Scale-Shift and Fractional Fourier Transform as Rotations over Finite Holographic Substrate

The paper constructs the fractional Fourier transform exactly on a finite holographic substrate. The underlying framework is the literal interpretation of the holographic principle, where the arithmetical substrate is a finite totality that bounds the information capacity of every embedded observers. A shell of capacity $\kap$ is the observational register, presented by the prime field with $4\kap+1$ elements and a frame of reference.
Every row of blocks B–E is exact finite-field arithmetic; the local scale chart of D3 is a continuum reading of a finite object, declared [chart], as in the body. The corpus master ledger carries B5, C3 and D4 under its row 00:C2, B2 under its row 00:C14, and C9 as the transform-layer content of its row 00:C7.
#DescriptionStatus Sources
A. Inputs: imported, not derived here
A1p06001The framed shell: $\Fp(\chron;0,1,\gen)$ with $\p=4\kap+1$, its register $\pi=2\kap$, $\im=-\gen^{\kap}=\gen^{-\kap}$, $e=\gen^{\im}$; the frame-relative labels and the atlas of presentations $\gen\mapsto\gen^{u}$.I
A2p06002The Weil representation $\omega_\p$ of $\SL(2,\Fp)$ and its Schrödinger model, the finite Fourier transform representing the rotation $w$, the finite harmonic oscillator; the scalar convention $\omega_\p(w)^{2}=\omega_\p(-I)$, $\omega_\p(w)^{4}=I$.I
A3p06003The sign of the quadratic Gauss sum: $G=\sum_k\gen^{k^{2}}=\varepsilon(1+\im)$ for a sign $\varepsilon\in\{\pm1\}$, the in-field form of the classical evaluation $\sum_k\zeta_n^{k^{2}}=(1+i)\sqrt n$ for $4\mid n$.I
[Berndt et al. 1998; Gurevich et al. 2010]
A5p06005The framed-rational zoom: the grids $G_n$ and their $(\p-1)$-periodicity $G_{n+(\p-1)}=G_n$, the meridian shift $m\mapsto m-1$ as the refinement step.I
[Akhtman 2025, §4.3, Lemma 2]
A7p06007The continuum fractional Fourier transform and the finite constructions the family is placed against: the eigenbasis transforms, the eigenvector-free and matrix-function transforms, the canonical DFT eigenbasis, the arithmetic FrFT of the rotation group.I
[Namias 1980; Ozaktas et al. 2001; Lima & Campello de Souza 2012; Pei et al. 2011; Lima & Campello de Souza 2016; Lima et al. 2017; Gurevich & Hadani 2009; Floratos & Pavlidis 2024]
A8p06055The framework’s definition of the substrate: the Carrier, a finite totality whose count bounds the coordinate plane $\Fp\times\Fp$ of an embedded observer, $\p^{2}<\Om$; the shell of capacity $\kap$ as the observer’s register, its phase cycle of $4\kap=\p-1$ points read as the area.I
master: 00:D15
B. The shell: the frame datum and the Fourier operator
