Optimal Polynomial Approximants (OPA)

• Optimal polynomial approximants (OPA) are finite-degree polynomials that minimize the norm of 1-pf, serving as precise projection mechanisms in analytic and Banach spaces.
• They use orthogonal projections in Hilbert spaces and nonlinear metric projections in H^p and ℓ_A^p spaces, linking cyclicity with invariant subspace theory.
• OPA theory underpins explicit coefficient computations, zero-set analysis, and applications in digital filter design and inverse filtering problems.

Optimal polynomial approximants (OPA) are finite-degree polynomial inverses defined by a best-approximation problem: for a nonzero analytic function $f$, one seeks a polynomial $p_n$ of degree at most $n$ that minimizes $\|1-pf\|$ in a prescribed function-space norm; more generally, one may minimize $\|g-pf\|$ for a fixed target $g$. In Hilbert spaces this is an orthogonal projection problem, while in $H^p$, $\ell_A^p$, and related Banach spaces it becomes a nonlinear metric projection problem. The subject lies at the intersection of approximation theory, cyclicity, invariant subspaces, orthogonal polynomials, reproducing kernels, and, in applied settings, least-squares inverse filter design (Bénéteau et al., 2021, Felder, 2020).

1. Definition and geometric formulation

In the classical Hilbert-space setting, especially weighted Hardy-type spaces $H^2_\omega$, an OPA is defined by

$p_n=\arg\min_{\deg p\le n}\|1-pf\|_{H^2_\omega}.$

If $p_n$0 and

$p_n$1

then $p_n$2 is the orthogonal projection of $p_n$3 onto the finite-dimensional subspace $p_n$4, where $p_n$5 denotes the polynomials of degree at most $p_n$6 (Bénéteau et al., 2019). This projection viewpoint is foundational: it makes existence and uniqueness immediate in Hilbert spaces and links the asymptotics of OPA to the closed shift-invariant subspace generated by $p_n$7 (Bénéteau et al., 2021).

A broader formulation replaces the target $p_n$8 by an arbitrary $p_n$9, defining the $n$0th OPA to $n$1 by

$n$2

In this form, OPA become a general projection framework on reproducing kernel Hilbert spaces with dense polynomials and bounded shift, and the special role of the normalized kernel at the origin $n$3 becomes explicit in spaces where $n$4 (Felder, 2020).

The Banach-space version is formally similar but geometrically different. In $n$5, $n$6, the OPA $n$7 minimizes $n$8, and uniqueness follows from uniform convexity rather than Hilbert orthogonality (Centner et al., 2023). The same metric-projection interpretation appears in $n$9, where

$\|1-pf\|$0

for $\|1-pf\|$1, and the approximant is the unique minimizer of $\|1-pf\|$2 (Cheng et al., 2021). This distinction between orthogonal and metric projection is one of the central structural divides in the subject.

2. Hilbert-space structure, orthogonal polynomials, and explicit computation

In Hilbert spaces, OPA admit several equivalent computational descriptions. Writing

$\|1-pf\|$3

the coefficients satisfy a Gram-system or normal-equation formulation. If

$\|1-pf\|$4

then $\|1-pf\|$5 in the classical $\|1-pf\|$6 problem (Bénéteau et al., 2021). In the generalized setting $\|1-pf\|$7, the coefficient vector solves

$\|1-pf\|$8

which is the basic finite-dimensional linear algebra model for OPA (Felder, 2020).

A second description uses orthogonal polynomials in the weighted inner product

$\|1-pf\|$9

If $\|g-pf\|$0 is an orthonormal basis for $\|g-pf\|$1, then

$\|g-pf\|$2

and in $\|g-pf\|$3 this identifies OPA with reproducing kernels and reversed orthogonal polynomials on the unit circle (Bénéteau et al., 2021). The formula

$\|g-pf\|$4

makes the zero set of OPA identical with the zero set of the associated reproducing kernels (Bénéteau et al., 2016).

For polynomial data, the computation can become especially explicit. If $\|g-pf\|$5 is a monic polynomial of degree $\|g-pf\|$6 with simple zeros $\|g-pf\|$7, then the residual $\|g-pf\|$8 can be written in terms of a fixed $\|g-pf\|$9 kernel Gram matrix

$g$0

and

$g$1

This reduces the problem to inversion of a matrix whose size depends on $g$2, not on $g$3, and yields explicit coefficient formulas and distance formulas such as

$g$4

(Bénéteau et al., 2019). For $g$5, the same program persists but the Gram matrix is replaced by a Hankel moment matrix, reflecting the appearance of kernel derivatives rather than kernel values (Bénéteau et al., 2019).

