PNCW Paper 05
全域計算、局部顯現:投影路由、解析度場與物化決策
Compute Globally, Materialize Selectively:
Projection Routing, Resolution Fields, and Materialization Decisions
版本:v0.1 日期:2026-08-28 系列:Projection-Native Computational World Series / 投影原生計算世界系列 定位:Series Paper 05 / Global Routing, Resolution, and Materialization Governance Layer 依賴:PNCW Paper 00–04、Global Computation Methodology(GCM)Series、GCM Phase A Reference Runtime、GCM Phase B Deterministic Compute Allocator 作者: Neo.K機構: EveMissLab/一言諾科技有限公司
摘要
本文延續 Projection-Native Computational World(PNCW)Series,建立 PNCW 的 Global Compute / Selective Materialization Governance Layer ,回答前四篇尚未統一解決的 Runtime 問題:
當 canonical world、Context World、projected carrier、visual canvas 與外部 resources 同時存在時,究竟由誰決定「現在要算什麼、在哪裡算、用什麼 representation、用多高解析度、哪些部分值得 materialize、哪些應維持 dormant、哪些應該被投影給哪個 observer」?
本文以 Global Computation Methodology(GCM)作為主要治理框架,將其核心原則:
Computation ≠ Observation \boxed{
\text{Computation}
\neq
\text{Observation}
} Computation = Observation
Global Dependency ≠ Full Materialization \boxed{
\text{Global Dependency}
\neq
\text{Full Materialization}
} Global Dependency = Full Materialization
Recursive Globality ≠ Recursive Full Expansion \boxed{
\text{Recursive Globality}
\neq
\text{Recursive Full Expansion}
} Recursive Globality = Recursive Full Expansion
以及:
Finite Active Realization + Unbounded Extensibility \boxed{
\text{Finite Active Realization}
+
\text{Unbounded Extensibility}
} Finite Active Realization + Unbounded Extensibility
正式嵌入 PNCW。
本文將 PNCW Runtime state 定義為:
W t = ⟨ W t , D t , P , L , Λ t , C t , S t , Π t , H t ⟩ . \boxed{
\mathfrak W_t
=
\left\langle
W_t,
\mathcal D_t,
\mathfrak P,
\mathfrak L,
\Lambda_t,
\mathcal C_t,
\mathcal S_t,
\Pi_t,
\mathcal H_t
\right\rangle.
} W t = ⟨ W t , D t , P , L , Λ t , C t , S t , Π t , H t ⟩ .
其中:
W t W_t W t :canonical world state;
D t \mathcal D_t D t :active/potential computational domains;
P \mathfrak P P :computational form / representation space;
L \mathfrak L L :transition-law family;
Λ t \Lambda_t Λ t :resolution / materialization field;
C t \mathcal C_t C t :global constraints / couplings;
S t \mathcal S_t S t :scheduling / routing / allocation;
Π t \Pi_t Π t :observer / carrier / context projections;
H t \mathcal H_t H t :history / provenance。
本文進一步將 GCM 的 Materialization Policy 形式化為:
M a t P o l i c y t : ( W t , q , O , R , B , R i s k ) → M t , \boxed{
\mathsf{MatPolicy}_t
:
(
W_t,
q,
O,
R,
B,
Risk
)
\rightarrow
\mathcal M_t,
} MatPolicy t : ( W t , q , O , R , B , R i s k ) → M t ,
其中 M t \mathcal M_t M t 是此時此刻值得 active realization 的 representation set。
本文特別區分:
λ c o m p u t e ≠ λ o b s e r v e ≠ λ c a r r i e r ≠ λ r e n d e r . \boxed{
\lambda^{compute}
\neq
\lambda^{observe}
\neq
\lambda^{carrier}
\neq
\lambda^{render}.
} λ co m p u t e = λ o b ser v e = λ c a r r i er = λ r e n d er .
也就是 compute resolution、observer resolution、carrier resolution 與 render resolution 不必相同。
本文亦將 GCM Phase B 的 deterministic allocation 思想納入 PNCW,使「顯示哪個 projection」不再由 UI 任意決定,而可以經:
Constraint → Feasible Plans → Objective Evaluation → Pareto Frontier → Explicit Policy Selection \boxed{
\text{Constraint}
\rightarrow
\text{Feasible Plans}
\rightarrow
\text{Objective Evaluation}
\rightarrow
\text{Pareto Frontier}
\rightarrow
\text{Explicit Policy Selection}
} Constraint → Feasible Plans → Objective Evaluation → Pareto Frontier → Explicit Policy Selection
選出一個可稽核的 active realization plan。
本文最終提出:
Global Coherence ≠ Global Simultaneous Materialization . \boxed{
\text{Global Coherence}
\neq
\text{Global Simultaneous Materialization}.
} Global Coherence = Global Simultaneous Materialization .
以及:
Compute Globally, Materialize Selectively, Observe Relatively . \boxed{
\text{Compute Globally, Materialize Selectively, Observe Relatively}.
} Compute Globally, Materialize Selectively, Observe Relatively .
這使 PNCW 的「一口氣看到」獲得真正的全域 runtime 基礎:不是把世界全部一次 render,而是讓 canonical world 維持全域一致,讓 Runtime 依 task、dependency、budget、risk、observer 與 resource state,選擇當下最值得 active realization 的部分。
關鍵詞: GCM、Selective Materialization、Resolution Field、Projection Routing、Finite Active Realization、Pareto Allocation、Global Computation、PNCW
0. 研究目的
PNCW Paper 00–04 已經建立:
M t o t a l → C q a c t i v e → E k → P k → V q , k . \boxed{
\mathcal M^{total}
\rightarrow
C_q^{active}
\rightarrow
\mathcal E_k
\rightarrow
P_k
\rightarrow
V_{q,k}.
} M t o t a l → C q a c t i v e → E k → P k → V q , k .
但仍缺一個更高層問題:
誰決定這條 projection chain 什麼時候啟動、啟動多大、走哪一條 representation route、用多少 resource?
本文引入:
G l o b a l R u n t i m e G o v e r n o r . \boxed{
\mathsf{GlobalRuntimeGovernor}.
} GlobalRuntimeGovernor .
1. Global 不等於 Everything Everywhere
PNCW 接受 GCM 的核心重定義:
Global = coherence relative to a designated World boundary . \boxed{
\text{Global}
=
\text{coherence relative to a designated World boundary}.
} Global = coherence relative to a designated World boundary .
它不等於:
all-to-all;
same algorithm;
same representation;
same timestep;
full materialization;
global barrier。
2. Global Computation
對 domains:
D 1 , … , D n , D_1,\ldots,D_n, D 1 , … , D n ,
每個 domain 可使用:
p i ( t ) ∈ P , p_i(t)\in\mathfrak P, p i ( t ) ∈ P ,
ℓ i ( t ) ∈ L . \ell_i(t)\in\mathfrak L. ℓ i ( t ) ∈ L .
局部演化:
Φ i p i , ℓ i . \Phi_i^{p_i,\ell_i}. Φ i p i , ℓ i .
全域演化:
Φ G ( t ) = Compose C G ( t ) ( Φ 1 p 1 , ℓ 1 , … , Φ n p n , ℓ n ) . \boxed{
\Phi_G(t)
=
\operatorname{Compose}_{\mathcal C_G(t)}
\left(
\Phi_1^{p_1,\ell_1},
\ldots,
\Phi_n^{p_n,\ell_n}
\right).
} Φ G ( t ) = Compose C G ( t ) ( Φ 1 p 1 , ℓ 1 , … , Φ n p n , ℓ n ) .