B1p06008Shell data, capacity-first: $\kap$ primary, $\p=4\kap+1$, the phase cycle $\Phit=\Z_{4\kap}$, the oriented quarter-turn $\im=-\gen^{\kap}$ (clockwise phase convention), $\pi=2\kap$, $e=\gen^{\im}$; the meridians $M_s=\{a\gen^{s}:a\in I_\p\}$; the six shells of Table 1 with their roots and quarter-turns.D
python source
§ 2, (2.1), (2.2), Def. 1
B2p06009The Euler identity is a property of the chart class: $e^{\im\pi}=\gen^{2\kap(\im\bmod2)}\equiv-1$ exactly when the quarter-turn residue is odd; the conjugate chart $(\gen,\im)\mapsto(\gen^{-1},-\im)$ toggles the parity, so one member of each conjugate pair carries it; on $\F_{13}(\chron;0,1,2)$, $\im=5$ and $6^{6}=-1$. Every primitive frame of the six shells.T
python sourceLean, the core
Rem. 2
master: 00:C14
B3p06010Chart covariance: under $\gen'=\gen^{u}$, $u\in\Zp^{\times}$, the quarter-turn name flips exactly on $u\equiv3\pmod4$ and the exponential-unit name moves already at $u\equiv1\pmod4$ whenever $\im(u-1)\not\equiv0\pmod{4\kap}$ ($\p=13$, $u=5$: $e'=2\neq6$); the objects of C3, C4 and C9 are invariant.T
python sourceLean, the core
Rem. 3
B4p06011The shell Fourier matrix $\Wt_{kj}=\gen^{jk}$ on $\Vt=\Fp^{\Phit}$, the reversal $J$, the normalized shell Fourier operator $\Ft=\im\Wt$.D
(3.2), (3.1), § 3
B5p06012$\Wt^{2}=-J$, $\Ft^{2}=J$, $\Ft^{4}=I$: the normalized operator generates the four-cycle, on the six shells of Table 1.T
python sourceLean, the coreLean, Mathlib
Lem. 1, Prop. 1
master: 00:C2
B6p06013The normalization is the unitary constant read in the field: $1/n\equiv-1$, its square roots in $\Fp$ are exactly $\pm\im$, and $(c\Wt)^{2}=J$ if and only if $c=\pm\im$.T
python sourceLean, the core
Rem. 4
B7p06014$\Wt J=J\Wt$, hence $\Ft J=J\Ft$; every $v$ is uniquely $v^{+}+v^{-}$ with $Jv^{\pm}=\pm v^{\pm}$, a symmetric vector determined by its $2\kap+1$ entries at $0,\dots,2\kap$ and an antisymmetric one by its $2\kap-1$ entries at $1,\dots,2\kap-1$.T
python sourceLean, the coreLean, Mathlib
Lem. 3
C. The fractional family
C1p06015The projectors $\Pi_\ell=\tfrac14\sum_{r=0}^{3}\im^{-\ell r}\Ft^{r}$ and the principal framed character lift $\Ft^{[s]}=\sum_\ell\gen^{-\ell s}\Pi_\ell$, $s\in\Phit$: the refinement base is the inverse root, $(\gen^{-1})^{\kap}=\im$, each projector’s character the $\ell$-th power of the meridian phase $z_s=\gen^{-s}$.D
(4.1), (4.2), Rem. 5
C2p06016$\Pi_\ell^{2}=\Pi_\ell$, $\Pi_\ell\Pi_m=0$ for $\ell\neq m$, $\sum_\ell\Pi_\ell=I$, $\Ft\Pi_\ell=\im^{\ell}\Pi_\ell$.T
python sourceLean, the coreLean, Mathlib
Lem. 2
C3p06017The exact finite-field FrFT: $s\mapsto\Ft^{[s]}$ is a representation of $\Phit$, $\Ft^{[s+r]}=\Ft^{[s]}\Ft^{[r]}$ for all $s,r$ (every pair on the six shells by the package), with the cardinal values $\Ft^{[0]}=I$, $\Ft^{[\kap]}=\Ft$, $\Ft^{[2\kap]}=J$, $\Ft^{[3\kap]}=\Ft^{-1}$, and $(\Ft^{[1]})^{\kap}=\Ft$.T
python sourceLean, the coreLean, Mathlib
Thm. 1
master: 00:C2
