The Setting
Changing the predual here means selecting a subspace of a real dual by perturbing its coordinate quotient map. The perturbation combines operators attached to pure and mixed families of coded paths.
The coordinate set is , the first uncountable ordinal. Fix a bijection .✦ The color classes and their parametrizations are defined in (1); each class has cardinality and is unbounded. (1) An array on is a finitely supported rational-valued function on . The operation families are the pure families , for , and the mixed family .✦ Supports are successive for a family when every coordinate of the first precedes every coordinate of the second; in the mixed family their colors must also increase.✦
For a family , let be its primal exponent, its family index, and its source color, as specified in (2). For a family , the path exponent is , and the source exponent is . (2) For a natural parameter , the Hölder exponent is when , and .
The Norming Completion
The family operations use block lengths and normalization factors at positive weights. Their underlying parameter sequences start at zero. Put .✦ The exponent scale is the first coordinate of the recursive parameter state in (3). (3) The scales and and their family versions are given by (4). Thus weight uses underlying parameter index ; subtraction of natural numbers stops at zero. Sequence position corresponds to positive position . (4)
A Type I expression in a family is a finite list of nonzero successive children , together with a positive weight and a length .✦ The normalized value of a Type I expression in a family of weight with children is .✦ The pure base at color consists of zero and the signed singleton arrays with .✦ Starting from this base, stage zero, each next stage retains the preceding stage and adjoins Type I values whose children belong to that stage. It also adjoins the values of Type II path expressions when every child of every original uncropped piece on each path belongs to that stage. The pure norming set is the union of these stages for .✦ The mixed family starts from and uses the same stage rule with mixed succession. Its union is the full norming set .✦ Write for the stage union belonging to a family .✦
A norming set on is a set of rational arrays containing every and having all coefficients bounded in absolute value by one.✦ For the full norming set, the bounded function space is .✦ The coordinate evaluation vector is the bounded function on .✦ The finite support inclusion , its closed evaluation span , and the coordinate predual are specified in (5). Here is evaluation at restricted to ✦, and is the coordinate functional.✦ (5) The space has the inherited norm, and also denotes the finite support inclusion with target restricted to .✦
Coded Paths and Their Functionals
The coherent finite closure system assigns to each a sequence of finite subsets of . Each sequence contains its endpoint, increases under inclusion, and exhausts ; coherence means when . A predecessor closure family at assigns a finite subset of to each and each natural index .✦ A predecessor closure family at is coherent when these sets lie in , each sequence contains its endpoint, increases under inclusion, exhausts , and satisfies the coherence condition below . A coherent endpoint extension of a predecessor closure family at is an increasing, exhaustive sequence of finite subsets of containing , whose restrictions at predecessor points agree with the predecessor closure family whenever those points occur.✦ The endpoint construction assigns a sequence of finite sets to a predecessor family, and the closure system is obtained by recursion over the endpoints. The positive-indexed closure is , with index zero also selecting .✦ The closure entry rank is the first positive for which when , and is zero otherwise.✦
A raw infinite path consists of nonzero rational arrays , declared weights , and metadata , indexed from zero.✦ Its prefix support , endpoint , and prefix closure are given by (6). (6) For a finite set , the order position of is the number of points of strictly below .✦ The positional array code of a rational array on relative to a finite set is the finite set of pairs consisting of the number of points of strictly below and , for ; zero coefficients are retained.✦ A finite path code is a list of triples recording weight, metadata, and a positional array code.✦ The code is the list of these triples for , all using the common closure .✦ The maximum metadata of a finite path code is the maximum of its metadata entries, with value zero for an empty list.✦ The coding map pairs the canonical natural-number code of its argument with one plus its maximum metadata.✦ It is injective, and its value exceeds every metadata entry of its argument.
A raw infinite path is admissible when its supports are successive in the ordinal order and its metadata are positive and strictly increasing, with the remaining conditions in (7). (7) An admissible prefix of length of a raw infinite path consists of its first entries, with successive supports in the ordinal order, positive and strictly increasing metadata, and the conditions in (7) at indices below , with the coding rule required only when the next index is also below .
A finite path in a family consists of a positive length , raw infinite data with an admissible prefix of that length, and original uncropped Type I expressions at indices below . Their values and weights are and , and the original rational pieces are successive for . The finite path sum of a finite path of length with rational pieces is .✦ An infinite path in a family consists of admissible raw data and original uncropped Type I expressions at every index , whose values are and whose weights are , with family succession and every child in .✦ The prefix through of an infinite path is the finite path retaining the raw data of and its first expressions; the zeroth prefix contains one piece.✦
For an infinite path , the partial functional is the norming evaluation functional associated with the sum of its prefix through .✦ The chosen path functional is its pointwise limit, as in (8).✦ The raw path index set consists of the raw data admitting an infinite path in ✦; for each , choose one such representative .✦ (8)
The Definition
Definition 1. The dual quotient is defined by (9). (9)✦ Write for the canonical bounded quotient map. For a family and , the path coset is given by (10). (10)✦ For a family , finite path synthesis is the linear map from finitely supported real families on to given by . Identify these coefficients with their natural images in the real sequence spaces .✦ The path synthesis operator in (11) is the continuous extension of finite path synthesis. (11)✦ For an exponent and an index set , the finitely supported sequence embedding is the linear inclusion of finitely supported real families on into that preserves every coordinate .
Assume the continuum hypothesis, . Choose a base path labelling injection .✦ The colored path labelling is defined in (12). (12)✦ For a family , finite color synthesis is the linear map from finitely supported real families on to that sends to . Its continuous extension is .✦ For an index set and conjugate exponents , the dual coefficient sequence of a functional is , where is the unit sequence.✦ For a bounded linear map , the dual analysis operator of is the map that sends to the dual coefficient sequence of .✦ The color analysis operator and the family perturbation operator are specified in (13). (13) Enumerate the families by and ✦, and put .✦ The bounded perturbation operator is defined by (14). (14)✦ The changed predual is the kernel in (15). (15)✦
What Is Checked
The definitions concern the real norming completion of the full pure and mixed norming set on ; colored labelling and the perturbation use the continuum hypothesis. Every finite prefix of an infinite path has its sum in ✦, and every raw prefix support is nonempty.✦ Every path functional has norm at most one. The source and path exponents are Hölder conjugate.✦ The quotient is complete.✦