並:
W t + 1 = Φ G ( t ) ( W t ) . W_{t+1}
=
\Phi_G(t)(W_t). W t + 1 = Φ G ( t ) ( W t ) .
3. Computational Form Space
令:
P = { sequence , graph , matrix , tensor , symbolic , simulation , image-carrier , canvas , … } . \boxed{
\mathfrak P
=
\{
\text{sequence},
\text{graph},
\text{matrix},
\text{tensor},
\text{symbolic},
\text{simulation},
\text{image-carrier},
\text{canvas},
\ldots
\}.
} P = { sequence , graph , matrix , tensor , symbolic , simulation , image-carrier , canvas , … } .
PNCW 不要求所有 domain 使用同一 representation。
4. Representation Routing
對 domain D i D_i D i 與 task (q):
p i ⋆ = R o u t e R e p ( D i , q , C o n s t r a i n t s , C o s t , R i s k ) . \boxed{
p_i^\star
=
\mathsf{RouteRep}
(
D_i,
q,
Constraints,
Cost,
Risk
).
} p i ⋆ = RouteRep ( D i , q , C o n s t r ain t s , C os t , R i s k ) .
因此:
Global Computation ≠ One Representation Everywhere . \boxed{
\text{Global Computation}
\neq
\text{One Representation Everywhere}.
} Global Computation = One Representation Everywhere .
5. Transition-Law Family
不同 domain 可以使用不同:
ℓ i ∈ L . \ell_i
\in
\mathfrak L. ℓ i ∈ L .
例如:
deterministic update;
probabilistic transition;
graph propagation;
optimization step;
symbolic rewrite;
neural inference;
event-driven transition。
6. Domain
定義 domain:
D i = ⟨ S t a t e , R e p r e s e n t a t i o n , L a w , R e s o l u t i o n , D e p e n d e n c i e s , A u t h o r i t y , H i s t o r y ⟩ . \boxed{
D_i
=
\left\langle
State,
Representation,
Law,
Resolution,
Dependencies,
Authority,
History
\right\rangle.
} D i = ⟨ S t a t e , R e p r ese n t a t i o n , L a w , R eso l u t i o n , D e p e n d e n c i es , A u t h or i t y , H i s t or y ⟩ .
Domain 是 GCM / PNCW 的基本 routing unit。
7. Domain 不等於 Process
一個 process 可以跨多 domain。
一個 domain 也可以由多 executor 共同處理。
因此:
Domain ≠ Process . \boxed{
\text{Domain}
\neq
\text{Process}.
} Domain = Process .
8. Global Dependency 不等於 Full Materialization
如果:
D i → D j D_i
\rightarrow
D_j D i → D j
存在 dependency,
不表示:
D i , D j D_i,
D_j D i , D j
所有 detail 必須同時 materialize。
因此:
Global Dependency ≠ Full Materialization . \boxed{
\text{Global Dependency}
\neq
\text{Full Materialization}.
} Global Dependency = Full Materialization .
9. Materialization
定義:
M a t e r i a l i z e : X → X r e s i d e n t . \boxed{
\mathsf{Materialize}
:
X
\rightarrow
X^{resident}.
} Materialize : X → X r es i d e n t .
其中 resident 可以是:
RAM;
VRAM;
local DB;
HDSRC tile;
Context Packet;
Canvas region;
rendered viewport。
10. Materialization 不等於 Existence
若:
X ∉ M t , X\notin M_t, X ∈ / M t ,
不表示:
X X X
不存在。
所以:
Not Materialized ≠ Nonexistent . \boxed{
\text{Not Materialized}
\neq
\text{Nonexistent}.
} Not Materialized = Nonexistent .
11. 五種 Runtime State
對 object (X):
R u n t i m e S t a t e ( X ) ∈ { A C T I V E , M A T E R I A L I Z E D , D O R M A N T , A R C H I V E D , P O T E N T I A L } . \boxed{
\mathsf{RuntimeState}(X)
\in
\{
\mathsf{ACTIVE},
\mathsf{MATERIALIZED},
\mathsf{DORMANT},
\mathsf{ARCHIVED},
\mathsf{POTENTIAL}
\}.
} RuntimeState ( X ) ∈ { ACTIVE , MATERIALIZED , DORMANT , ARCHIVED , POTENTIAL } .
12. Active
X ∈ A t X\in A_t X ∈ A t
表示目前參與 computation。
13. Materialized
X ∈ M t X\in M_t X ∈ M t
表示細節已展開、可直接 access / render。
14. Dormant
存在 canonical state,但暫不 high-frequency update。
15. Archived
只保存:
checkpoint;
seed;
summary;
history pointer;
compressed carrier。
16. Potential
structure 尚未 materialize,但 grammar / generator / rule 允許未來生成。
17. Active 不等於 Materialized
一般:
A t ≠ M t . \boxed{
A_t
\neq
M_t.
} A t = M t .
可能:
X ∈ A t , X ∉ M t , X\in A_t,
\quad
X\notin M_t, X ∈ A t , X ∈ / M t ,
例如 hidden low-resolution solver。
也可能:
X ∈ M t , X ∉ A t , X\in M_t,
\quad
X\notin A_t, X ∈ M t , X ∈ / A t ,
例如 frozen UI snapshot。
18. Finite Active Realization
真實 Runtime 必須:
∣ A t ∣ < ∞ ∀ t . \boxed{
|A_t|<\infty
\qquad
\forall t.
} ∣ A t ∣ < ∞ ∀ t .
更強可要求:
∣ A t ∣ ≤ B A . |A_t|
\le
B_A. ∣ A t ∣ ≤ B A .
19. Unbounded Extensibility
World 可以:
W 0 → W 1 → W 2 → ⋯ W_0
\rightarrow
W_1
\rightarrow
W_2
\rightarrow
\cdots W 0 → W 1 → W 2 → ⋯
而不預設固定最大深度。
但:
Unbounded Extensibility ⇏ Infinite Simultaneous Computation . \boxed{
\text{Unbounded Extensibility}
\not\Rightarrow
\text{Infinite Simultaneous Computation}.
} Unbounded Extensibility ⇒ Infinite Simultaneous Computation .
20. Recursive Globality / Recursive Full Expansion Non-Collapse
Recursive Globality ≠ Recursive Full Expansion . \boxed{
\text{Recursive Globality}
\neq
\text{Recursive Full Expansion}.
} Recursive Globality = Recursive Full Expansion .
這直接適用於:
Context World;
Recursive Canvas;
HDSRC multi-scale carrier;
nested simulation;
project graph。
21. Bounded Frontier
定義 active frontier:
∂ A t \boxed{
\partial A_t
} ∂ A t
為目前 active world 與 potential / dormant world 的邊界。
Runtime 只需管理 bounded frontier。
22. Rolling Horizon
對長期 evolution:
H t = [ t , t + h ] . H_t
=
[t,t+h]. H t = [ t , t + h ] .
只 materialize有限 rolling horizon:
∣ H t ∣ < ∞ . \boxed{
|H_t|<\infty.
} ∣ H t ∣ < ∞.
23. Checkpoint Reconstruction
如果某 region dormant / archived:
X t X_t X t
可以從:
C h e c k p o i n t + S e e d + E v e n t s Checkpoint
+
Seed
+
Events C h ec k p o in t + S ee d + E v e n t s
重建。
因此:
Persistence ≠ Full Residency . \boxed{
\text{Persistence}
\neq
\text{Full Residency}.
} Persistence = Full Residency .
24. Resolution Field
本文引入:
Λ t : D t → L λ . \boxed{
\Lambda_t
:
\mathcal D_t
\rightarrow
\mathcal L_\lambda.