C4p06018Faithfulness: $s\mapsto\Ft^{[s]}$ is injective on $\Z_{4\kap}$ for every $\kap\ge1$; at $\p=5$ the surviving odd projector carries the faithful character.T
python sourceLean, the coreLean, Mathlib
Thm. 3
C5p06019The projector sums $\Pi_0+\Pi_2=\tfrac12(I+J)$ and $\Pi_1+\Pi_3=\tfrac12(I-J)$; $\Pi_0,\Pi_2\neq0$ for every $\kap\ge1$ and $\Pi_1,\Pi_3\neq0$ for $\kap\ge2$, each by one explicit entry; at $\p=5$ exactly one of $\Pi_1,\Pi_3$ vanishes.T
python sourceLean, the coreLean, Mathlib
Lem. 4
C6p06020The projectors are chart data: for $u$ odd and a unit mod $4\kap$ the index relabelling $m\mapsto um$ carries $\Ft(\gen)$ to $\Ft(\gen^{u^{2}})$ by a coordinate permutation, $\Ft(\gen^{u^{2}})_{kj}=\Ft(\gen)_{uk,uj}$; at $\p=13$ the frames $\gen=2$ and $\gen=6$ give the trace tuples $(3,3,4,2)$ and $(4,2,3,3)$, with $\operatorname{Tr}\Ft=4,9$.T
python sourceLean, the core
Rem. 6
C7p06021The multiplicity dichotomy in trace form: with $G=\sum_{k}\gen^{k^{2}}$ and $G^{*}=\sum_k\gen^{-k^{2}}$ the traces are $\operatorname{Tr}\Ft=\im G$, $\operatorname{Tr}\Ft^{2}=2$, $\operatorname{Tr}\Ft^{3}=\im G^{*}$ and $\operatorname{Tr}\Pi_\ell=\tfrac14\sum_r\im^{-\ell r}\operatorname{Tr}\Ft^{r}$; $GG^{*}=-2$ and $\varepsilon(\gen^{-1})=-\varepsilon(\gen)$; when $G=\varepsilon(1+\im)$ (A3) the tuple of traces is $(\kap,\kap,\kap+1,\kap-1)$ for $\varepsilon=+1$ and $(\kap+1,\kap-1,\kap,\kap)$ for $\varepsilon=-1$, read in $\Fp$; the sign and the tuples on the six shells of Table 1. Exact on the $38$ primitive frames of $\p\in\{5,13,17,29,37\}$ and the $16$ of $\p=41$.T
python sourceLean, the coreLean, Mathlib
Thm. 2
C8p06022Exponent lifts: every lift $a_\ell\equiv\ell\pmod4$, $U^{(a)}_s=\sum_\ell\gen^{-a_\ell s}\Pi_\ell$, is additive with the same cardinal skeleton, $U^{(a)}_{s+r}=U^{(a)}_sU^{(a)}_r$ and $U^{(a)}_{m\kap}=\Ft^{m}$.T
python sourceLean, the coreLean, Mathlib
Rem. 5
C9p06023The conjugate chart $(\gen,\im)\mapsto(\gen^{-1},-\im)$: exactly $\Ft'=-\Ft^{-1}$ and $\Pi'_\ell=\Pi_{\ell+2}$; the operator relations, cardinal values, additivity and faithfulness hold on the conjugate frame, and its trace tuple is the other pattern of C7.T
python sourceLean, the coreLean, Mathlib
Rem. 3, Thm. 2
master: 00:C7
C11p06052Charts outside the classified lifts: for $u$ a unit mod $4\kap$, $\Wt(\gen^{u})=\Wt P_u$ with $P_u$ the index relabelling $j\mapsto uj$, so $\Wt\,\Wt(\gen^{u})=-JP_u$ and $\Wt(\gen^{u})\,\Wt=-JP_{u^{-1}}$, and the two differ at the entry $(-u,1)$ when $u^{2}\not\equiv1\pmod{4\kap}$: a family canonical in such a chart does not commute with $\Ft$; $\p=29$, $u=5$, and every such chart of the six shells of Table 1 checked.T
python sourceLean, the core
Rem. 5
D. Meridional domains and the scale-shift
