symmetric monoidal (∞,1)-category of spectra
Laurent series generalize power series by allowing both positive and negative powers. In particular, Laurent series with complex coefficients generalize Taylor series of analytic functions to meromorphic functions. A Laurent series for a meromorphic function at finite has the form
where is merely constrained to be finite and is often negative.
Or, in some contexts one wants to take . Here such a formal sum (with powers extending infinitely in both directions) is suggestive notation for an element belonging to the dual of a ring (see Laurent polynomials below). In such contexts, Laurent series can be likened to distributions, i.e., functionals on the algebra of functions with compact support.
A Laurent series in one variable over a commutative unital ring is a doubly infinite series
where . Equivalently: a Laurent series is a function . The -module of Laurent series is denoted .
A Laurent polynomial is a Laurent series for which all but finitely many are zero. Laurent polynomials form a ring which may be described as or abstractly as . Observe that Laurent series in the generality discussed here do not analogously form a ring: the obvious definition of the coefficients of the product of two Laurent series ,
doesn’t make sense in general (although it could sometimes make sense in topological contexts where some such infinite sums can converge, as in the case ).
The -vector space of Laurent series does however form a module over the ring of Laurent polynomials, i.e., if and is a Laurent series, then the product as defined above always makes sense.
In general, questions of convergence are treated as separate issues. In complex analysis, the Laurent series describes a meromorphic function in a neighborhood around the point (possibly with a pole there) if all but finitely many negatively indexed terms are zero. Similarly, series of the form describe meromorphic functions in a neighborhood of with poles of order at most .
On the other hand, in algebra one often hears of the ring of formal Laurent series. Here, the presence of the word “ring” signifies that we are restricting the coefficients so that multiplication makes sense. Thus,
The ring of formal Laurent series over a commutative ring in an indeterminate consists of Laurent series , with but where all but finitely many for vanish.
Multiplication defined as above clearly makes sense. If is a field , then this ring is usually denoted and is in fact a field; indeed it is the field of fractions of the ring of formal power series, where is often viewed as a discrete valuation ring.
It would perhaps be clearer if we used the term “restricted Laurent series” to cover the Laurent series considered in Definition 2, and let “Laurent series” be the term that covers doubly infinite series.
Another point of view on Laurent series is given in the following alternative definition.
A Laurent series over is a -linear functional
on the algebra of Laurent polynomials.
The associated series is . But here we emphasize the point of view that Laurent series have a distribution-like character, with Laurent polynomials being considered a space of functions with compact (finite) support, via the evident inclusion
Of particular interest as a Laurent series is the formal Dirac distribution,
which intuitively is like the Fourier transform of the pointwise multiplicative identity given by for all . It has the following property:
Indeed, notice that is the distribution which evaluates a Laurent polynomial at the multiplicative identity .
More generally, for an invertible unit, the series satisfies
Just as it doesn’t make sense in general to multiply distributions (at least not without heavy qualifications, as in the theory of Colombeau), we cannot make sense of expressions like . However, one can meaningfully work with derivatives of distributions, so we have for example
which has the property
Indeed, there is a formal calculus of the Dirac distribution that plays an important role in the theory of vertex operator algebras. See for example the treatment in Frenkel, Lepowsky, Meurman.
One can also work with convolution products , defined by multiplying coefficients degree-wise:
For discussion of products of distributions, see