PNCW Paper 07
投影原生軟體時空:從狀態切片到因果—時間世界投影
Projection-Native Software Spacetime:
From State Slices to Causal-Temporal World Projection
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 07 / Causal-Temporal and Multi-Temporal Projection Layer 依賴:PNCW Paper 00–06、Software Spacetime Theory Series 01–04 主要來源接口:Software Spacetime、Multi-Temporal Computing、Software Causal Topology、Fractal AI Spacetime Governance 作者: Neo.K機構: EveMissLab/一言諾科技有限公司
摘要
Projection-Native Computational World(PNCW)Paper 00–06 已建立一條從 canonical world、active context、stable carrier 到 visual computational surface 與 atomic reveal 的統一架構:
W t → C q , t a c t i v e → E k → P k → V q , k → U q , k . \boxed{
W_t
\rightarrow
C_{q,t}^{active}
\rightarrow
\mathcal E_k
\rightarrow
P_k
\rightarrow
V_{q,k}
\rightarrow
U_{q,k}.
} W t → C q , t a c t i v e → E k → P k → V q , k → U q , k .
這條鏈成立,但其中的 canonical source 常被簡寫為單一時間索引下的世界狀態:
W t . W_t. W t .
本文指出,這個形式對許多 AI-native runtime、simulation、agent workflow、compiler pipeline、database transaction、game world、replay system 與分支式計算而言仍過度簡化。真正需要被投影的來源往往不是「某個瞬間的高維 state」,而是具有:
local temporal domains;
software worldlines;
event histories;
causal partial orders;
replay / branch structures;
cross-domain synchronization;
observer-relative temporal resolution;
的 Software Spacetime(軟體時空) 。
本文因此將 PNCW 的 canonical source 從單一 state slice 提升為:
S = ( D , G C , G R , R , P , H ) , \boxed{
\mathfrak S
=
(
\mathcal D,
G_C,
G_R,
\mathcal R,
\mathcal P,
\mathcal H
),
} S = ( D , G C , G R , R , P , H ) ,
其中每個 software spacetime domain:
D i = ( X i , T i , E i , C i , O i , R i , A i ) \boxed{
D_i
=
(
X_i,
T_i,
E_i,
C_i,
O_i,
R_i,
A_i
)
} D i = ( X i , T i , E i , C i , O i , R i , A i )
具有自己的狀態空間 X i X_i X i 、局部時間結構 T i T_i T i 、事件集合 E i E_i E i 、因果關係 C i C_i C i 、observer / projection structure O i O_i O i 、resources R i R_i R i 與 control capabilities A i A_i A i 。
本文引入 Spacetime Projection Operator :
Π q , O S T : S → Y q , O S T , \boxed{
\Pi^{ST}_{q,O}
:
\mathfrak S
\rightarrow
Y^{ST}_{q,O},
} Π q , O S T : S → Y q , O S T ,
更完整地:
Π q , O , σ , Λ , Δ T , K , B , π S T : S → Y S T . \boxed{
\Pi^{ST}_{
q,
O,
\sigma,
\Lambda,
\Delta T,
\mathcal K,
\mathcal B,
\pi
}
:
\mathfrak S
\rightarrow
Y^{ST}.
} Π q , O , σ , Λ , Δ T , K , B , π S T : S → Y S T .
其中:
(q):task / query;
(O):observer;
σ \sigma σ :domain / object scope;
Λ \Lambda Λ :state / temporal / causal / observation resolution policy;
Δ T \Delta T Δ T :temporal window / multi-time selection;
K \mathcal K K :causal cone / dependency region;
B \mathcal B B :branch / replay selection;
π \pi π :projection frame / presentation contract。
本文的核心新命題是:
State Projection ⊂ Spacetime Projection . \boxed{
\text{State Projection}
\subset
\text{Spacetime Projection}.
} State Projection ⊂ Spacetime Projection .
單一時刻的 state view 只是 spacetime projection 的退化特例。更一般的 PNCW output 可以一次投影:
一段 worldline;
一個 causal cone;
一組互不比較的 parallel events;
多個 local temporal domains;
replay 與 live state 的差異;
speculative branch ensemble;
critical path 與 off-critical work;
branch point、join point、commit boundary。
本文進一步採用 Software Causal Topology 的核心區分:
Serialization Order ≠ Necessary Causal Order . \boxed{
\text{Serialization Order}
\neq
\text{Necessary Causal Order}.
} Serialization Order = Necessary Causal Order .
token stream、event ledger sequence、UI list order 或 timestamp sorting 都可以是某種合法 linearization,但不應被誤認為 world 的 causal ontology。若原始 computation 是一個 partial order:
( E , ≺ ) , (E,\prec), ( E , ≺ ) ,
則一條 sequence:
L ( E , ≺ ) = ( e π ( 1 ) , … , e π ( n ) ) L(E,\prec)
=
(e_{\pi(1)},\ldots,e_{\pi(n)}) L ( E , ≺ ) = ( e π ( 1 ) , … , e π ( n ) )
只是其一個 total-order extension / presentation,而不是因果結構本身。
因此,「AI 為什麼一定要一個 token 一個 token輸出」的問題,在本文中被提升為:
High-Dimensional + Multi-Temporal + Causal + Branched + Cross-Structural World ⇏ One Linear Observation Timeline . \boxed{
\begin{aligned}
&\text{High-Dimensional}\\
+&\text{Multi-Temporal}\\
+&\text{Causal}\\
+&\text{Branched}\\
+&\text{Cross-Structural World}
\\
&\not\Rightarrow
\text{One Linear Observation Timeline}.