} Λ t : D t → L λ .
其中:
Λ t ( D i ) = λ i ( t ) . \Lambda_t(D_i)
=
\lambda_i(t). Λ t ( D i ) = λ i ( t ) .
25. Resolution 是 Field,不是 Global Constant
不同 domain:
λ i ≠ λ j \lambda_i
\neq
\lambda_j λ i = λ j
完全合法。
因此 World 可以同時:
one domain coarse;
one domain fine;
one dormant;
one fully materialized;
one checkpoint-only。
26. Compute Resolution
定義:
λ i c o m p u t e . \lambda_i^{compute}. λ i co m p u t e .
表示 domain 真正計算所需 fidelity。
27. Observe Resolution
定義:
λ i o b s e r v e . \lambda_i^{observe}. λ i o b ser v e .
表示 observer projection fidelity。
28. Carrier Resolution
定義:
λ i c a r r i e r . \lambda_i^{carrier}. λ i c a r r i er .
表示 projected carrier 的 spatial / semantic / quantization fidelity。
29. Render Resolution
定義:
λ i r e n d e r . \lambda_i^{render}. λ i r e n d er .
表示最後 viewport / human output render fidelity。
30. 四種 Resolution Non-Collapse
因此:
λ c o m p u t e ≠ λ o b s e r v e ≠ λ c a r r i e r ≠ λ r e n d e r . \boxed{
\lambda^{compute}
\neq
\lambda^{observe}
\neq
\lambda^{carrier}
\neq
\lambda^{render}.
} λ co m p u t e = λ o b ser v e = λ c a r r i er = λ r e n d er .
它們可以相等,但不是理論必然。
31. Hidden High-Fidelity Compute
可以:
λ c o m p u t e > λ o b s e r v e . \lambda^{compute}
>
\lambda^{observe}. λ co m p u t e > λ o b ser v e .
例如內部高精度 simulation,但 UI 只顯示 summary。
32. High-Resolution View / Low-Fidelity Compute
也可以:
λ r e n d e r > λ c o m p u t e . \lambda^{render}
>
\lambda^{compute}. λ r e n d er > λ co m p u t e .
例如 frozen data 以高解析圖形 presentation,但 underlying compute 已停止。
33. Adaptive Refinement
如果:
R i s k ( D i ) Risk(D_i) R i s k ( D i )
或:
U n c e r t a i n t y ( D i ) Uncertainty(D_i) U n cer t ain t y ( D i )
升高,
可以:
λ i ↑ . \lambda_i
\uparrow. λ i ↑ .
34. De-Refinement
如果 relevance 下降:
λ i ↓ . \lambda_i
\downarrow. λ i ↓ .
因此:
Refinement is reversible runtime policy . \boxed{
\text{Refinement}
\text{ is reversible runtime policy}.
} Refinement is reversible runtime policy .
35. Materialization Policy
定義:
M a t P o l i c y t : ( W t , q , O , R , B , R i s k ) → M t . \boxed{
\mathsf{MatPolicy}_t
:
(
W_t,
q,
O,
R,
B,
Risk
)
\rightarrow
\mathcal M_t.
} MatPolicy t : ( W t , q , O , R , B , R i s k ) → M t .
36. Materialization Decision Factors
可包含:
query frequency;
observer demand;
dependency;
cost;
predicted future use;
safety criticality;
uncertainty;
storage pressure;
latency;
provenance requirement;
authority。
37. What Deserves Active Realization Now?
GCM materialization policy 的核心是:
What representation deserves active realization now? \boxed{
\text{What representation deserves active realization now?}
} What representation deserves active realization now?
這也是 PNCW Paper 05 的中心問題。
38. Context Materialization
對 PNCW Paper 02:
M a t C o n t e x t : M t o t a l → C q r e s i d e n t / a c t i v e . \mathsf{MatContext}
:
\mathcal M^{total}
\rightarrow
C_q^{resident/active}. MatContext : M t o t a l → C q r es i d e n t / a c t i v e .
39. Carrier Materialization
對 Paper 03:
M a t C a r r i e r : C q a c t i v e → P k a c t i v e − r e g i o n s . \mathsf{MatCarrier}
:
C_q^{active}
\rightarrow
P_k^{active-regions}. MatCarrier : C q a c t i v e → P k a c t i v e − r e g i o n s .
40. Visual Materialization
對 Paper 04:
M a t V i s u a l : P k → V v i e w p o r t . \mathsf{MatVisual}
:
P_k
\rightarrow
V_{viewport}. MatVisual : P k → V v i e w p or t .
41. Materialization Stack
因此:
M a t = M a t C o n t e x t ∘ M a t C a r r i e r ∘ M a t V i s u a l . \boxed{
\mathsf{Mat}
=
\mathsf{MatContext}
\circ
\mathsf{MatCarrier}
\circ
\mathsf{MatVisual}.
} Mat = MatContext ∘ MatCarrier ∘ MatVisual .
這些不是同一層。
42. Materialization / Projection Non-Collapse
Projection:
Π ( X ) \Pi(X) Π ( X )
可以只產生 logical view。
Materialization:
M a t ( Π ( X ) ) \mathsf{Mat}(\Pi(X)) Mat ( Π ( X ))
才把所需 detail resident。
所以:
Projection ≠ Materialization . \boxed{
\text{Projection}
\neq
\text{Materialization}.
} Projection = Materialization .
43. Projection / Observation Non-Collapse
同樣:
Projection ≠ Observation . \boxed{
\text{Projection}
\neq
\text{Observation}.
} Projection = Observation .
一個 projection 可以先 machine-ready,但還沒 human-visible。
44. Observation / Presentation Non-Collapse
Observer 取得 state 後,還可有不同 presentation。
因此:
Observation ≠ Presentation . \boxed{
\text{Observation}
\neq
\text{Presentation}.
} Observation = Presentation .
45. GCM Projection Operator
對 observer (O):
Π O : W t → Y O . \boxed{
\Pi_O:
W_t
\rightarrow
Y_O.
} Π O : W t → Y O .
Projection 可以:
hide;
aggregate;
filter;
quantize;
resample;
transform modality;
summarize history;
select relation layer。
46. Read-Side by Default
一般:
Π O ( W t ) is read-side by default . \boxed{
\Pi_O(W_t)
\text{ is read-side by default}.
} Π O ( W t ) is read-side by default .
Projection 不應偷偷 mutate world。
47. Observer Operation
U O : O → O ′ . \mathcal U_O:
O
\rightarrow
O'. U O : O → O ′ .
例如:
zoom;
focus;
layer toggle;
resolution change。
通常:
Δ W = 0. \Delta W=0. Δ W = 0.
48. World Operation
U W : W t → W t + 1 . \mathcal U_W:
W_t
\rightarrow
W_{t+1}. U W : W t → W t + 1 .
例如:
state update;
topology event;
database mutation;
agent action;
transaction commit。
49. Observer / World Operation Non-Collapse
U O ≠ U W . \boxed{
\mathcal U_O
\neq
\mathcal U_W.
} U O = U W .
這與 MRMIC Canvas 的 view operation / world operation 分離一致。
50. Compute Globally, Observe Selectively
本文採用:
Compute Globally, Observe Selectively . \boxed{
\text{Compute Globally, Observe Selectively}.
} Compute Globally, Observe Selectively .
更精確:
global dependency、constraint、history 與 domain evolution 由 canonical Runtime 管理;每個 observer 只取得所需 projection。
51. PNCW 擴張版
本文進一步提出:
Compute Globally, Materialize Selectively, Observe Relatively . \boxed{
\text{Compute Globally, Materialize Selectively, Observe Relatively}.