D1p06024The meridional domain $D_s=(\Vt,\mathcal B_s)$, $\mathcal B_s=\Ft^{[s]}\mathcal B_0$, every $\Ft^{[s]}$ invertible with inverse $\Ft^{[n-s]}$; the cardinal domains spatial, spectral, parity, inverse-spectral.D
python sourceLean, the core
Def. 5
D2p06025The $4\kap$ framed domains are pairwise distinct; read as unordered bases $B_{s+2\kap}=B_s$, since $\Ft^{[s+2\kap]}=\Ft^{[s]}J$ and $J$ permutes the standard basis.T
python sourceLean, the coreLean, Mathlib
Thm. 3, Rem. 7
D3p06026The meridian-step interval $I_\p=\{0,1,\dots,\pi\}$, the scale-shift $S_r(x)=\gen^{r}x$, the dilation by $\gen^{r}$, and the local scale chart $\gen\mapsto\lambda>1$ reading the effective step $\gen^{m}$ as $\lambda^{\tilde m}$ [chart]: the forward shift a zoom-out and the inverse shift a zoom-in, the return of the lift the scale-periodicity of A5, the readings valid while $w\gen^{r}<\p$.D
(5.1), Def. 6, Def. 7
D4p06027Scale-shift covariance: $S_r(M_m)=M_{m+r}$ for all $m$ and $r$, as ordered lists; consecutive entries of $M_m$ differ by the effective step $\gen^{m}$; $S_{r+(\p-1)}=S_r$, the periodicity of A5 in meridian form; at $\p=13$, $\gen=2$ the ladder $M_0,\dots,M_3$ at the steps $1,2,4,8$ (Section 7).T
python sourceLean, the core
Prop. 2, Cor. 1
master: 00:C2
D6p06029Scope of the two readings: the representation-side rotation and the coordinate-side scale-shift act on carriers of different dimension and are not equal as operators; they share the index group $\Phit$ and the cardinal labels, and the operator-level relation is E5–E9.D
Rem. 8
D7p06053The cycle carries exactly $2\kap$ unordered bases: every entry $(0,k)$, $k\neq0$, of $\Ft^{[t]}$ is $\tfrac{\im}{2}(1-\gen^{-2t})$, nonzero for $t\notin\{0,2\kap\}$, so $\Ft^{[t]}$ is not monomial and $B_{s+t}\neq B_s$; the six shells of Table 1.T
python sourceLean, the core
Rem. 7
E. The rotations, the Weil dictionary and the operator-level comparison
E1p06030The Weil-side data: $z_s=\gen^{-s}$, $c_s=(z_s+z_s^{-1})/2$, $d_s=(z_s-z_s^{-1})/(2\im)$, the rotation $R_s$; the pulled-back family $\operatorname{FrFT}^{\rm Weil}_\kap(s)=\omega_\p(R_s)$; cardinal Weil correspondence of two families.D
(6.1), Def. 8
E2p06031$R_s\in SO(2,\Fp)$, and $s\mapsto R_s$ is an isomorphism $\Phit\simeq SO(2,\Fp)$ with $|SO(2,\Fp)|=\p-1=4\kap$: the full rotation group of the $\p^{2}$-point plane is the phase cycle.T
python sourceLean, the coreLean, Mathlib
Prop. 3
E3p06032The cardinal-skeleton dictionary $M_s\leftrightarrow s\leftrightarrow z_s\leftrightarrow R_s$ with $R_0=I$, $R_\kap=w$, $R_{2\kap}=-I$, $R_{3\kap}=w^{-1}$ (the $R_\kap$ column of Table 1), $z_\kap=\im$; the shell’s family and the Weil family stand in cardinal Weil correspondence, the Weil side by the normalization of A2 (an import).T
python sourceLean, the coreLean, Mathlib
Thm. 4 and the paragraph after it