\end{aligned}
} + + + + High-Dimensional Multi-Temporal Causal Branched Cross-Structural World ⇒ One Linear Observation Timeline .
本文最終提出:
Projection-Native Spacetime = Selective State + Selective Time + Selective Causality + Selective Branches + Observer-Relative Resolution . \boxed{
\text{Projection-Native Spacetime}
=
\text{Selective State}
+
\text{Selective Time}
+
\text{Selective Causality}
+
\text{Selective Branches}
+
\text{Observer-Relative Resolution}.
} Projection-Native Spacetime = Selective State + Selective Time + Selective Causality + Selective Branches + Observer-Relative Resolution .
這使 PNCW 從「非序列狀態輸出架構」進一步升級為「因果—時間世界投影架構」。
關鍵詞: Projection-Native Software Spacetime、Software Worldline、Multi-Temporal Computing、Causal Topology、Branch Projection、Replay、Partial Order、Temporal Projection、PNCW
0. 研究目的與邊界
本文只處理:
Software Spacetime → Projection-Native Observation . \boxed{
\text{Software Spacetime}
\rightarrow
\text{Projection-Native Observation}.
} Software Spacetime → Projection-Native Observation .
本文暫不處理下一篇 Paper 08 的:
CSPMF cross-structural perception;
persistent perceptual memory;
APR attention;
PHOSPHOR actuation;
full perception–action loop。
因此 Paper 07 的任務非常單純:
把 PNCW 的 source ontology 從 state slice 擴展成 causal-temporal software world。
1. PNCW 既有形式的限制
PNCW 前六篇常寫:
W t . W_t. W t .
這在形式上方便,也適合表示:
時間 (t) 的 canonical world state。
但它容易讓人誤解成:
World = Snapshot . \boxed{
\text{World}
=
\text{Snapshot}.
} World = Snapshot .
本文否定這個等同。
2. World 不等於 Snapshot
更一般地:
World = State + Evolution + Events + Causality + History + Branches . \boxed{
\text{World}
=
\text{State}
+
\text{Evolution}
+
\text{Events}
+
\text{Causality}
+
\text{History}
+
\text{Branches}.
} World = State + Evolution + Events + Causality + History + Branches .
Snapshot 只是 World 的一個截面。
3. Software Spacetime Domain
本文採用:
D i = ( X i , T i , E i , C i , O i , R i , A i ) . \boxed{
D_i
=
(
X_i,
T_i,
E_i,
C_i,
O_i,
R_i,
A_i
).
} D i = ( X i , T i , E i , C i , O i , R i , A i ) .
其中:
X i X_i X i :state space;
T i T_i T i :local temporal structure;
E i E_i E i :events;
C i C_i C i :causal relations;
O i O_i O i :observer / projection structure;
R i R_i R i :resources;
A i A_i A i :control capabilities。
4. Software Spacetime 不等於 Physical Spacetime
Software Spacetime ≠ Physical Spacetime . \boxed{
\text{Software Spacetime}
\neq
\text{Physical Spacetime}.
} Software Spacetime = Physical Spacetime .
本文只借用「spacetime」表示:
state、local time、causal relations、observation 與 worldline 被統一納入同一 runtime abstraction。
不搬用 Lorentz geometry,也不宣稱軟體服從相對論物理定律。
5. Global Software Spacetime
令:
S = ( D , G C , G R , R , P , H ) . \boxed{
\mathfrak S
=
(
\mathcal D,
G_C,
G_R,
\mathcal R,
\mathcal P,
\mathcal H
).
} S = ( D , G C , G R , R , P , H ) .
其中:
D = { D 1 , … , D n } \mathcal D=\{D_1,\ldots,D_n\} D = { D 1 , … , D n } :software spacetime domains;
G C G_C G C :cross-domain causal / dependency graph;
G R G_R G R :resource contention / mapping graph;
R \mathcal R R :shared physical / virtual resources;
P \mathcal P P :governance policies;
H \mathcal H H :history / branch / replay lineage。
6. State Slice
對某 domain D i D_i D i ,在 local time t i t_i t i :
S i ( t i ) ∈ X i . \boxed{
S_i(t_i)
\in
X_i.
} S i ( t i ) ∈ X i .
這只是一個 state slice。
7. Software Worldline
對可識別 software entity (x):
γ x : T i → X i . \boxed{
\gamma_x:
T_i
\rightarrow
X_i.
} γ x : T i → X i .
其中:
γ x ( t ) \gamma_x(t) γ x ( t )
表示 entity 在 local time (t) 的 state。
8. Worldline 的工程含義
worldline 可以表示:
process lifecycle;
thread execution;
Agent task;
game NPC;
database transaction;
compiler job;
simulation object;
virtual machine;
workflow branch。
9. Worldline 不等於單一 Log Sequence
一條 execution log:
L x = ( l 1 , … , l m ) L_x
=
(l_1,\ldots,l_m) L x = ( l 1 , … , l m )
可以記錄 worldline 的 observations。
但:
Worldline ≠ Log Sequence . \boxed{
\text{Worldline}
\neq
\text{Log Sequence}.