} Compute Globally, Materialize Selectively, Observe Relatively .
52. Global Coherence / Global Materialization Non-Collapse
Global Coherence ≠ Global Simultaneous Materialization . \boxed{
\text{Global Coherence}
\neq
\text{Global Simultaneous Materialization}.
} Global Coherence = Global Simultaneous Materialization .
53. Projection Route
定義:
ρ = ⟨ S o u r c e D o m a i n , R e p r e s e n t a t i o n , R e s o l u t i o n , C a r r i e r , O b s e r v e r , R e s o u r c e , P o l i c y ⟩ . \boxed{
\rho
=
\left\langle
SourceDomain,
Representation,
Resolution,
Carrier,
Observer,
Resource,
Policy
\right\rangle.
} ρ = ⟨ S o u r ceD o main , R e p r ese n t a t i o n , R eso l u t i o n , C a r r i er , O b ser v er , R eso u r ce , P o l i cy ⟩ .
54. Route 不是單一 path
Route 可以是一個 DAG:
D 0 → { D 1 , D 2 } → D 3 . D_0
\rightarrow
\{D_1,D_2\}
\rightarrow
D_3. D 0 → { D 1 , D 2 } → D 3 .
因此:
Projection Route ≠ Single Linear Pipeline . \boxed{
\text{Projection Route}
\neq
\text{Single Linear Pipeline}.
} Projection Route = Single Linear Pipeline .
55. Route Candidate
ρ i ∈ R c a n d . \rho_i
\in
\mathcal R^{cand}. ρ i ∈ R c an d .
每個 candidate 有:
feasibility;
latency;
cost;
quality;
risk;
migration cost;
materialization cost。
56. Constraint First
先求:
R f e a s i b l e = { ρ i : C o n s t r a i n t s ( ρ i ) = P A S S } . \boxed{
\mathcal R^{feasible}
=
\{
\rho_i:
Constraints(\rho_i)=PASS
\}.
} R f e a s ib l e = { ρ i : C o n s t r ain t s ( ρ i ) = P A S S } .
再做 optimization。
57. Feasible 不等於 Preferred
Feasible ≠ Preferred . \boxed{
\text{Feasible}
\neq
\text{Preferred}.
} Feasible = Preferred .
58. Objective Vector
對 route:
o ( ρ i ) = ( L a t e n c y , M e m o r y , I O , E n e r g y , C o s t , Q u a l i t y , R i s k , M i g r a t i o n , V i s i b i l i t y L a t e n c y ) . \boxed{
\mathbf o(\rho_i)
=
(
Latency,
Memory,
IO,
Energy,
Cost,
Quality,
Risk,
Migration,
VisibilityLatency
).
} o ( ρ i ) = ( L a t e n cy , M e m or y , I O , E n er g y , C os t , Q u a l i t y , R i s k , M i g r a t i o n , V i s ibi l i t y L a t e n cy ) .
59. Pareto Frontier
定義:
P = ParetoFront ( R f e a s i b l e ) . \boxed{
\mathcal P
=
\operatorname{ParetoFront}
(
\mathcal R^{feasible}
).
} P = ParetoFront ( R f e a s ib l e ) .
60. Pareto 不等於 Weighted Score
Pareto ≠ Weighted Score . \boxed{
\text{Pareto}
\neq
\text{Weighted Score}.
} Pareto = Weighted Score .
61. Explicit Policy Selection
對 frontier:
P , \mathcal P, P ,
由 policy:
P P P
選:
ρ ⋆ = P ( P ) . \boxed{
\rho^\star
=
P(\mathcal P).
} ρ ⋆ = P ( P ) .
62. Preferred 不等於 Universal Optimum
Policy Preferred ≠ Universally Optimal . \boxed{
\text{Policy Preferred}
\neq
\text{Universally Optimal}.
} Policy Preferred = Universally Optimal .
63. Selection / Execution / Commit Non-Collapse
即使:
ρ ⋆ \rho^\star ρ ⋆
被選中,
也:
Selected Route ≠ Execution Authority ≠ Commit Authority . \boxed{
\text{Selected Route}
\neq
\text{Execution Authority}
\neq
\text{Commit Authority}.
} Selected Route = Execution Authority = Commit Authority .
64. Allocation Plan
定義:
A P l a n = ⟨ T a s k s , R e s o u r c e s , B i n d i n g s , C a p a c i t y , C o n s t r a i n t s , O b j e c t i v e s , P o l i c y , P r o v e n a n c e ⟩ . \boxed{
\mathsf{APlan}
=
\left\langle
Tasks,
Resources,
Bindings,
Capacity,
Constraints,
Objectives,
Policy,
Provenance
\right\rangle.
} APlan = ⟨ T a s k s , R eso u r ces , B in d in g s , C a p a c i t y , C o n s t r ain t s , O bj ec t i v es , P o l i cy , P r o v e nan ce ⟩ .
65. Physical Resource Binding
可能綁定:
CPU;
GPU;
VRAM;
RAM;
SSD;
NAS;
remote compute;
browser process;
agent session。
66. Resource Snapshot
allocation 必須基於:
R t s n a p s h o t . R_t^{snapshot}. R t s na p s h o t .
不能基於模糊的「大概有資源」。
67. Frozen Evaluation Context
多 objective 比較必須在:
E e v a l \boxed{
\mathcal E^{eval}
} E e v a l
中固定:
resource snapshot;
objective profile;
units;
policy;
bounded candidate set。
68. Same Frozen Inputs → Same Selection
要求:
Same Frozen Inputs + Same ObjectiveProfile + Same Evaluation Context + Same Policy ⇒ Same Result . \boxed{
\text{Same Frozen Inputs}
+
\text{Same ObjectiveProfile}
+
\text{Same Evaluation Context}
+
\text{Same Policy}
\Rightarrow
\text{Same Result}.
} Same Frozen Inputs + Same ObjectiveProfile + Same Evaluation Context + Same Policy ⇒ Same Result .
69. PNCW Projection Plan
本文定義:
P P l a n t = ⟨ C o n t e x t P l a n , C a r r i e r P l a n , V i s u a l P l a n , R e s o u r c e P l a n , R e v e a l P l a n ⟩ . \boxed{
\mathsf{PPlan}_t
=
\left\langle
ContextPlan,
CarrierPlan,
VisualPlan,
ResourcePlan,
RevealPlan
\right\rangle.
} PPlan t = ⟨ C o n t e x tP l an , C a r r i er P l an , V i s u a l P l an , R eso u r ce P l an , R e v e a l P l an ⟩ .
70. Context Plan
決定:
active objects;
page-in;
working set;
context budget;
query overlay。
71. Carrier Plan
決定:
HDSRC profile;
scale;
relation representation;
materialized tiles;
integrity mode。
72. Visual Plan
決定:
viewport;
panel set;
portals;
visual overlays;
render resolution。
73. Resource Plan
決定:
executor;
memory tier;
IO path;
compute budget。
74. Reveal Plan
決定:
stream;
batch;
atomic artifact;
canvas;
hybrid。
75. Plan Non-Collapse
Context Plan ≠ Carrier Plan ≠ Visual Plan ≠ Reveal Plan . \boxed{
\text{Context Plan}
\neq
\text{Carrier Plan}
\neq
\text{Visual Plan}
\neq
\text{Reveal Plan}.
} Context Plan = Carrier Plan = Visual Plan = Reveal Plan .
76. Hierarchical Planning
可先:
G l o b a l P l a n \mathsf{GlobalPlan} GlobalPlan
再:
D o m a i n P l a n i . \mathsf{DomainPlan}_i. DomainPlan i .