E4p06033Scope of the dictionary: agreement on the four cardinal indices only, the two families on spaces of different dimension over different coefficient rings, the Weil model one dimension larger as a count; no relation at an intermediate index is asserted.D
Rem. 9
E5p06034Spectral obstruction: for $\kap\ge2$ no invertible $T$ on $\Vt$ satisfies $T\sigma T^{-1}=\Ft^{[s]}$ for any $s$; the cyclic groups $\langle\sigma\rangle$ and $\langle\Ft^{[1]}\rangle$, each of order $4\kap$ ($\Ft^{[1]}$ by C4; $\sigma$ permutes the $4\kap$ labels cyclically), are not conjugate.T
python sourceLean, the coreLean, Mathlib
Prop. 4
E6p06035Common character sector: $\Ft^{[s]}\Pi_1=\gen^{-s}\Pi_1$, so on $E_1=\operatorname{im}\Pi_1\neq0$ ($\kap\ge2$) the family acts by the meridian character; the intertwiner $\Ft^{[s]}T_v=T_vS_{-s}$ on every $x\in\Fp$ and $s$, and $R_s(1,-\im)^{\mathsf T}=\gen^{-s}(1,-\im)^{\mathsf T}$.T
python sourceLean, the coreLean, Mathlib
Prop. 5
E7p06036Cardinal Heisenberg covariance: $\Ft\sigma\Ft^{-1}=D_1$, $\Ft D_1\Ft^{-1}=\sigma^{-1}$; $\Ft^{r}\sigma=\sigma_r\Ft^{r}$ with $(\sigma,D_1,\sigma^{-1},D_1^{-1})$; the expansion (6.3), $\Ft^{[s]}=\sum_r c_r(s)\Ft^{r}$ with $c_r(s)=\tfrac14\sum_\ell(\gen^{r\kap-s})^{\ell}$.T
python sourceLean, the coreLean, Mathlib
Prop. 6, § 6.1
E8p06050Cardinal exclusivity: the conjugate $\Ft^{[s]}\sigma\Ft^{[s]\,-1}$ of (6.3) is not monomial for every $s\notin\{0,\kap,2\kap,3\kap\}$ — off $j\in\{-1,0,1\}$ its row $0$ is $x^{-1}$ times a quadratic in $x=\gen^{j}$ with leading coefficient $\im\,[c_0(s)+c_2(s)]\,c_3(-s)\neq0$, $c_0(s)+c_2(s)=\tfrac12(1+\gen^{-2s})$, so the row vanishes at no three distinct indices off those three and, by counting, has at least $4\kap-5$ nonzero entries; the count checked on every shell below $200$.T
python sourceLean, the core
Thm. 5
E9p06054The spectra behind E5: every nonzero residue $b$ is an eigenvalue of $\sigma$, with the geometric eigenvector $(b^{-j})_j$ (so $\sigma$ is diagonalizable over $\Fp$ with the $4\kap$ simple eigenvalues $\Fpx$, the Mathlib reading); an eigenvalue of $\Ft^{[s]}=\sum_\ell\gen^{-\ell s}\Pi_\ell$ is one of the four $\gen^{-\ell s}$ ($\Ft^{[1]}$ on the six shells).T
python sourceLean, the core
Prop. 4
E10p06056The plane as orbits of the cycle: $R_s$ scales $u=x+\im y$ by $z_s$ and $v=x-\im y$ by $z_s^{-1}$, so every point off the origin has a free orbit of size $4\kap=|SO(2,\Fp)|$, and $\p^{2}=1+(4\kap+2)\,4\kap$ is the count of the origin and $4\kap+2$ orbits (the $4\kap$ circles of nonzero norm and the two isotropic lines, by counting); $\dim\Vt=4\kap$.T
python sourceLean, the core
Cor. 2

Status

I
Import. A standard result, a corpus result, or a convention used here without reproof.
D
Definition. A naming, a declared dictionary, or a set-up move, carrying no empirical content.
