Abstract
The Weyl procedure associates a function of two ordinary variables, called the c-function or symbol, with an operator, called the Weyl operator of the symbol. One generally formulates this association by defining the operator corresponding to a given symbol. In this paper we consider the reverse problem: Given the Weyl operator, what is the matching symbol? We give a number of explicit formulas for obtaining the symbol that would generate an arbitrary Weyl operator, and we illustrate each form with an example.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Although there were antecedents in the operational calculus of Heaviside, the concept of associating ordinary functions with operators took on particular importance with the development of quantum mechanics. Since ordinary functions commute and operators generally do not, there is an infinite number of ways to associate a function of two ordinary variables with an operator. The earliest proposed rules were that of Born and Jordan [2] and of Weyl [14], and subsequently other rules have been proposed. Such rules are called rules of association, correspondence rules, ordering rules, among other names. The ordinary function is commonly called a c-function or symbol. The infinite number of rules may be characterized and generated in a simple manner [3, 6]. In this paper we restrict ourselves to the Weyl rule. Historically the issue was posed as to how to define the operator for a given symbol. We consider the inverse problem, namely how to obtain the c-function given the Weyl operator. We present a number of explicit formulas and give an example for each one.
Notation Except for the operator D, all operators will be denoted by boldface type. For a symbol we shall generally use g(x, k), and use \(\mathbf {G}(\mathbf {x}, D)\) for the corresponding operator. The Weyl operator is generally a function of the operators \(\mathbf {x}\) and D where,
The fundamental relation between \(\mathbf {x}\) and D is the commutator,
Depending on the field of study these operators may be appropriately called position and spatial frequency, or position and momentum, and in time-frequency analysis, they correspond to the time and frequency operators [4, 5]. To avoid confusion in some equations, instead of the pair \((\mathbf {x,}\, D)\) as defined above we often use the pair \((\mathbf y \mathbf {,}\, D_{y})\). All integrals range from \(-\infty \) to \(\infty \) unless otherwise indicated.
Throughout the paper all functions will be supposed to belong to suitably regular function spaces in order that all performed operations makes sense. We use the delta function and its properties freely.
2 The Weyl operator
There are two standard definitions of the Weyl procedure, the first, originally given by Weyl, and the second, by the action of the Weyl operator on a function [17].
Definition 1
For the symbol g(x, k), the corresponding operator, \(\mathbf {G}\left( \mathbf {x},D\right) \), is:
where \(\widehat{g}(\theta ,\tau )\) is the Fourier transform of g(x, k),
In going from Eqs. (2.1) to (2.2) we used the well known identity [12, 16]
Combining Eqs. (2.2) and (2.3), we also have
We call \(\mathbf {G}(\mathbf {x},D)\,\)the Weyl operator corresponding to the symbol g(x, k).
Definition 2
Alternatively, one can define the Weyl procedure by how the operator transforms a function, say \(\psi (x),\)
Hence the Weyl procedure is often called the Weyl transform because it transforms \(\psi (x)\) into the right hand side of Eq. (2.8) [17].
Equivalence of the two definitions For completeness we show the equivalence of the above two definitions. Using Eq. (2.6) and operating on a function \(\psi (x)\) we have
Using the fact that \(e^{i\tau D}\) is the translation operator
we have
Making the transformation
Equation (2.8) follows straight forwardly.
In this paper, we deal mostly with the first definition, but all the formulas we derive can be transcribed to the second.
3 Preliminaries
Central to our results is the ordinary function R(x, k) which is defined by
Using Eq. (2.1) we have
giving
Expressing \(R\mathbf {(}x,k\mathbf {)}\) in terms of the symbol directly, we have
Rearrangement form of an operator For an operator \(\mathbf {G}{(}\mathbf {x},D{),}\) one defines the function R(x, k) by the following procedure [6]:
To perform the rearrangement, one generally uses the commutation relation, Eq. (1.3), and variations of it. Conversely, if we have R(x, k), one gets the operator, \(\mathbf {G}{(}\mathbf {x},D{),}\) by:
Notice that we have used the same notation R(x, k) in both Eqs. (3.6) and (3.1) because indeed they are the same. To prove this, consider Eq. (2.2) which we repeat here for convenience
which indeed has the \(\mathbf {x}\) operators to the left of the D operators. Hence
which is the same as Eq. (3.4).
Example
Consider the operator
By making repeated use of the commutation relation, Eq. (1.3), we place all the D operators to the right of the \(\mathbf {x}\) operators,
and therefore
Alternatively, using Eq. (3.1) for \(R\mathbf {(}x,k\mathbf {)}\) we have
which is the same as Eq. (3.12).
We note that the symbol that gives the operator in Eq. (3.10) is
To show that, consider the Weyl correspondence for \(\left( x^{2} k^{2}+1/2\right) .\) First we calculate \(\widehat{g}(\theta ,\tau )\,\)
Therefore, the corresponding operator, according to Eq. (2.2) is
which evaluates to Eq. (3.10).