} Worldline = Log Sequence .
因為 log 可以:
sparse;
sampled;
reordered;
aggregated;
partial。
10. State / Observation Non-Collapse
對 observer:
O i = ( F i , Z i , Q i , L i ) , O_i
=
(
F_i,
Z_i,
Q_i,
L_i
), O i = ( F i , Z i , Q i , L i ) ,
其中:
F i F_i F i :focus;
Z i Z_i Z i :zoom;
Q i Q_i Q i :resolution;
L i L_i L i :enabled semantic layers。
定義:
Φ i : ( S i , C T S i , O i ) → V i . \boxed{
\Phi_i:
(
S_i,
CTS_i,
O_i
)
\rightarrow
V_i.
} Φ i : ( S i , C T S i , O i ) → V i .
所以:
State ≠ Observation . \boxed{
\text{State}
\neq
\text{Observation}.
} State = Observation .
11. Not Observed 不等於 Not Existing
Not Observed ≠ Not Existing . \boxed{
\text{Not Observed}
\neq
\text{Not Existing}.
} Not Observed = Not Existing .
這與 PNCW 的 selective materialization / selective observation 完全相容。
12. State Projection 是退化特例
如果只選:
Δ T = { t ⋆ } , \Delta T
=
\{t^\star\}, Δ T = { t ⋆ } ,
只選一個 branch:
B = { b ⋆ } , \mathcal B
=
\{b^\star\}, B = { b ⋆ } ,
且 causal scope 只包含當前 state 所需 relations,
則:
Π S T ( S ) \Pi^{ST}(\mathfrak S) Π S T ( S )
退化成:
Π ( W t ⋆ ) . \Pi(W_{t^\star}). Π ( W t ⋆ ) .
因此:
State Projection ⊂ Spacetime Projection . \boxed{
\text{State Projection}
\subset
\text{Spacetime Projection}.
} State Projection ⊂ Spacetime Projection .
13. Local Time
對 domain D i D_i D i ,local time:
t i ( τ ) = b i + ∫ 0 τ α i ( s ) d s . \boxed{
t_i(\tau)
=
b_i
+
\int_0^\tau
\alpha_i(s)\,ds.
} t i ( τ ) = b i + ∫ 0 τ α i ( s ) d s .
其中:
τ \tau τ :reference / physical time;
b i b_i b i :offset;
α i \alpha_i α i :temporal rate。
14. Temporal Rate / Compute Rate Non-Collapse
α i ≠ physical compute speed . \boxed{
\alpha_i
\neq
\text{physical compute speed}.
} α i = physical compute speed .
clock multiplier 不會憑空增加硬體 throughput。
15. Temporal Feasibility
令:
κ i ( t i ) \kappa_i(t_i) κ i ( t i )
為每單位 local time 所需 physical work density,
c i ( τ ) c_i(\tau) c i ( τ )
為有效 compute service rate。
若不跳事件、不降 fidelity、不改 semantics:
κ i ( t i ) α i r e a l ≤ c i . \boxed{
\kappa_i(t_i)
\alpha_i^{real}
\le
c_i.
} κ i ( t i ) α i r e a l ≤ c i .
16. Requested / Realized Temporal Rate Non-Collapse
α i c m d ≠ α i r e a l . \boxed{
\alpha_i^{cmd}
\neq
\alpha_i^{real}.
} α i c m d = α i r e a l .
要求 100 × 100\times 100 × 不代表能實現 100 × 100\times 100 × 。
17. 六類 Temporal Domain
本文採用:
T D = { A n c h o r e d , E l a s t i c , E v e n t J u m p , R e p l a y , S p e c u l a t i v e , F r o z e n } . \boxed{
\mathfrak T_D
=
\{
Anchored,
Elastic,
EventJump,
Replay,
Speculative,
Frozen
\}.
} T D = { A n c h or e d , E l a s t i c , E v e n t J u m p , R e pl a y , S p ec u l a t i v e , F r oz e n } .
18. Anchored Domain
與 wall time / external time 有強同步需求:
human interaction;
audio;
network protocol;
external sensor;
market feed;
real deadline。
可要求:
∣ t i − g i ( τ ) ∣ ≤ ϵ i . \boxed{
|t_i-g_i(\tau)|
\le
\epsilon_i.
} ∣ t i − g i ( τ ) ∣ ≤ ϵ i .
19. Elastic Domain
允許:
α i ∈ [ α i m i n , α i m a x ] . \boxed{
\alpha_i
\in
[
\alpha_i^{min},
\alpha_i^{max}
].
} α i ∈ [ α i min , α i ma x ] .
適合 offline simulation、background world model、batch analysis 等。
20. Event-Jump Domain
若:
t ⋆ > t t^\star
>
t t ⋆ > t
是下一個會改變 observable state 的有效事件,
則可:
t i → t ⋆ \boxed{
t_i
\rightarrow
t^\star
} t i → t ⋆
而不逐 tick 執行空狀態。
21. Fast-Forward / Event-Jump Non-Collapse
FastForward ≠ EventSkipping . \boxed{
\text{FastForward}
\neq
\text{EventSkipping}.
} FastForward = EventSkipping .