The relation with orthogonal polynomials extends beyond one variable. In reproducing kernel Hilbert spaces on the ball and bidisk, weighted orthogonal polynomial systems still control OPA, but the recovery of the full orthogonal family from the OPA sequence can fail because the approximants may probe only a thin monomial sector (Sargent et al., 2020). Explicit closed forms are nevertheless available in special multivariable models, such as $g$6 in a scale of spaces on the unit ball with kernel $g$7 (Sargent et al., 2020).

3. Banach-space theory in $g$8, $g$9, and $H^p$0

Outside the Hilbert setting, the OPA problem becomes genuinely nonlinear. In $H^p$1, $H^p$2, the minimizer exists and is unique because $H^p$3 is uniformly convex, but orthogonality must be replaced by Birkhoff–James orthogonality (Centner et al., 2023). James’s criterion gives

$H^p$4

and this criterion underlies the characterization of OPA in $H^p$5 and $H^p$6 alike (Cheng et al., 2023).

A central substitute for the Hilbert-space Pythagorean theorem is a family of $H^p$7-Pythagorean inequalities. The $H^p$8 theory uses these inequalities repeatedly to control OPA errors, coefficients, and roots when no linear projection formula is available (Centner et al., 2023). In this regime, even low-degree approximants reflect Banach-space geometry. The degree-zero approximant is a constant $H^p$9 minimizing $\ell_A^p$0, and the degree-one approximant satisfies nonlinear identities involving integral quantities such as

$\ell_A^p$1

with analogous formulas for $\ell_A^p$2 and $\ell_A^p$3, from which exact expressions for the root and leading coefficient can be derived (Centner et al., 2023).

The paper "More properties of optimal polynomial approximants in Hardy spaces" develops the asymptotic and continuity theory in this Banach setting. For fixed $\ell_A^p$4, the metric projections $\ell_A^p$5 converge in norm to the metric projection of $\ell_A^p$6 onto the invariant subspace $\ell_A^p$7, and the map $\ell_A^p$8 is continuous for fixed $\ell_A^p$9; for bounded $H^2_\omega$0, $H^2_\omega$1 uniformly on $H^2_\omega$2 when $H^2_\omega$3 (Cheng et al., 2023).

The $H^2_\omega$4 extension places OPA in a still wider framework. For $H^2_\omega$5, $H^2_\omega$6 is defined by

$H^2_\omega$7

Existence holds for all $H^2_\omega$8, uniqueness holds for $H^2_\omega$9, and uniqueness can fail for $p_n=\arg\min_{\deg p\le n}\|1-pf\|_{H^2_\omega}.$0 and $p_n=\arg\min_{\deg p\le n}\|1-pf\|_{H^2_\omega}.$1 (Centner, 2021). In $p_n=\arg\min_{\deg p\le n}\|1-pf\|_{H^2_\omega}.$2, one recovers the Hilbert-space orthogonality system and a first-degree zero-free criterion: $p_n=\arg\min_{\deg p\le n}\|1-pf\|_{H^2_\omega}.$3 For general $p_n=\arg\min_{\deg p\le n}\|1-pf\|_{H^2_\omega}.$4, the characterizing equations become

$p_n=\arg\min_{\deg p\le n}\|1-pf\|_{H^2_\omega}.$5

which is the $p_n=\arg\min_{\deg p\le n}\|1-pf\|_{H^2_\omega}.$6 analogue of the normal equations (Centner, 2021).

4. Zeros, extra zeros, and geometric constraints

The zero set of OPA is one of the most intensively studied aspects of the theory. In $p_n=\arg\min_{\deg p\le n}\|1-pf\|_{H^2_\omega}.$7, the standard picture is rigid: OPA zeros lie outside the closed unit disk, and this can be read either from orthogonal-polynomial theory or from reproducing-kernel representations (Bénéteau et al., 2021). More generally, in weighted Hilbert spaces $p_n=\arg\min_{\deg p\le n}\|1-pf\|_{H^2_\omega}.$8, the minimal possible modulus of an OPA zero is governed by the nonlinear extremal quantity

$p_n=\arg\min_{\deg p\le n}\|1-pf\|_{H^2_\omega}.$9

and a major result identifies $p_n$00 with half the norm of a Jacobi matrix $p_n$01: $p_n$02 Hence there exists an OPA zero in the open unit disk if and only if $p_n$03 (Bénéteau et al., 2016). In Dirichlet-type spaces $p_n$04, this yields a dichotomy: if $p_n$05, OPA zeros stay outside $p_n$06; if $p_n$07, zeros may occur inside $p_n$08 (Bénéteau et al., 2016).