4 Inversion: from Weyl operator to symbol
The issue we address is finding the symbol g(x, k) for a given operator, \(\mathbf {G}\left( \mathbf {x},D\right) ,\) assuming that the operator was obtained by the Weyl procedure. We have found a variety of expressions. We list these results, and for each one we provide a proof, and an example. For the example, we use the one considered in Sect. 3 which we repeat here for convenience. For the symbol
the Weyl operator is
and the corresponding rearrangement operator is
4.1 Direct substitution method in G(x, D)
For an operator \(\mathbf {G(x,}\,D\mathbf {)}\) make the substitution
then the symbol is given by
where the right hand side is the operation on the number one as indicated.
Proof
In Eq. (2.2) make the substitution indicated by Eq. (4.4),
Now, operate on an arbitrary function, f(x, k),
and therefore
If we take
then Eq. (4.10) becomes identical to Eq. (4.10), and we have that
Also, we note, that the operators indicated by Eq. (4.4) have these same commutation relations as \(\mathbf {x}\) and D:
In this regard we mention that the Weyl operator may be obtained from g(x, k) by way of
Example
Consider the operator
Making the substitution indicated by Eq. (4.5) we have
4.2 Integral transformation of R(x, k)
For an operator \(\mathbf {G}(\mathbf {x},D)\) with a corresponding R(x, k) , as defined in Sect. 3, the symbol g(x, k) may obtained by way of
or, translating the integration variables, via
In addition,
Proof
Inverting the Fourier transform in Eq. (3.4) to find \({\widehat{g}(\theta ,\tau )}\) in terms of R, we have
and using Eq. (2.4) to obtain the symbol g from \(\widehat{g}\),
which simplifies to
which is (4.19).
To obtain Eq. (4.21), we make the substitution
which gives
\(\square \)
Example
Substituting R(x, k) from Eq. (4.3) into (4.20), we have
which straightforwardly evaluates to
4.3 Operator form on \(R\mathbf {(}x,k\mathbf {)}\)
Equation (4.19) can be put into an operator form giving
Proof
From Eq. (4.19) we have
which gives Eq. (4.29). \(\square \)
Example
Using Eq. (4.3) for R(x, k)
Expanding the exponential in a power series in its argument, only the first three terms contribute to the sum
giving
4.4 R(x, k) operating on \(e^{-2ixk}\)
In R(x, k) make the substitution
which defines the operator \({R\left( \frac{i}{2}\frac{\partial }{\partial k},\frac{i}{2}\frac{\partial }{\partial x}\right) }\). Then, the symbol g(x, k) is given by
Proof
In Eq. (3.4) for R(x, k), make the substitutions given by Eq. (4.36) to obtain that
which is g(x, k). \(\square \)
Example
For our example, where R(x, k) \(=x^{2}k^{2}-2ixk,\) the expression for the symbol g, as per Eq. (4.36), gives
4.5 Operating on a delta function
Since the delta function forms a complete set, one would expect that the operation of the operator on the delta function would allow one to obtain the symbol. Explicitly we show that
Proof
Consider
witch evaluates to g(x, k). \(\square \)
Example
Using our standard example
which evaluates to \(g(x,k)=\,x^{2}k^{2}+1/2.\)
4.6 Operating on the exponential
Similar to Eq. (4.45), we have
Proof
Substituting Eq. (2.2) for \(\mathbf {G} (\mathbf y ,D_{y})\) gives for the right hand side of Eq. (4.51),
which evaluates to g(x, k). \(\square \)
Example
For our standard example
we have
which simplifies to
4.7 Operating on the delta function directly
Another form involving the delta function is to define
then the symbol is found via
A convenient form to evaluate Eq. (4.60) is
Proof
Using Eq. (2.8), we have that
A few manipulations and an inverse Fourier transform leads to Eq. (4.60).
Example
For our usual example,
we have
Straightforward evaluation leads to \(g(x,k)=\,x^{2}k^{2}+1/2.\)
4.8 Trace of the Wigner and Weyl operators
The symbol may be obtained from
where \(\mathbf {W}\) and \(\mathbf {G}\) are the matrix elements of the Wigner operator and the Weyl operator respectively, terms that we now define. This method is essentially the one presented by Englert [8] and in reference [1]. A similar result was obtained by Duan and Wong [7]; see discussion after Eq. (4.87).
For any complete set \(u_{n}(y)\), we define, as is standard in quantum mechanics, the matrix elements, \(G_{nk}\) of an operator \(\mathbf {G} (\mathbf y ,D\mathbf {_{y}})\) by
We define the Wigner operator by
where the subscripts x and k are parameters. The reason this is called the Wigner operator is that its expectation value gives the Wigner distribution [1, 8, 15]. That is,
which is the Wigner distribution in x and k. For the Wigner operator, the matrix elements are
and are functions of the parameters x and k.
The nk matrix elements of the Weyl operator, \(\mathbf {G}(\mathbf y ,\,D\mathbf {_{y}}),\) given by Eq. (2.2), are
Trace The nm matrix elements of the product of two matrices is
and their trace is given by
Since the \(u_{n}\) form a complete set, we have that in general
Applying Eq. (4.85) to Eq. (4.84) we have
which simplifies to
and hence Eq. (4.70).
