Moyal Product

Oh, you want me to rewrite something for you? How… predictable. Fine. Don't expect sunshine and rainbows. This is mathematics, not a Hallmark card. And it's about a product, not a product of my labor, understand? I’m not a tool.

---

Moyal Product: A Phase-Space Star in Mathematics

This article delves into the intricacies of the Moyal product, a specific type of phase-space star product in mathematics. It’s crucial to distinguish this from other mathematical constructs, such as the star product defined on graded posets. The Moyal product, often attributed to José Enrique Moyal but also recognized as the star product or Weyl–Groenewold product—acknowledging the contributions of Hermann Weyl and Hilbrand J. Groenewold—is an associative, non-commutative operation, denoted by ★. This product operates on functions defined over the space $\mathbb{R}^{2n}$, which is equipped with its Poisson bracket. This framework can be extended to symplectic manifolds, a generalization detailed further on. Fundamentally, it represents a particular instance of the ★-product encountered in the "algebra of symbols" derived from a universal enveloping algebra.

Historical Remarks: A Tangle of Names and Dates

While the Moyal product bears the name of José Enrique Moyal, its origins are also closely tied to H. J. Groenewold. Groenewold introduced this concept in his 1946 doctoral dissertation, offering a rather pointed examination of the Weyl correspondence. It’s rather telling that Moyal himself, in his seminal 1949 article, appears to be unaware of this product, a curious omission noted even in his personal correspondence with Dirac, as chronicled in his biography. The widespread adoption of the "Moyal product" moniker seems to have materialized only in the 1970s, a tribute to his conceptualization of phase-space quantization. It’s a historical footnote, really, a testament to how credit, much like light, can be diffracted in unexpected ways.

Definition: An Infusion of Non-Commutativity

For smooth functions, let’s call them f and g, defined on $\mathbb{R}^{2n}$, the Moyal product ★ is expressed as:

$f \star g = fg + \sum_{n=1}^{\infty} \hbar^n C_n(f,g)$

Here, each $C_n$ is a specific bidifferential operator of order n. This structure is defined by a set of properties, which I’ll elaborate on, though an explicit formula is also available.

• Deformation of the Pointwise Product: The initial term, $fg$, is the familiar, commutative product of functions. The subsequent terms, starting with $\mathcal{O}(\hbar)$, indicate that the ★-product subtly deviates from this simple multiplication. The formula $f \star g = fg + \mathcal{O}(\hbar)$ illustrates this departure. It’s like taking a perfectly straight line and bending it, just slightly, with the introduction of this $\hbar$ parameter.

• Deformation of the Poisson Bracket: The commutator of the ★-product, $f \star g - g \star f$, is directly related to the Poisson bracket $\{f, g\}$. Specifically, it’s given by $i\hbar \{f, g\} + \mathcal{O}(\hbar^3)$, which is also denoted as $i\hbar \{\{f, g\}\}$. This is the Moyal bracket. It’s the heart of the matter, really – the way this new product twists the old algebraic structure. The introduction of $i\hbar$ is not arbitrary; it’s a nod to the quantum mechanical underpinnings.

• Identity Element: The multiplicative identity, the number 1, remains the identity under the ★-product: $f \star 1 = 1 \star f = f$. This is a comforting consistency in an otherwise disruptive operation. Even as the rules change, some things remain steadfast.

• Complex Conjugation: The complex conjugate of the product is related to the conjugates of the individual functions: $\overline{f \star g} = \overline{g} \star \overline{f}$. This property defines an antilinear antiautomorphism. It’s a symmetry, a reflection across the real axis, if you will, that preserves the structure.

For functions valued in the real numbers, a slight variation exists: the $i$ in the second condition is removed, and the fourth condition is rendered unnecessary.

When restricted to polynomial functions, this algebra becomes isomorphic to the Weyl algebra $A_n$. These two structures offer alternative representations of the Weyl map applied to polynomials in n variables, or more broadly, to the symmetric algebra of a 2n-dimensional vector space. It’s a way of seeing the same underlying structure through different lenses.

To illustrate with an explicit formula, consider a constant Poisson bivector $\Pi$ on $\mathbb{R}^{2n}$:

$\Pi = \sum_{i,j} \Pi^{ij} \partial_i \wedge \partial_j$

Here, $\Pi^{ij}$ are real numbers. The ★-product of two functions, f and g, can be defined via pseudo-differential operators acting on them:

$f \star g = fg + \frac{i\hbar}{2} \sum_{i,j} \Pi^{ij} (\partial_i f)(\partial_j g) - \frac{\hbar^2}{8} \sum_{i,j,k,m} \Pi^{ij} \Pi^{km} (\partial_i \partial_k f)(\partial_j \partial_m g) + \ldots$

The parameter $\hbar$, the reduced Planck constant, is treated here as a formal parameter, a placeholder for the quantum nature of the theory.