因此:
Planning can be hierarchical . \boxed{
\text{Planning}
\text{ can be hierarchical}.
} Planning can be hierarchical .
77. Global Plan / Local Autonomy
Global plan 管:
budget;
hard constraints;
dependencies;
authority。
Local domain 可以在範圍內自治。
78. Constraint Envelope
定義:
E i c o n s t r a i n t \boxed{
\mathcal E_i^{constraint}
} E i co n s t r ain t
作為 domain 的合法操作 envelope。
79. Local Freedom Within Envelope
只要:
F i ∈ E i c o n s t r a i n t , F_i
\in
\mathcal E_i^{constraint}, F i ∈ E i co n s t r ain t ,
domain 可自行:
refine;
route;
cache;
rearrange local tiles。
80. Authority Envelope
同樣:
E i a u t h o r i t y \boxed{
\mathcal E_i^{authority}
} E i a u t h or i t y
限定:
read;
propose;
execute;
commit。
81. Materialization Budget
對每 domain:
B i m a t . \boxed{
B_i^{mat}.
} B i ma t .
要求:
C o s t m a t ( D i ) ≤ B i m a t . Cost_{mat}(D_i)
\le
B_i^{mat}. C os t ma t ( D i ) ≤ B i ma t .
82. Active Support Budget
∣ A i ( t ) ∣ ≤ B i a c t i v e . \boxed{
|A_i(t)|
\le
B_i^{active}.
} ∣ A i ( t ) ∣ ≤ B i a c t i v e .
83. Resolution Budget
λ i ≤ B i r e s o l u t i o n . \boxed{
\lambda_i
\le
B_i^{resolution}.
} λ i ≤ B i r eso l u t i o n .
84. Visibility Budget
對 observer:
∣ V O ∣ ≤ B O v i e w . \boxed{
|V_O|
\le
B_O^{view}.
} ∣ V O ∣ ≤ B O v i e w .
所以:
observer overload \text{observer overload} observer overload
也可以是 runtime constraint。
85. Multi-Observer Allocation
如果:
O 1 , … , O m , O_1,\ldots,O_m, O 1 , … , O m ,
各自要求不同 projection,
Runtime 不必 duplicate 整個 world。
可共享:
canonical state;
carrier regions;
cached context;
portal state。
86. Shared Substrate / Observer-Specific Projection
Shared Substrate + Observer-Specific Projection . \boxed{
\text{Shared Substrate}
+
\text{Observer-Specific Projection}.
} Shared Substrate + Observer-Specific Projection .
87. Human / AI Resource Competition
Human viewport 與 AI query 可能同時要求同一 GPU / carrier region。
GCM 可做:
Resource Arbitration . \boxed{
\text{Resource Arbitration}.
} Resource Arbitration .
88. Priority
對 request:
r i r_i r i
可有:
P r i o r i t y ( r i ) . Priority(r_i). P r i or i t y ( r i ) .
但 priority 不得繞過 hard authority / safety constraints。
89. Safety Criticality
某些 region:
S a f e t y C r i t i c a l = 1 SafetyCritical=1 S a f e t y C r i t i c a l = 1
可提高:
resolution;
redundancy;
verification;
residency priority。
90. Prediction of Future Use
Materialization Policy 可以包含:
U ^ f u t u r e ( X ) . \hat U_{future}(X). U ^ f u t u r e ( X ) .
例如 HDSRC predictive materialization / Context prefetch。
91. Prediction 不等於 Authority
Predicted Useful ≠ Authorized . \boxed{
\text{Predicted Useful}
\neq
\text{Authorized}.
} Predicted Useful = Authorized .
92. Predictive Materialization / GCM Routing
GCM 可以使用:
C ^ m a t , L ^ , U ^ . \hat C_{mat},
\quad
\hat L,
\quad
\hat U. C ^ ma t , L ^ , U ^ .
但 high uncertainty 時回 oracle / conservative path。
93. Fast / Oracle Selection
S e l e c t M o d e ∈ { F A S T , O R A C L E } . \boxed{
\mathsf{SelectMode}
\in
\{
\mathsf{FAST},
\mathsf{ORACLE}
\}.
} SelectMode ∈ { FAST , ORACLE } .
94. Selection Regret
R e g r e t = C ( ρ s e l e c t e d ) − C ( ρ b e s t − k n o w n ) . \boxed{
Regret
=
C(\rho_{selected})
-
C(\rho_{best-known}).
} R e g r e t = C ( ρ se l ec t e d ) − C ( ρ b es t − k n o w n ) .
95. System Regret / Prediction Error Non-Collapse
如果 predictor miss,但 fallback 防止 bad route:
Prediction Error ≠ System Regret . \boxed{
\text{Prediction Error}
\neq
\text{System Regret}.
} Prediction Error = System Regret .
96. Replanning
當:
resource stale;
context change;
carrier debt;
task change;
portal loss;
observer change;
可以:
R e p l a n . \boxed{
\mathsf{Replan}.
} Replan .
97. Replanning / Commit Non-Collapse
新 plan:
P 1 P_1 P 1
不自動替代:
P 0 . P_0. P 0 .
需要 lifecycle / authority。
98. Stale Plan
如果:
R e s o u r c e S n a p s h o t ( P 0 ) ≠ R t , ResourceSnapshot(P_0)
\neq
R_t, R eso u r ce S na p s h o t ( P 0 ) = R t ,
則:
P 0 = S T A L E . \boxed{
P_0
=
\mathsf{STALE}.
} P 0 = STALE .
99. Plan Lineage
P 0 → P 1 → P 2 P_0
\rightarrow
P_1
\rightarrow
P_2 P 0 → P 1 → P 2
必須記錄 supersession / cause。
100. Projection Schedule
定義:
Σ P N C W = ( O b s e r v e W o r l d , B u i l d P l a n , A l l o c a t e , P r o j e c t C o n t e x t , F r e e z e C a r r i e r , M a t e r i a l i z e C a r r i e r , B u i l d V i s u a l W o r l d , V e r i f y , R e v e a l ) . \boxed{
\Sigma_{PNCW}
=
(
\mathsf{ObserveWorld},
\mathsf{BuildPlan},
\mathsf{Allocate},
\mathsf{ProjectContext},
\mathsf{FreezeCarrier},
\mathsf{MaterializeCarrier},
\mathsf{BuildVisualWorld},
\mathsf{Verify},
\mathsf{Reveal}
).
} Σ P N C W = ( ObserveWorld , BuildPlan , Allocate , ProjectContext , FreezeCarrier , MaterializeCarrier , BuildVisualWorld , Verify , Reveal ) .
101. Schedule Non-Commutativity
一般:
A ∘ B ≠ B ∘ A . A\circ B
\neq
B\circ A. A ∘ B = B ∘ A .
例如先 Reveal 再 Verify 是非法的。
102. No Premature Reveal
R e v e a l ≺ V e r i f y is forbidden for authoritative mode . \boxed{
\mathsf{Reveal}
\prec
\mathsf{Verify}
\quad
\text{is forbidden for authoritative mode}.
} Reveal ≺ Verify is forbidden for authoritative mode .
103. No Premature Full Materialization
同樣:
Materialize Everything First \boxed{
\text{Materialize Everything First}
} Materialize Everything First
不應成為 default。
104. Lazy Materialization
優先:
Plan → Materialize Required Support → Expand on Demand . \boxed{
\text{Plan}
\rightarrow
\text{Materialize Required Support}
\rightarrow
\text{Expand on Demand}.
} Plan → Materialize Required Support → Expand on Demand .
105. Speculative Materialization
Runtime 可以 speculative prefetch:
X s p e c . X^{spec}. X s p ec .