In Banach spaces the geometry changes substantially. In $p_n$09, $p_n$10, $p_n$11, the set of all possible OPA zeros is exactly

$p_n$12

for some constant $p_n$13; thus the excluded disk has radius strictly between $p_n$14 and $p_n$15, and extra zeros inside $p_n$16 do occur (Cheng et al., 2021). The first-degree case is extremal in this theory: a point is an OPA zero for some degree if and only if it is a zero of an optimal linear approximant (Cheng et al., 2021). The analysis proceeds through a Lagrange-multiplier recurrence and a dynamical system for the coefficient ratios of extremal polynomials (Cheng et al., 2021).

For Hardy spaces $p_n$17, the full zero-free theory for $p_n$18 remains incomplete, but several strong results are known. If $p_n$19 is inner, or if $p_n$20 is an even integer, then the root of the nontrivial degree-one OPA is bounded away from the origin by a radius depending only on $p_n$21 (Centner et al., 2023). More generally, if $p_n$22, $p_n$23, and $p_n$24, then all OPA $p_n$25 are zero-free in a disk centered at the origin whose radius is controlled by the degree-zero error: $p_n$26 (Centner, 2021). The same work states the conjectural picture explicitly: if $p_n$27, $p_n$28, and $p_n$29, then $p_n$30 should be zero-free in $p_n$31 (Centner, 2021).

Recent work on metric projections in $p_n$32 reframes the zero question through invariant subspaces. If $p_n$33 with $p_n$34, then the zeros of $p_n$35 eventually leave every compact subset of $p_n$36 as $p_n$37, and if $p_n$38 are the zeros of $p_n$39 in $p_n$40, then their product satisfies an explicit lower bound involving $p_n$41 and the inner factor $p_n$42 of $p_n$43 (Bénéteau et al., 11 Nov 2025). The same paper identifies as a central open problem whether OPA in $p_n$44, $p_n$45, can have zeros in $p_n$46; in $p_n$47, they cannot (Bénéteau et al., 11 Nov 2025).

In the Hardy $p_n$48 case, the zeros of OPA can also be studied through orthogonal polynomials on the unit circle. For the boundary weight $p_n$49, the OPA are reversed OPUC, and their zeros therefore lie outside $p_n$50. For generalized Jacobi-type weights, the zeros satisfy explicit electrostatic balance laws, with repelling charges at singular points on $p_n$51, an attracting charge at the origin, and additional charges at zeros of an electrostatic partner polynomial $p_n$52 (Orive et al., 21 Jul 2025).

5. Convergence, cyclicity, and projections onto invariant subspaces

OPA are closely tied to cyclicity. In the Hilbert-space literature, $p_n$53 is cyclic if and only if its polynomial multiples are dense, and OPA provide a concrete finite-dimensional approximation scheme for testing this density (Bénéteau et al., 2019). In the survey formulation, for Dirichlet-type and related Hilbert spaces,

$p_n$54

(Bénéteau et al., 2021). This connects approximation of $p_n$55 to shift-invariant subspaces and to the larger problem of identifying cyclic vectors.

For polynomial $p_n$56 in weighted Hardy-type Hilbert spaces, boundary and compact-set convergence can be made highly explicit. If $p_n$57 is a polynomial with simple zeros and no zeros in $p_n$58, then in $p_n$59 or $p_n$60 the sequence $p_n$61 is uniformly bounded in the Wiener algebra norm, hence uniformly bounded on $p_n$62, and

$p_n$63

uniformly on compact subsets of $p_n$64 (Bénéteau et al., 2019). The same paper derives explicit rate information from determinant estimates on the kernel Gram matrix and treats the previously unknown higher-multiplicity case $p_n$65 (Bénéteau et al., 2019).

The projection onto the full invariant subspace generated by $p_n$66 is the limit object behind these finite approximants. In general reproducing kernel Hilbert spaces on the disk, the projections $p_n$67 converge strongly to $p_n$68, and stabilization occurs precisely when this limiting projection is already achieved by a finite polynomial multiple (Felder, 2020). For $p_n$69, the following are equivalent after some index $p_n$70: the OPA are truncations of a single power series, the OPA stabilize, and $p_n$71 (Felder, 2020). Stabilization is further characterized by a rigid inner-type factorization

$p_n$72

where $p_n$73 is $p_n$74-inner (Felder, 2020).