T
Theorem. Derived within this paper, checked in exact arithmetic by the named python check, and proved in Lean 4 in the framework’s core library with no axioms.

Sources

python source
Gold: the python check that decides the predicate, at its line in the script (the source, rendered on this site); on hover the deciding family and the families that corroborate it, and the verdict of the last run (red when it fails). The script page's bar runs the predicate's cell of the package's notebook on Google Colab (one cell per predicate; the whole notebook from the header badge).
Lean, the coreLean, Mathlib
The predicate as one Lean declaration, named by its accession key (p06008) — the conjunction of the theorems that prove it — at its line in the rendered module, one mark per library that proves it, the core's first. Green: the core (lean/FrcCore) — the statement on the shell ring, by computation and induction, no Mathlib, no axioms: the finitist witness. Violet: Mathlib (lean/FrcLedger) — the classical statement for every shell on Mathlib's hierarchy (its axioms, propext, Classical.choice, Quot.sound, on hover). Red: a proof the gate rejects (an unfinished proof, native_decide) — not a witness. A predicate proved in both libraries shows both marks; one mark, one proof. The module page's bar opens the module as one executable file in Lean Live (the core on plain Lean, no Mathlib; the file elaborates in the browser and prints every predicate's axioms at its end); the header's two Lean badges count the marks of each colour and open the two executables.
section
The statement in the paper's body that carries the predicate: a proposition, a remark or a section.
rests on
The predicates it rests on, by label: realisations, definitions and imports of this ledger, predicates of other ledgers (32:B3) and of the master (00:D15).
Collecting the tags separates what a reader must grant from what is derived. Imported are the framed shell (A1), the Weil representation with its scalar convention (A2), the sign of the Gauss sum (A3), the framed-rational zoom (A5), the continuum transform with the finite constructions it is compared with (A7) and the framework’s definition of the substrate (A8). Declared are the shell data (B1), the Fourier matrix and its normalization (B4), the projectors and the principal lift (C1), the domains (D1), the scale map and its chart (D3), the scope of the two readings (D6), the Weil-side data (E1) and the scope of the dictionary (E4); no realisation row is claimed, since the paper makes no physical identification of its own. Theorems are the twenty-five rows tagged T: the shell arithmetic (B2, B3, B5–B7), the fractional family (C2–C9, C11), the domains and the scale-shift (D2, D4, D7) and the rotations, the dictionary and the operator-level comparison (E2, E3, E5–E10). Every theorem row, and the definition D1, is proved in Lean 4 in the framework’s core library with no axioms: each such row is the conjunction of the core declarations its source cell names, up to four kinds of clause stated as such in the row. The clauses that name a shell or a chart (the six shells, the $38+16$ frames of C7, the $\p=13$ instances of C6 and D4) are decided by the kernel or by the package. The imports a row consumes are named in it: A3’s sign in C7, A2’s normalization in E3; D4 cites A5 as the prior form of the periodicity its key proves. The one-line consequences of the declarations are drawn in the row, among them B2’s toggle of the parity under the conjugate chart, B3’s law of the exponential unit and the invariance of C3, C4 and C9 (checked by the package), C9’s relations on the conjugate frame, C11’s non-commutation, D2’s distinct domains from the injectivity of C4, D7’s count $2\kap$ from the transforms that are not monomial, E2’s count of the rotation group, E5’s order $4\kap$ of the two cyclic groups, E6’s $E_1\neq0$ from C5, E7’s expansion, E9’s diagonalizability from the simple eigenvalues, and the two counts E8’s $4\kap-5$ and E10’s $4\kap+2$ orbits. The one reading a row draws from its declarations is marked as a reading: D3’s chart. Of the theorem rows, B5, B7, C2–C5, C7–C9, D2, E2, E3 and E5–E7 are also proved on Mathlib’s hierarchy for every shell; C7’s Mathlib key keeps the rank form of the dichotomy, and C5’s its rank sums.