但:
Speculative Residency ≠ Semantic Activation . \boxed{
\text{Speculative Residency}
\neq
\text{Semantic Activation}.
} Speculative Residency = Semantic Activation .
106. Eviction
如果:
C o s t k e e p ( X ) > E x p e c t e d B e n e f i t ( X ) , Cost_{keep}(X)
>
ExpectedBenefit(X), C os t k ee p ( X ) > E x p ec t e d B e n e f i t ( X ) ,
可以:
E v i c t ( X ) . \boxed{
\mathsf{Evict}(X).
} Evict ( X ) .
107. Evict / Forget Non-Collapse
Evict ≠ Forget . \boxed{
\text{Evict}
\neq
\text{Forget}.
} Evict = Forget .
evict 只改 residency。
108. Archive
如果長期不 active:
A r c h i v e ( X ) \boxed{
\mathsf{Archive}(X)
} Archive ( X )
保留重建所需資訊。
109. Archive / Delete Non-Collapse
Archive ≠ Delete . \boxed{
\text{Archive}
\neq
\text{Delete}.
} Archive = Delete .
110. PNCW Materialization Ledger
記錄:
e M = ⟨ O b j e c t , O l d S t a t e , N e w S t a t e , R e a s o n , C o s t , P o l i c y , T i m e ⟩ . \boxed{
e_M
=
\left\langle
Object,
OldState,
NewState,
Reason,
Cost,
Policy,
Time
\right\rangle.
} e M = ⟨ O bj ec t , O l d S t a t e , N e w S t a t e , R e a so n , C os t , P o l i cy , T im e ⟩ .
111. Resolution Ledger
e λ = ⟨ D o m a i n , O l d R e s o l u t i o n , N e w R e s o l u t i o n , C a u s e , R i s k , T i m e ⟩ . \boxed{
e_\lambda
=
\left\langle
Domain,
OldResolution,
NewResolution,
Cause,
Risk,
Time
\right\rangle.
} e λ = ⟨ D o main , O l d R eso l u t i o n , N e w R eso l u t i o n , C a u se , R i s k , T im e ⟩ .
112. Projection Routing Ledger
e ρ = ⟨ R e q u e s t , C a n d i d a t e s , F e a s i b l e S e t , F r o n t i e r , P o l i c y , S e l e c t e d R o u t e , E v i d e n c e ⟩ . \boxed{
e_\rho
=
\left\langle
Request,
Candidates,
FeasibleSet,
Frontier,
Policy,
SelectedRoute,
Evidence
\right\rangle.
} e ρ = ⟨ R e q u es t , C an d i d a t es , F e a s ib l e S e t , F r o n t i er , P o l i cy , S e l ec t e d R o u t e , E v i d e n ce ⟩ .
113. History / Current State Non-Collapse
即使:
C u r r e n t P l a n = P 2 , CurrentPlan=P_2, C u r r e n tP l an = P 2 ,
仍保留:
P 0 , P 1 . P_0,P_1. P 0 , P 1 .
所以:
Current Route ≠ Routing History . \boxed{
\text{Current Route}
\neq
\text{Routing History}.
} Current Route = Routing History .
114. GCM × Context VM
GCM 決定:
budget;
resource;
representation;
global dependency。
Context MMU 決定:
page;
working set;
active context。
因此:
GCM + Context MMU = Cognitive Materialization Governance . \boxed{
\text{GCM}
+
\text{Context MMU}
=
\text{Cognitive Materialization Governance}.
} GCM + Context MMU = Cognitive Materialization Governance .
115. GCM × SPET/HDSRC
GCM 決定:
carrier candidate;
scale;
materialization budget;
resource route。
SPET/HDSRC 決定:
frame validity;
carrier semantics;
native operator correctness。
116. GCM × MRMIC/NVCL
GCM 決定:
viewport budget;
portal priority;
render resources;
active panels。
MRMIC/NVCL 提供:
visual world;
interaction;
resource portals;
observer projection。
117. PNCW Global Runtime Architecture
W t → G C M P l a n P P l a n t → C o n t e x t M M U C q a c t i v e → S P E T / H D S R C P k → M R M I C / N V C L V q , k → R e v e a l P o l i c y U q , k . \boxed{
\begin{aligned}
W_t
&\xrightarrow{\mathsf{GCMPlan}}
\mathsf{PPlan}_t
\\
&\xrightarrow{\mathsf{ContextMMU}}
C_q^{active}
\\
&\xrightarrow{\mathsf{SPET/HDSRC}}
P_k
\\
&\xrightarrow{\mathsf{MRMIC/NVCL}}
V_{q,k}
\\
&\xrightarrow{\mathsf{RevealPolicy}}
U_{q,k}.
\end{aligned}
} W t GCMPlan PPlan t ContextMMU C q a c t i v e SPET/HDSRC P k MRMIC/NVCL V q , k RevealPolicy U q , k .
118. Global Runtime Invariant
No layer may silently assume authority owned by another layer . \boxed{
\text{No layer may silently assume authority owned by another layer}.
} No layer may silently assume authority owned by another layer .
119. Planner / Executor Non-Collapse
Planner ≠ Executor . \boxed{
\text{Planner}
\neq
\text{Executor}.
} Planner = Executor .
120. Executor / Committer Non-Collapse
Executor ≠ Committer . \boxed{
\text{Executor}
\neq
\text{Committer}.
} Executor = Committer .
121. Observer / Planner Non-Collapse
Observer demand 可以影響 materialization,但:
Observer Request ≠ Global Plan Authority . \boxed{
\text{Observer Request}
\neq
\text{Global Plan Authority}.
} Observer Request = Global Plan Authority .
122. AI Proposal / Deterministic Runtime Non-Collapse
未來 AI 可提議:
P o l i c y , O b j e c t i v e s , R e p r e s e n t a t i o n , R e s o l u t i o n . Policy,
Objectives,
Representation,
Resolution. P o l i cy , O bj ec t i v es , R e p r ese n t a t i o n , R eso l u t i o n .
但:
AI Proposal ≠ Deterministic Runtime Decision . \boxed{
\text{AI Proposal}
\neq
\text{Deterministic Runtime Decision}.
} AI Proposal = Deterministic Runtime Decision .
123. Phase C / Future AI Orchestration
只有 deterministic substrate 成熟後,才值得研究:
AI-assisted Compute Orchestrator . \boxed{
\text{AI-assisted Compute Orchestrator}.
} AI-assisted Compute Orchestrator .
它應提出 plan,不直接繞過 hard constraints。
124. PNCW Global Objective
定義:
J = w L L + w M M + w I I + w E E + w R R + w Q Q + w V V . \boxed{
J
=
w_L L
+
w_M M
+
w_I I
+
w_E E
+
w_R R
+
w_Q Q
+
w_V V.
} J = w L L + w M M + w I I + w E E + w R R + w Q Q + w V V .
但 scalar objective 只是一種 policy,不是 universal truth。
125. Multi-Objective Default
因此更穩妥的是:
Feasibility → Pareto → Explicit Policy . \boxed{
\text{Feasibility}
\rightarrow
\text{Pareto}
\rightarrow
\text{Explicit Policy}.
} Feasibility → Pareto → Explicit Policy .
126. Proposition 1 — Global Coherence Does Not Require Global Materialization
只要 canonical dependency / constraints 可被 Runtime 管理,observer/task 所需 support 有效,則:
Global Coherence ⇏ Full Materialization . \boxed{
\text{Global Coherence}
\not\Rightarrow
\text{Full Materialization}.
} Global Coherence ⇒ Full Materialization .