Inner functions occupy an extreme position in this theory. In reproducing kernel Hilbert spaces with orthogonal monomials, an $p_n$75-inner function $p_n$76 satisfies

$p_n$77

and then every OPA to $p_n$78 is constant; moreover,

$p_n$79

(Bénéteau et al., 2017). In several variables, the analogue is the class of weakly inner functions, for which all OPA are likewise constant, even though classical innerness and weak innerness need not coincide (Sargent et al., 2020).

The same projection mechanism underlies newer results in $p_n$80. For a closed $p_n$81-invariant subspace $p_n$82, the metric projection

$p_n$83

governs the asymptotics of the finite-dimensional approximants $p_n$84 (Bénéteau et al., 11 Nov 2025). When $p_n$85 is factored into inner and outer parts, the exact distance from $p_n$86 to the invariant subspace is

$p_n$87

and, for $p_n$88, the projection is generally not inner but of the form

$p_n$89

(Bénéteau et al., 11 Nov 2025). This marks a sharp departure from the $p_n$90 theory.

A nonlinear universality phenomenon also appears. If $p_n$91 is closed of measure zero, then the set of $p_n$92 whose OPA have subsequences universal on $p_n$93 is $p_n$94-dense in $p_n$95; analogous statements hold in the Dirichlet space under a logarithmic-capacity-zero hypothesis (Bénéteau et al., 2018). This result is driven by simultaneous zero-free approximation on $p_n$96 (Bénéteau et al., 2018).

6. Multivariable, operator-valued, and applied extensions

The multivariable theory retains the formal definition of OPA but acquires new algebraic and geometric complications. In a reproducing kernel Hilbert space $p_n$97 with dense polynomials and bounded coordinate shifts, one fixes an ordering of monomials $p_n$98, defines $p_n$99, and sets

$n$00

The coefficients solve a Gram system

$n$01

but zero geometry and orthogonal-polynomial recovery become much more intricate than in one variable (Sargent et al., 2020). The strong form of the Shanks conjecture fails in multivariable weighted spaces: even zero-free target polynomials can have OPA with zeros in the bidisk (Sargent et al., 2020).

Despite these difficulties, explicit multivariable models exist. For $n$02 in a scale of unit-ball spaces with kernel $n$03, one can write down closed expressions for the weighted orthogonal polynomials, their norms, the corresponding OPA, and the optimal distance, all without reduction to the one-variable case (Sargent et al., 2020). In other cases, symmetry does permit reduction: in the Drury–Arveson space, $n$04 leads to OPA of the form

$n$05

for a one-variable approximant $n$06 (Sargent et al., 2020).

The notion also extends beyond scalar analytic-function spaces. For rational matrix functions $n$07, one may seek the best polynomial approximation to $n$08 from the Krylov space $n$09. The Arnoldi-OR method computes

$n$10

so $n$11 for the degree-$n$12 polynomial that is optimal in the $n$13-norm (Chen et al., 2023). The resulting least-squares problem is built from Arnoldi Hessenberg matrices and requires $n$14 extra Arnoldi steps (Chen et al., 2023).

A different operator-valued extension appears in the $n$15-product treatment of non-autonomous linear ODEs. There the target is the $n$16-resolvent action $n$17, approximated by $n$18-polynomials

$n$19

that minimize a $n$20-norm induced by the $n$21-inner product (Pozza, 2024). Spectral reduction converts this to a classical best polynomial approximation problem for the exponential on a compact interval, yielding geometric error bounds of the form

$n$22

(Pozza, 2024).

The applied lineage of OPA is equally explicit. In digital filter design, the least-squares inverse problem for a stable filter is mathematically identical to the OPA problem in $n$23: one seeks a polynomial $n$24 minimizing $n$25, and the location of OPA zeros controls stability after reversal (Bénéteau et al., 2021). This identification is one reason OPA occupy a distinctive place between classical complex analysis and computational approximation theory.

OPA therefore form not a single theorem but a research program: a projection-theoretic mechanism that, depending on the ambient space, becomes a problem in Gram matrices, reproducing kernels, orthogonal polynomials, Jacobi matrices, Birkhoff–James geometry, invariant-subspace structure, Krylov approximation, or stable inverse design (Bénéteau et al., 2021, Bénéteau et al., 2016).