前者可能仍計算所有 intermediate work;後者依 event semantics 省略無狀態變化區間。
22. Replay Domain
Replay domain 可與 live world time 解耦。
其目標可以是:
max Replay Throughput \boxed{
\max
\text{Replay Throughput}
} max Replay Throughput
subject to determinism / evidence / resource constraints。
23. Speculative Domain
由 snapshot:
Σ t \Sigma_t Σ t
建立:
{ D ( 1 ) , D ( 2 ) , … , D ( n ) } . \boxed{
\{
D^{(1)},
D^{(2)},
\ldots,
D^{(n)}
\}.
} { D ( 1 ) , D ( 2 ) , … , D ( n ) } .
24. Candidate Branch / Commit Non-Collapse
Candidate Branch ≠ World Commit . \boxed{
\text{Candidate Branch}
\neq
\text{World Commit}.
} Candidate Branch = World Commit .
這是 PNCW branch projection 必須保留的 authority boundary。
25. Frozen Domain
α i = 0. \alpha_i=0. α i = 0.
但:
Temporal Inactivity ≠ Ontological Deletion . \boxed{
\text{Temporal Inactivity}
\neq
\text{Ontological Deletion}.
} Temporal Inactivity = Ontological Deletion .
26. Multi-Temporal World
若:
D = { D 1 , … , D n } , \mathcal D
=
\{D_1,\ldots,D_n\}, D = { D 1 , … , D n } ,
則同一 global system 在 physical time τ \tau τ 可以具有:
( t 1 ( τ ) , t 2 ( τ ) , … , t n ( τ ) ) . \boxed{
(
t_1(\tau),
t_2(\tau),
\ldots,
t_n(\tau)
).
} ( t 1 ( τ ) , t 2 ( τ ) , … , t n ( τ )) .
不要求:
t i = t j . t_i=t_j. t i = t j .
27. Multi-Temporal Projection
因此 observer projection 不應只問:
現在是幾點?
而應問:
Which domain? Which local time? Which temporal contract? \boxed{
\text{Which domain? Which local time? Which temporal contract?}
} Which domain? Which local time? Which temporal contract?
28. Temporal Selection Operator
定義:
Θ O : S → Δ T O . \boxed{
\Theta_O:
\mathfrak S
\rightarrow
\Delta T_O.
} Θ O : S → Δ T O .
Δ T O \Delta T_O Δ T O 可以包含多個 domain-specific temporal windows。
29. Temporal Window
例如:
Δ T O = { ( D 1 , [ t 1 a , t 1 b ] ) , ( D 2 , t 2 ⋆ ) , ( D 3 , replay [ u , v ] ) } . \boxed{
\Delta T_O
=
\{
(D_1,[t_1^a,t_1^b]),
(D_2,t_2^\star),
(D_3,\text{replay }[u,v])
\}.
} Δ T O = {( D 1 , [ t 1 a , t 1 b ]) , ( D 2 , t 2 ⋆ ) , ( D 3 , replay [ u , v ])} .
所以一個 observation 可以同時橫跨不同 local times。
30. Local-Time Freedom / Boundary Freedom Non-Collapse
LocalTimeFreedom ≠ BoundaryFreedom . \boxed{
\text{LocalTimeFreedom}
\neq
\text{BoundaryFreedom}.
} LocalTimeFreedom = BoundaryFreedom .
一個 domain 可以本地 fast-forward,但跨 external I/O、commit、human boundary 時仍需 synchronization contract。
31. Cross-Domain Synchronization
若:
e i → e j e_i
\rightarrow
e_j e i → e j
跨 domain 有 causal exchange,
需:
Γ i j \boxed{
\Gamma_{ij}
} Γ ij
描述 synchronization / mapping / tolerance contract。
32. Temporal Drift
對 anchored / synchronized domain:
δ i ( τ ) = t i ( τ ) − g i ( τ ) . \boxed{
\delta_i(\tau)
=
t_i(\tau)
-
g_i(\tau).
} δ i ( τ ) = t i ( τ ) − g i ( τ ) .
33. Temporal Debt
若實際 progress 低於 target progress,可以定義:
B i = max ( 0 , p i ⋆ − p i ) . \boxed{
B_i
=
\max(
0,
p_i^\star-p_i
).
} B i = max ( 0 , p i ⋆ − p i ) .
Temporal debt 可影響 materialization、priority、observation 與 resource allocation。
34. Temporal Observation Resolution
觀察頻率:
ρ i o b s \rho_i^{obs} ρ i o b s
不必等於 execution rate:
α i . \alpha_i. α i .
因此:
Execution Frequency ≠ Observation Frequency . \boxed{
\text{Execution Frequency}
\neq
\text{Observation Frequency}.
} Execution Frequency = Observation Frequency .
35. Adaptive Temporal Observation
可根據:
information change;
uncertainty;
risk;
temporal debt;
causal importance;
global objective;
調整:
ρ i o b s . \boxed{
\rho_i^{obs}.
} ρ i o b s .
36. Observation Resolution / Temporal Rate Non-Collapse
λ o b s e r v e ≠ α t i m e . \boxed{
\lambda^{observe}
\neq
\alpha^{time}.
} λ o b ser v e = α t im e .
domain 跑很快,不代表 observer 要高頻看它。
37. Event Set
令:
E = { e 1 , … , e m } . \boxed{
E
=
\{e_1,\ldots,e_m\}.