127. Proposition 2 — Different Domains May Use Different Resolutions
存在:
D i , D j D_i,D_j D i , D j
使:
λ i ≠ λ j \lambda_i\neq\lambda_j λ i = λ j
而 global constraints 仍成立。
128. Proposition 3 — Observer Resolution Does Not Determine Compute Resolution
存在:
λ o b s e r v e ≠ λ c o m p u t e . \lambda^{observe}
\neq
\lambda^{compute}. λ o b ser v e = λ co m p u t e .
因此 UI zoom 不應直接綁定 solver fidelity。
129. Proposition 4 — Projection Route Selection Can Be Deterministic Without Being Unique in Objective Space
同一 feasible frontier 可有多個 nondominated routes,但 explicit policy 可 deterministically 選一個。
因此:
Multiple Valid Alternatives ≠ Nondeterministic System . \boxed{
\text{Multiple Valid Alternatives}
\neq
\text{Nondeterministic System}.
} Multiple Valid Alternatives = Nondeterministic System .
130. Proposition 5 — Materialization State Can Change Without Semantic State Change
如果:
X : R e s i d e n t → A r c h i v e d X:
Resident
\rightarrow
Archived X : R es i d e n t → A r c hi v e d
但 canonical content 不變,
則:
Δ R e s i d e n c y ⇏ Δ S e m a n t i c s . \boxed{
\Delta Residency
\not\Rightarrow
\Delta Semantics.
} Δ R es i d e n cy ⇒ Δ S e man t i cs .
131. Proposition 6 — Visibility Can Change Without Recomputing the World
不同 observer projection / viewport 可:
V 1 → V 2 V_1
\rightarrow
V_2 V 1 → V 2
同時:
W t W_t W t
不變。
132. PNCW Paper 05 規範 v0.1
PNCW-G1 — Global / Everywhere Non-Collapse
Global 不得被解釋為 all-to-all/full materialization。
PNCW-G2 — Dependency / Materialization Separation
存在 dependency 不等於同時展開全部 detail。
PNCW-G3 — Finite Active Realization
任一真實 Runtime 的 active support 必須 finite / bounded。
PNCW-G4 — Unbounded / Infinite Active Separation
無界可展開不等於無限同時活動。
PNCW-G5 — Resolution Typing
compute / observe / carrier / render resolution 必須分開。
PNCW-G6 — Projection / Materialization Separation
logical projection 不等於 physical residency。
PNCW-G7 — Constraint Before Optimization
所有 route selection 必須先通過 hard constraints。
PNCW-G8 — Feasible / Preferred / Optimal Separation
不得把 policy-selected route 宣稱 universal optimum。
PNCW-G9 — Plan / Execution / Commit Separation
選中 plan 不等於有 execution / commit authority。
PNCW-G10 — Explicit Materialization Policy
active realization 必須有可解釋 policy / evidence。
PNCW-G11 — Replanning Lineage
stale / superseded plan 必須保存 lineage。
PNCW-G12 — Observer Relativity
不同 observer 可以合法取得不同 projection / resolution。
133. 對「一口氣看到」的真正全域解釋
到了 Paper 05,「一口氣看到」可以被更精確地描述成:
Global world remains coherent, \boxed{
\text{Global world remains coherent,}
} Global world remains coherent,
only task-relevant structures are materialized, \boxed{
\text{only task-relevant structures are materialized,}
} only task-relevant structures are materialized,
and the observer receives a logically complete projection of that active world . \boxed{
\text{and the observer receives a logically complete projection of that active world}.
} and the observer receives a logically complete projection of that active world .
134. 它不是全世界一次展開
不是:
W t → Render Everything . \boxed{
W_t
\rightarrow
\text{Render Everything}.
} W t → Render Everything .
而是:
W t → P l a n → A t → M t → V O . \boxed{
W_t
\rightarrow
\mathsf{Plan}
\rightarrow
A_t
\rightarrow
M_t
\rightarrow
V_O.
} W t → Plan → A t → M t → V O .
135. Active / Materialized / Visible Chain
W t ⊇ A t ⊇ M t t a s k ⊇ V O v i s i b l e \boxed{
W_t
\supseteq
A_t
\supseteq
M_t^{task}
\supseteq
V_O^{visible}
} W t ⊇ A t ⊇ M t t a s k ⊇ V O v i s ib l e
注意這只是常見 inclusion,不是所有架構必須嚴格如此。
136. Global Structure First
Observer 可以先拿到:
manifest;
domain graph;
active regions;
panel identities;
unresolved areas;
projection routes。
再按需 materialize detail。
137. First Useful Global View
因此新的 latency 指標:
L g l o b a l − v i e w . \boxed{
L_{global-view}.
} L g l o ba l − v i e w .
它不同於:
L f i r s t − t o k e n . L_{first-token}. L f i r s t − t o k e n .
138. First Actionable State
另一個指標:
L f i r s t − a c t i o n a b l e . \boxed{
L_{first-actionable}.
} L f i r s t − a c t i o nab l e .
表示 observer 何時能對整體 artifact 做有意義操作。
139. Sequence vs Projection Benchmark
Sequence mode 優化:
L f i r s t − t o k e n . L_{first-token}. L f i r s t − t o k e n .
Projection mode可能優化:
L g l o b a l − v i e w , L f i r s t − a c t i o n a b l e . L_{global-view},
L_{first-actionable}. L g l o ba l − v i e w , L f i r s t − a c t i o nab l e .
兩者不是同一產品目標。
140. PNCW Vertical Slice — Global Phase
完整最小實驗:
建立 10 4 ∼ 10 5 10^4\sim10^5 1 0 4 ∼ 1 0 5 node canonical world;
建多個 task/query;
GCM 建 candidate routes;
constraints filter;
Pareto frontier;
explicit policy select;
Context MMU materialize active support;
SPET Freeze;
HDSRC materialize selected carrier regions;
MRMIC 建 visual world;
Atomic Reveal;
observer zoom / query;
dynamic refinement;
resource pressure 觸發 eviction;
task change 觸發 replanning。
141. Metrics
M G = ⟨ A c t i v e F r a c t i o n , M a t e r i a l i z e d F r a c t i o n , G l o b a l L a t e n c y , F i r s t G l o b a l V i e w , F i r s t A c t i o n a b l e , P e a k M e m o r y , I O , R e p l a n C o u n t , M i g r a t i o n C o s t , S e l e c t i o n R e g r e t , S t a l e R e j e c t R a t e , R e u s e R a t e ⟩ . \boxed{
\mathbf M_G
=
\left\langle
ActiveFraction,
MaterializedFraction,
GlobalLatency,
FirstGlobalView,
FirstActionable,
PeakMemory,
IO,
ReplanCount,
MigrationCost,
SelectionRegret,
StaleRejectRate,
ReuseRate
\right\rangle.
} M G = ⟨ A c t i v e F r a c t i o n , M a t er ia l i z e d F r a c t i o n , Gl o ba l L a t e n cy , F i r s tGl o ba l V i e w , F i r s t A c t i o nab l e , P e ak M e m or y , I O , R e pl an C o u n t , M i g r a t i o n C os t , S e l ec t i o n R e g r e t , S t a l e R e j ec tR a t e , R e u se R a t e ⟩ .
142. Resolution Metrics
M λ = ⟨ C o m p u t e R e s , C a r r i e r R e s , O b s e r v e R e s , R e n d e r R e s , R e f i n e m e n t C o u n t , D e R e f i n e m e n t C o u n t ⟩ . \boxed{
\mathbf M_\lambda
=
\left\langle
ComputeRes,
CarrierRes,
ObserveRes,
RenderRes,
RefinementCount,
DeRefinementCount
\right\rangle.