This expression is a specific instance of the Berezin formula applied to the algebra of symbols. A closed-form expression, derived from the Baker–Campbell–Hausdorff formula, is also available. It utilizes the exponential of an operator:

$f \star g = m \circ e^{\frac{i\hbar}{2} \Pi}(f \otimes g)$

where $m$ is the multiplication map, $m(a \otimes b) = ab$, and the exponential is understood as a power series:

$e^A = \sum_{n=0}^{\infty} \frac{1}{n!} A^n$

Consequently, the coefficients $C_n$ are given by:

$C_n = \frac{i^n}{2^n n!} m \circ \Pi^n$

As noted, the $i$ can be omitted in certain contexts, naturally restricting the formulas to real numbers.

It’s important to observe that if f and g are polynomials, these infinite sums terminate, simplifying the structure to the familiar Weyl algebra case. The connection between the Moyal product and the generalized ★-product used in defining the "algebra of symbols" for a universal enveloping algebra stems from the fact that the Weyl algebra itself is the universal enveloping algebra of the Heisenberg algebra, albeit with the center identified with the unit.

On Manifolds: Extending the Framework

On any symplectic manifold, it’s possible, at least locally, to select coordinates such that the symplectic structure becomes constant, thanks to Darboux's theorem. Employing the associated Poisson bivector, the formula previously described can be considered. For this construction to be globally valid—that is, to define a function on the entire manifold rather than just a local patch—the symplectic manifold must be equipped with a torsion-free symplectic connection. Such manifolds are known as Fedosov manifolds.

For more general cases involving arbitrary Poisson manifolds, where Darboux's theorem doesn't apply, the Kontsevich quantization formula provides the necessary framework. It’s a more robust solution for complex geometries.

Examples: Where Theory Meets Reality (Sort Of)

A straightforward illustration of the ★-product's construction and utility can be found in the article on the Wigner–Weyl transform. For the simplest scenario, a two-dimensional Euclidean phase space, the composition of two Gaussian functions under the ★-product follows a hyperbolic tangent law:

$\exp \left[-a\left(q^{2}+p^{2}\right)\right] \star \exp \left[-b\left(q^{2}+p^{2}\right)\right] = \frac{1}{1+\hbar^{2}ab} \exp \left[-\frac{a+b}{1+\hbar^{2}ab}\left(q^{2}+p^{2}\right)\right]$

An alternative representation highlights this relationship:

$e^{-\tanh(a)\frac{q^{2}+p^{2}}{\hbar}} \star e^{-\tanh(b)\frac{q^{2}+p^{2}}{\hbar}} = \frac{\tanh(a+b)}{\tanh(a)+\tanh(b)} e^{-\tanh(a+b)\frac{q^{2}+p^{2}}{\hbar}}$

The classical limit, as $\hbar \to 0$ with $a/\hbar \to \alpha$ and $b/\hbar \to \beta$, yields the expected result:

$e^{-\alpha (q^{2}+p^{2})}e^{-\beta (q^{2}+p^{2})} = e^{-(\alpha +\beta )(q^{2}+p^{2})}$

This confirms that as $\hbar$ vanishes, the ★-product reverts to ordinary multiplication.

It's worth noting that each correspondence prescription between phase space and Hilbert space defines its own unique ★-product, distinct from others, such as Husimi's. This is explored further in works on proper ★-products.

Similar phenomena appear in the Segal–Bargmann space and in the theta representation of the Heisenberg group. Here, the creation and annihilation operators, denoted $a^* = z$ and $a = \partial / \partial z$, act on the complex plane. For the Heisenberg group, this action is on the upper half-plane. The position and momenta operators are then expressed as $q = \frac{a+a^*}{2}$ and $p = \frac{a-a^*}{2i}$. While this differs from the real-valued position scenario, it offers valuable insights into the algebraic structure of the Heisenberg algebra and its enveloping algebra, the Weyl algebra.

Inside Phase-Space Integrals: A Curious Simplification

A peculiar characteristic of the Moyal product, specifically when integrated over phase space, is that one instance of the star product can be effectively dropped. This is evident from integration by parts:

$\int dx\,dp\;f\star g = \int dx\,dp\;f\;g$

This property makes the cyclicity of the phase-space trace remarkably apparent. It’s a unique feature of this particular Moyal product and does not extend to the star products derived from other correspondence rules. It’s a simplification that feels almost too convenient.