} E = { e 1 , … , e m } .
若 e a e_a e a 是 e b e_b e b 合法發生/完成/驗證/commit 的必要前置:
e a ≺ e b . \boxed{
e_a
\prec
e_b.
} e a ≺ e b .
38. Causal Partial Order
若:
e a ⊀ e b e_a\nprec e_b e a ⊀ e b
且:
e b ⊀ e a , e_b\nprec e_a, e b ⊀ e a ,
則在目前 causal model:
e a ∥ C e b . \boxed{
e_a
\parallel_C
e_b.
} e a ∥ C e b .
39. Incomparability 不等於 Safe Parallelism
即使:
e a ∥ C e b , e_a\parallel_C e_b, e a ∥ C e b ,
仍可能:
write same state;
compete same GPU;
share lock;
violate resource compatibility。
因此:
S a f e P a r a l l e l ( a , b ) = C ( a , b ) ∧ R ( a , b ) ∧ S ( a , b ) . \boxed{
SafeParallel(a,b)
=
C(a,b)
\land
R(a,b)
\land
S(a,b).
} S a f e P a r a l l e l ( a , b ) = C ( a , b ) ∧ R ( a , b ) ∧ S ( a , b ) .
40. Causal Graph / Resource Graph / Containment Graph
至少區分:
G C ≠ G R ≠ G O . \boxed{
G_C
\neq
G_R
\neq
G_O.
} G C = G R = G O .
其中:
G C G_C G C :causal graph;
G R G_R G R :resource contention graph;
G O G_O G O :ownership / containment graph。
41. Containment / Causality Non-Collapse
Containment ≠ Causality . \boxed{
\text{Containment}
\neq
\text{Causality}.
} Containment = Causality .
UI parent-child tree 不能直接當 execution dependency。
42. Serialization / Causality Non-Collapse
若:
l e d g e r _ s e q ( u ) < l e d g e r _ s e q ( v ) , ledger\_seq(u)
<
ledger\_seq(v), l e d g er _ se q ( u ) < l e d g er _ se q ( v ) ,
不自動表示:
u ≺ v . u\prec v. u ≺ v .
因此:
Serialization Order ≠ Necessary Causal Order . \boxed{
\text{Serialization Order}
\neq
\text{Necessary Causal Order}.
} Serialization Order = Necessary Causal Order .
43. Token Stream 是 Linearization
令 causal structure:
P = ( E , ≺ ) . \mathcal P
=
(E,\prec). P = ( E , ≺ ) .
序列化輸出:
L ( P ) = ( e π ( 1 ) , … , e π ( n ) ) . \boxed{
L(\mathcal P)
=
(
e_{\pi(1)},
\ldots,
e_{\pi(n)}
).
} L ( P ) = ( e π ( 1 ) , … , e π ( n ) ) .
其中 π \pi π 必須至少滿足需要保留的 happens-before constraints。
44. Linearization 不等於 Causal Structure
L ( P ) ≠ P . \boxed{
L(\mathcal P)
\neq
\mathcal P.
} L ( P ) = P .
linearization 可以攜帶 causal annotation,但一條裸 sequence 本身不等於原 partial order。
45. 多個合法 Linearization
對同一 poset,可存在:
L 1 ( P ) , L 2 ( P ) , … L_1(\mathcal P),
L_2(\mathcal P),
\ldots L 1 ( P ) , L 2 ( P ) , …
而它們都尊重 hard causal order。
因此:
One Causal World → Many Valid Sequences . \boxed{
\text{One Causal World}
\rightarrow
\text{Many Valid Sequences}.
} One Causal World → Many Valid Sequences .
46. Sequence Non-Uniqueness
這直接支持 PNCW:
Presentation Order is not uniquely determined by causal ontology . \boxed{
\text{Presentation Order}
\text{ is not uniquely determined by causal ontology}.
} Presentation Order is not uniquely determined by causal ontology .
47. Causal Work
對 causal DAG:
G = ( V , E ) , G=(V,E), G = ( V , E ) ,
權重:
w : V → R ≥ 0 . w:V\rightarrow\mathbb R_{\ge0}. w : V → R ≥ 0 .
定義:
W ( G , w ) = ∑ v ∈ V w ( v ) . \boxed{
W(G,w)
=
\sum_{v\in V}w(v).
} W ( G , w ) = v ∈ V ∑ w ( v ) .
48. Critical Depth
D ( G , w ) = max p ∈ P ( G ) ∑ v ∈ p w ( v ) . \boxed{
D(G,w)
=
\max_{p\in\mathcal P(G)}
\sum_{v\in p}w(v).
} D ( G , w ) = p ∈ P ( G ) max v ∈ p ∑ w ( v ) .
49. Structural Parallelism Signal
Π s = W D . \boxed{
\Pi_s
=
\frac{W}{D}.
} Π s = D W .
但:
W D ≠ CPU Count Recommendation . \boxed{
\frac{W}{D}
\neq
\text{CPU Count Recommendation}.
} D W = CPU Count Recommendation .
50. Poset Width
DAG reachability 誘導偏序:
u ⪯ v . u\preceq v. u ⪯ v .
定義:
ω ( G ) = max A ∣ A ∣ \boxed{
\omega(G)
=
\max_A |A|
} ω ( G ) = A max ∣ A ∣
其中 (A) 為 antichain。
51. Width / Work-Depth Non-Collapse
ω ( G ) ≠ W D . \boxed{
\omega(G)
\neq
\frac{W}{D}.