As mentioned above, a similar result was obtained by Duan and Wong [7] but with different approach and terminology.Footnote 1 It is of interest to relate their result to that of Eq. (4.70). Using our notation, they define integral representation of the Weyl transform by
which is what we have called the Weyl operator, \(\mathbf {G}\left( \mathbf {x},D\right) \). Also, they define the operator \(\rho ^*\) by what we have called the Wigner operator \(\mathbf {W}_{xk}(\mathbf {y,}{D}_{y}).\) Their Theorem 1.8 is that the symbol is the trace of \(\rho ^* W_{g}\) which corresponds to Eq. (4.70)
Continuous case If the complete set is continuous, \(u_{\alpha }(y),\) the above summations become integrations and we have that
and
and further
which is the continuous analog of Eq. (4.87).
Examples of complete sets Considering the continuous complete set given by
The Wigner operator matrix elements are
Evaluation leads to
For the matrix element of the Weyl operator we have
which evaluates to
Now consider the trace of the product of \({W_{\alpha ,\beta }(}\) x, k ) and \({G_{\alpha ,\beta }}\)
which evaluates to
Complete set: delta function
We consider the complete set
For the Wigner matrix elements, we have
which simplifies to
The \(\mathbf {G}\) matrix elements are,
which gives
Taking the trace, we have
which simplifies to
5 Conclusion
We have presented a number of different expressions for obtaining the symbol that generates a given Weyl operator. The Weyl procedure is not the only possible one and many others have been studied, including the Born and Jordan, standard, anti-standard symmetrization rule, among others [9, 11, 13]. A unified approach that generates all rules has been developed [3, 6, 10]. For a symbol g(x, k), the operator is given by
where \(\Phi (\theta ,\tau )\) is a two dimensional function, called the kernel. By choosing different kernels, different rules are obtained. In a future paper, we will deal with the inverse problem for the generalized correspondence rule indicated by Eq. (5.1).
Notes
The authors thank the referee for making us aware of the paper by Duan and Wong.
References
Ben-Benjamin, J.S., Kim, M.B., Schleich, W.P., Case, W.B., Cohen, L.: Working in phase-space with Wigner and Weyl. Fortschr. Phys. (2016). doi:10.1002/prop.201600092
Born, M., Jordan, P.: Zur Quantenmechanik. Zeit. f. Phys. 34, 858–888 (1925)
Cohen, L.: Generalized phase-space distribution functions. J. Math. Phys. 7, 781–786 (1966)
Cohen, L.: Time–frequency distributions—a review. Proc. IEEE. 77, 941–981 (1989)
Cohen, L.: Time–Frequency Analysis. Prentice-Hall, Englewood Cliffs (1995)
Cohen, L.: The Weyl Operator and its Generalization. Springer, Basel (2013)
Duan, X., Wong, M.W.: Pseudo-differential operators and Weyl transforms. Sci. Bull. Univ. Politehn. Buchar. Ser. A Appl. Math. Phys. 75, 3 (2013)
Englert, B.-G.: On the operator bases underlying Wigner’s, Kirkwood’s and Glauber’s phase space functions. J. Phys. A Math. Gen. 22, 625 (1989)
Kirkwood, J.G.: Quantum statistics of almost classical ensembles. Phys. Rev. 44, 31–37 (1933)
Lee, H.W.: Theory and application of quantum phase-space distribution functions. Phys. Rep. 259, 147–211 (1995)
Margenau, H., Hill, R.N.: Correlation between measurements in quantum theory. Prog. Theor. Phys. 26, 722–738 (1961)
Moyal, J.E.: Quantum mechanics as a statistical theory. Proc. Camb. Philos. Soc. 45, 99–124 (1949)
Shewell, J.R.: On the formation of quantum-mechanical operators. Am. J. Phys. 27, 16–21 (1959)
Weyl, H.: The Theory of Groups and Quantum Mechanics. E. P. Dutton and Co., Boston (1928)
Wigner, E.P.: On the quantum correction for thermodynamic equilibrium. Phys. Rev. 40, 749–759 (1932)
Wilcox, R.M.: Exponential operators and parameter differentiation in quantum physics. J. Math. Phys. 8, 962–982 (1967)
Wong, M.W.: Weyl Transforms. Springer, New York (1998)
Acknowledgements
This paper originated a number of years ago with discussions that one of the authors (LC) had with Prof. Wong on the problem of the Weyl operator and its inverse. L. Cohen would like to thank Prof. Wong for the many subsequent discussions. The authors would like to thank Profs. Schleich and Scully for the lively discussions we had at the Wyoming summer school. The work of M. Kim and J. Ben-Benjamin is supported by the Office of Naval Research (Award No. N00014-16-1-3054) and the Robert A. Welch Foundation (Grant No. A-1261).
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Kim, M., Ben-Benjamin, J.S. & Cohen, L. Inverse Weyl transform/operator. J. Pseudo-Differ. Oper. Appl. 8, 661–678 (2017). https://doi.org/10.1007/s11868-017-0225-9
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11868-017-0225-9