} M λ = ⟨ C o m p u t e R es , C a r r i er R es , O b ser v e R es , R e n d er R es , R e f in e m e n tC o u n t , D e R e f in e m e n tC o u n t ⟩ .
143. Materialization Metrics
M M = ⟨ P a g e I n , P a g e O u t , C a c h e H i t , P r e f e t c h H i t , E v i c t i o n , A r c h i v e , R e c o n s t r u c t , R e s i d e n t B y t e s ⟩ . \boxed{
\mathbf M_M
=
\left\langle
PageIn,
PageOut,
CacheHit,
PrefetchHit,
Eviction,
Archive,
Reconstruct,
ResidentBytes
\right\rangle.
} M M = ⟨ P a g e I n , P a g e O u t , C a c h eH i t , P r e f e t c h H i t , E v i c t i o n , A r c hi v e , R eco n s t r u c t , R es i d e n tB y t es ⟩ .
144. Failure Conditions
若:
active support bounded 但 runtime 還每 tick scan 全 world;
lazy materialization 導致 page-fault storm;
replan churn;
resolution oscillation;
observer view inconsistent;
route predictor distribution shift 無 fallback;
migration cost 超過收益;
global constraint 被 local optimization 破壞;
則 PNCW global-runtime claim 必須降低。
145. Bounded Semantics / Bounded Runtime Cost Non-Collapse
即使:
∣ A t ∣ ≤ B A , |A_t|\le B_A, ∣ A t ∣ ≤ B A ,
如果每 tick 還:
scan dormant;
clone full world;
render all;
rehash all;
仍不能說 cost bounded。
所以:
Bounded Active Semantics ≠ Bounded Runtime Cost . \boxed{
\text{Bounded Active Semantics}
\neq
\text{Bounded Runtime Cost}.
} Bounded Active Semantics = Bounded Runtime Cost .
146. Runtime Closure Requirement
真正 bounded runtime 還要:
Bounded Active + Bounded Scan + Bounded Materialization + Bounded Commit . \boxed{
\text{Bounded Active}
+
\text{Bounded Scan}
+
\text{Bounded Materialization}
+
\text{Bounded Commit}.
} Bounded Active + Bounded Scan + Bounded Materialization + Bounded Commit .
147. 與 Paper 06 的接口
Paper 05 回答:
誰決定什麼被算、被物化、被投影、被顯示?
下一篇 Paper 06 將整個系列收束成:
一個真正可實作的 Non-Sequential AI Output Architecture 應如何把 GCM、Context VM、SPET/HDSRC、MRMIC/NVCL、Visibility Commit 變成一條完整 runtime pipeline?
即:
World → Plan → Active Context → Carrier → Canvas → Atomic Visible Result . \boxed{
\text{World}
\rightarrow
\text{Plan}
\rightarrow
\text{Active Context}
\rightarrow
\text{Carrier}
\rightarrow
\text{Canvas}
\rightarrow
\text{Atomic Visible Result}.
} World → Plan → Active Context → Carrier → Canvas → Atomic Visible Result .
148. 系列位置
P 00 : Projection-Native World Foundations P 01 : Visibility / Atomic Reveal P 02 : Virtual Context Projection P 03 : Stable High-D Projection Carrier P 04 : Visual Computational Canvas P 05 : Global Compute / Local Materialization P 06 : Non-Sequential AI Output Architecture \boxed{
\begin{aligned}
P00 &: \text{Projection-Native World Foundations}\\
P01 &: \text{Visibility / Atomic Reveal}\\
P02 &: \text{Virtual Context Projection}\\
P03 &: \text{Stable High-D Projection Carrier}\\
P04 &: \text{Visual Computational Canvas}\\
P05 &: \text{Global Compute / Local Materialization}\\
P06 &: \text{Non-Sequential AI Output Architecture}
\end{aligned}
} P 00 P 01 P 02 P 03 P 04 P 05 P 06 : Projection-Native World Foundations : Visibility / Atomic Reveal : Virtual Context Projection : Stable High-D Projection Carrier : Visual Computational Canvas : Global Compute / Local Materialization : Non-Sequential AI Output Architecture
149. 結論
本文把 GCM 正式放回 PNCW 的最上層 runtime governance。
它的核心不是:
把所有東西都算完再顯示。
而是:
Global Coherence + Finite Active Realization + Selective Materialization + Observer-Relative Projection . \boxed{
\text{Global Coherence}
+
\text{Finite Active Realization}
+
\text{Selective Materialization}
+
\text{Observer-Relative Projection}.
} Global Coherence + Finite Active Realization + Selective Materialization + Observer-Relative Projection .
因此:
Global Computation ≠ Global Simultaneous Materialization . \boxed{
\text{Global Computation}
\neq
\text{Global Simultaneous Materialization}.
} Global Computation = Global Simultaneous Materialization .
一個世界可以保持全域一致,但:
只有少數 domain active;
只有部分 active domain materialized;
只有部分 materialized state 進 carrier;
只有部分 carrier region 進 viewport;
不同 observer 取得不同 projection。
本文也正式建立:
λ c o m p u t e ≠ λ o b s e r v e ≠ λ c a r r i e r ≠ λ r e n d e r . \boxed{
\lambda^{compute}
\neq
\lambda^{observe}
\neq
\lambda^{carrier}
\neq
\lambda^{render}.
} λ co m p u t e = λ o b ser v e = λ c a r r i er = λ r e n d er .
因此「看得更細」不自動要求「算得更細」,「算得更細」也不自動要求「全部顯示」。
透過 GCM Phase B 類型的 deterministic allocation,PNCW 可以進一步將:
Constraint → Feasible Routes → Pareto Frontier → Explicit Policy Selection \boxed{
\text{Constraint}
\rightarrow
\text{Feasible Routes}
\rightarrow
\text{Pareto Frontier}
\rightarrow
\text{Explicit Policy Selection}
} Constraint → Feasible Routes → Pareto Frontier → Explicit Policy Selection
用於 projection/materialization planning。
這使 PNCW 不只是一套 visual metaphor,而開始具有一個真正可實作的 global runtime selection semantics。
本文最終提出:
Compute Globally, Materialize Selectively, Observe Relatively . \boxed{
\text{Compute Globally, Materialize Selectively, Observe Relatively}.
} Compute Globally, Materialize Selectively, Observe Relatively .
這句話正是「一口氣看到」在 PNCW 中最準確的全域版本。
不是讓整個宇宙同時 render。
而是:
全域世界保持一致;Runtime 只把現在真正有用的部分算到適合的解析度、物化成適合的 representation,再讓 observer 一次取得完整而可操作的當前世界投影。
內部理論與工程血統
本文主要承接:
PNCW Paper 00–04;
Global Computation Methodology Series 00–06;
GCM Paper 04 — Computation ≠ Observation / Selective Materialization;
GCM Paper 05 — Finite Active Realization and Unbounded Extensibility;
GCM Phase A — Reference Runtime MVP;
GCM Phase B — Deterministic Compute Allocator;
B1 — Intent & Deterministic Planning Contract;
B2 — Resource Snapshot & Binding;
B3 — Capacity-Constrained Allocation;
B4 — Multi-Objective / Pareto Allocation;
AI Context Virtual Memory / Context MMU;
SPET / HDSRC carrier routing;
MRMIC/NVCL visual computational surface。
本文不宣稱 GCM Phase B 全部完成;目前以已驗證的 Phase A 與已完成之 B1–B4 engineering evidence 作為可引用工程基礎,B5 lifecycle/replanning、B6 final conformance、Phase C AI-assisted orchestration 與 Phase D native compute fabric 仍屬後續階段。