} ω ( G ) = D W .
width 描述最大不可比較集合;(W/D) 描述平均結構比例。
52. Causal Projection
定義:
K O = Π C a u s a l ( G C , q , O , S c o p e ) . \boxed{
\mathcal K_O
=
\Pi_Causal(
G_C,
q,
O,
Scope
).
} K O = Π C a u s a l ( G C , q , O , S co p e ) .
它可以是一個 causal cone / dependency region。
53. Backward Causal Cone
對 result (r):
K − ( r ) = { e : e ⪯ r } . \boxed{
K^{-}(r)
=
\{
e:
e\preceq r
\}.
} K − ( r ) = { e : e ⪯ r } .
表示所有可能對 (r) 有必要因果路徑的 events。
54. Forward Causal Cone
K + ( e ) = { x : e ⪯ x } . \boxed{
K^{+}(e)
=
\{
x:
e\preceq x
\}.
} K + ( e ) = { x : e ⪯ x } .
表示 event (e) 可能影響的 downstream region。
55. Causal Cone Projection
Observer 可以要求:
給我導致這個錯誤的 causal cone。
即:
Y = Π S T ( S ∣ K − ( e r r o r ) ) . \boxed{
Y
=
\Pi^{ST}(
\mathfrak S
\mid
K^{-}(error)
).
} Y = Π S T ( S ∣ K − ( er r or )) .
56. Critical-Path Projection
也可以:
Y c r i t = Π S T ( S ∣ C r i t i c a l P a t h ( G C , w ) ) . \boxed{
Y_{crit}
=
\Pi^{ST}(
\mathfrak S
\mid
CriticalPath(G_C,w)
).
} Y cr i t = Π S T ( S ∣ C r i t i c a l P a t h ( G C , w )) .
57. Antichain Projection
可一次顯示:
A ⋆ = arg max A ∣ A ∣ . \boxed{
A^\star
=
\arg\max_A |A|.
} A ⋆ = arg A max ∣ A ∣.
這是一組彼此不可比較的 events。
58. Parallel Region / Parallel Execution Non-Collapse
即使 antichain 可視為 potential parallel region,也仍需 resource / state compatibility。
所以:
Parallel Projection ≠ Safe Parallel Execution . \boxed{
\text{Parallel Projection}
\neq
\text{Safe Parallel Execution}.
} Parallel Projection = Safe Parallel Execution .
59. Snapshot
令:
Σ t \boxed{
\Sigma_t
} Σ t
為可恢復 snapshot。
60. Replay
R ( Σ t , U t : t + k ) → S ^ t + k . \boxed{
\mathcal R(
\Sigma_t,
U_{t:t+k}
)
\rightarrow
\hat S_{t+k}.
} R ( Σ t , U t : t + k ) → S ^ t + k .
61. Branch
B ( Σ t ) = { Σ t ( 1 ) , … , Σ t ( n ) } . \boxed{
\mathcal B(\Sigma_t)
=
\{
\Sigma_t^{(1)},
\ldots,
\Sigma_t^{(n)}
\}.
} B ( Σ t ) = { Σ t ( 1 ) , … , Σ t ( n ) } .
62. Branch Dimensions
不同 branch 可具有不同:
inputs;
policy;
resource allocation;
time rate;
AI decision;
model hypothesis。
63. Branch Projection
定義:
Π B : { Σ ( 1 ) , … , Σ ( n ) } → V B . \boxed{
\Pi_B:
\{
\Sigma^{(1)},
\ldots,
\Sigma^{(n)}
\}
\rightarrow
V_B.
} Π B : { Σ ( 1 ) , … , Σ ( n ) } → V B .
64. Branch Ensemble
observer 可以一次取得:
V B = ⟨ S h a r e d P r e f i x , B r a n c h P o i n t s , D i v e r g e n c e , O u t c o m e s , C o s t s , R i s k s , C o m m i t S t a t u s ⟩ . \boxed{
V_B
=
\left\langle
SharedPrefix,
BranchPoints,
Divergence,
Outcomes,
Costs,
Risks,
CommitStatus
\right\rangle.
} V B = ⟨ S ha r e d P r e f i x , B r an c h P o in t s , D i v er g e n ce , O u t co m es , C os t s , R i s k s , C o mmi tS t a t u s ⟩ .
65. 一次看見多個未來
因此「一口氣看到」可以是:
One Branch Point → Multiple Projected Futures . \boxed{
\text{One Branch Point}
\rightarrow
\text{Multiple Projected Futures}.
} One Branch Point → Multiple Projected Futures .
而不是依序讀:
Future A...
Future B...
Future C...
66. Replay / Live Non-Collapse
Replay State ≠ Live World State . \boxed{
\text{Replay State}
\neq
\text{Live World State}.
} Replay State = Live World State .
兩者可同時被投影到同一 Canvas,但必須有明確 temporal identity。
67. Candidate Branch / World Commit Non-Collapse
再次要求:
Candidate Branch ≠ World Commit . \boxed{
\text{Candidate Branch}
\neq
\text{World Commit}.
} Candidate Branch = World Commit .
