unifies disparate inequalities (Gardner, Young, Peierls–Bogoliubov, etc.) under a common framework of "violation constants," showing that while equivalent for traces, they diverge significantly in their sensitivity to non-traciality. The paper does not claim to solve open problems regarding the exact constants for all cases (noting that some remain numerical or conjectured) but provides a rigorous catalogue of exact results and sharp bounds for the majority of known characterisations.

More like this