視覺上看見 speculative future 不代表它已成為現實 world state。
68. Spacetime Projection Operator
本文正式定義:
Π q , O , σ , Λ , Δ T , K , B , π S T : S → Y S T . \boxed{
\Pi^{ST}_{
q,
O,
\sigma,
\Lambda,
\Delta T,
\mathcal K,
\mathcal B,
\pi
}
:
\mathfrak S
\rightarrow
Y^{ST}.
} Π q , O , σ , Λ , Δ T , K , B , π S T : S → Y S T .
69. Projection Scope
σ \sigma σ 決定:
domains;
entities;
resources;
semantic regions。
70. Resolution Bundle
定義:
Λ = ( λ s t a t e , λ t i m e , λ c a u s a l , λ b r a n c h , λ o b s e r v e , λ r e n d e r ) . \boxed{
\Lambda
=
(
\lambda^{state},
\lambda^{time},
\lambda^{causal},
\lambda^{branch},
\lambda^{observe},
\lambda^{render}
).
} Λ = ( λ s t a t e , λ t im e , λ c a u s a l , λ b r an c h , λ o b ser v e , λ r e n d er ) .
71. State Resolution
λ s t a t e \lambda^{state} λ s t a t e
控制 state detail。
72. Temporal Resolution
λ t i m e \lambda^{time} λ t im e
控制:
sample frequency;
interval granularity;
event density;
replay detail。
73. Causal Resolution
λ c a u s a l \lambda^{causal} λ c a u s a l
控制:
hard verified edges only;
inferred edges;
full provenance;
summarized causal groups。
74. Branch Resolution
λ b r a n c h \lambda^{branch} λ b r an c h
控制:
show all branches;
top-k branches;
cluster equivalent futures;
show only committed branch。
75. Observe / Render Resolution
沿用 PNCW:
λ o b s e r v e ≠ λ r e n d e r . \lambda^{observe}
\neq
\lambda^{render}. λ o b ser v e = λ r e n d er .
76. 六種 Resolution Non-Collapse
一般:
λ s t a t e ≠ λ t i m e ≠ λ c a u s a l ≠ λ b r a n c h ≠ λ o b s e r v e ≠ λ r e n d e r . \boxed{
\lambda^{state}
\neq
\lambda^{time}
\neq
\lambda^{causal}
\neq
\lambda^{branch}
\neq
\lambda^{observe}
\neq
\lambda^{render}.
} λ s t a t e = λ t im e = λ c a u s a l = λ b r an c h = λ o b ser v e = λ r e n d er .
77. Spacetime Projection Contract
定義:
S T P r o j C o n t r a c t = ⟨ Q u e r y , O b s e r v e r , D o m a i n s , T e m p o r a l S e l e c t i o n , C a u s a l S e l e c t i o n , B r a n c h S e l e c t i o n , R e s o l u t i o n B u n d l e , A u t h o r i t y , H i s t o r y P o l i c y , F a l l b a c k ⟩ . \boxed{
\mathsf{STProjContract}
=
\left\langle
Query,
Observer,
Domains,
TemporalSelection,
CausalSelection,
BranchSelection,
ResolutionBundle,
Authority,
HistoryPolicy,
Fallback
\right\rangle.
} STProjContract = ⟨ Q u er y , O b ser v er , D o main s , T e m p or a l S e l ec t i o n , C a u s a l S e l ec t i o n , B r an c h S e l ec t i o n , R eso l u t i o n B u n d l e , A u t h or i t y , H i s t or y P o l i cy , F a l l ba c k ⟩ .
78. Temporal Anchor
每個 projection 必須聲明:
T e m p o r a l A n c h o r . \boxed{
\mathsf{TemporalAnchor}.
} TemporalAnchor .
例如:
LIVE;
SNAPSHOT;
REPLAY;
SPECULATIVE;
FROZEN;
MULTI-TIME。
79. Branch Anchor
同樣:
B r a n c h A n c h o r . \boxed{
\mathsf{BranchAnchor}.
} BranchAnchor .
不能讓不同 branch 的 state 無標記混在一起。
80. Causal Evidence Class
因果 edge 應有 epistemic class:
{ o b s e r v e d , v e r i f i e d , i n f e r r e d , h y p o t h e s i z e d , u n k n o w n , r e j e c t e d } . \boxed{
\{
observed,
verified,
inferred,
hypothesized,
unknown,
rejected
\}.
} { o b ser v e d , v er i f i e d , in f er r e d , h y p o t h es i z e d , u nk n o w n , r e j ec t e d } .
81. Missing Edge / Proven Independence Non-Collapse
MissingEdge ≠ ProvenIndependence . \boxed{
\text{MissingEdge}
\neq
\text{ProvenIndependence}.
} MissingEdge = ProvenIndependence .
這對 AI-generated causal projection尤其重要。
82. Hard / Soft Causal Layer
Hard causal scheduling / commit:
E h a r d E^{hard} E ha r d
應要求較強證據。
Observer explanatory view 可額外顯示:
E s o f t . E^{soft}. E so f t .
83. Causal Projection / Causal Authority Non-Collapse
Visible Causal Hypothesis ≠ Scheduling / Commit Authority . \boxed{
\text{Visible Causal Hypothesis}
\neq
\text{Scheduling / Commit Authority}.
} Visible Causal Hypothesis = Scheduling / Commit Authority .
84. Spacetime Materialization
materialization 不只決定「哪些 state」。
還要決定:
which times, events, edges and branches become resident . \boxed{
\text{which times, events, edges and branches become resident}.
} which times, events, edges and branches become resident .
85. Temporal Materialization
例如只保留:
every 100th snapshot;
anomaly neighborhoods;
commit boundaries;
branch points;
critical-path events。
86. Causal Materialization
只 materialize:
K − ( t a r g e t ) K^{-}(target) K − ( t a r g e t )
而不展開整個 history graph。
87. Branch Materialization
只 materialize top-k branches:
B ⋆ ⊂ B . \mathcal B^\star
\subset
\mathcal B. B ⋆ ⊂ B .
88. Selective Spacetime Materialization
因此:
Selective Materialization = State + Time + Causality + Branch . \boxed{
\text{Selective Materialization}
=
\text{State}
+
\text{Time}
+
\text{Causality}
+
\text{Branch}.
} Selective Materialization = State + Time + Causality + Branch .
89. GCM Integration
GCM Paper 05 已決定:
what domain;
what representation;
what resolution;
what materialization。
Paper 07 將 temporal / causal axes 加入 GCM plan:
P P l a n S T = ( D o m a i n P l a n , T e m p o r a l P l a n , C a u s a l P l a n , B r a n c h 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 v e a l P l a n ) . \boxed{
\mathsf{PPlan}^{ST}
=
(
DomainPlan,
TemporalPlan,
CausalPlan,
BranchPlan,
CarrierPlan,
VisualPlan,
RevealPlan
).
} PPlan S T = ( D o main P l an , T e m p or a l P l an , C a u s a l P l an , B r an c h P l an , C a r r i er P l an , V i s u a l P l an , R e v e a l P l an ) .
90. Temporal Plan
決定:
temporal class;
time window;
replay/live;
sample rate;
temporal debt tolerance;
synchronization constraints。
91. Causal Plan
決定:
target causal cone;
edge evidence threshold;
work/depth summary;
critical path;
branch/join points。
92. Branch Plan
決定:
committed branch;
speculative branches;
comparison set;
branch materialization budget。
93. Context Projection Integration
PNCW Paper 02 的:
C q a c t i v e C_q^{active} C q a c t i v e
現在可由:
C q a c t i v e = Π C ( S ∣ Δ T , K , B ) . \boxed{
C_q^{active}
=
\Pi_C(
\mathfrak S
\mid
\Delta T,
\mathcal K,
\mathcal B
).
} C q a c t i v e = Π C ( S ∣ Δ T , K , B ) .
active cognition 可以包含一段 causal-temporal world,而不是只含「當前資料」。
94. Context 可包含 Worldline
例如:
C q a c t i v e = { γ x [ t a : t b ] , K − ( r ) , S n a p s h o t t , B r a n c h D i f f } . C_q^{active}
=
\{
\gamma_x[t_a:t_b],
K^{-}(r),
Snapshot_t,
BranchDiff
\}. C q a c t i v e = { γ x [ t a : t b ] , K − ( r ) , S na p s h o t t , B r an c h D i f f } .
95. SPET Integration
SPET Freeze 不應被誤解為凍結 source world evolution。
它 Freeze 的是:
projection frame / scope contract . \boxed{
\text{projection frame / scope contract}.
} projection frame / scope contract .
96. Spacetime Source 可繼續演化
source:
S t → S t + 1 \mathfrak S_t
\rightarrow
\mathfrak S_{t+1} S t → S t + 1
仍可持續。
但某 projected epoch:
E k \mathcal E_k E k
綁定 temporal / causal anchor。
97. Temporal Anchor / Frame ID
Carrier FrameID 應可加入:
T e m p o r a l A n c h o r + B r a n c h A n c h o r + C a u s a l S c o p e D i g e s t . \boxed{
TemporalAnchor
+
BranchAnchor
+
CausalScopeDigest.
} T e m p or a l A n c h or + B r an c h A n c h or + C a u s a l S co p eD i g es t .
避免同座標混入不同時域/分支 state。
98. HDSRC Integration
HDSRC carrier 可以不只保存:
還可投影:
temporal layers;
event layers;
branch IDs;
causal edge layers;
critical path overlays。
99. Time as Carrier Dimension
時間不一定要映射成 horizontal x-axis。
可選:
separate layers;
recursive subcanvas;
color / phase;
animation;
branch plane;
explicit time tiles。
因此:
Temporal Projection ≠ Timeline UI only . \boxed{
\text{Temporal Projection}
\neq
\text{Timeline UI only}.
} Temporal Projection = Timeline UI only .
100. MRMIC / NVCL Integration
Canvas 可同時顯示:
live world;
frozen snapshot;
replay;
speculative branch;
causal cone;
critical path;
off-critical parallel regions。
101. Multi-Time Canvas
定義:
V M T = Ψ ( D 1 ( t 1 ) , D 2 ( t 2 ) , … , D n ( t n ) ) . \boxed{
V^{MT}
=
\Psi(
D_1(t_1),
D_2(t_2),
\ldots,
D_n(t_n)
).
} V M T = Ψ ( D 1 ( t 1 ) , D 2 ( t 2 ) , … , D n ( t n )) .
同一 Canvas 不要求所有 panels 共享同一 local time。
102. Time Label Requirement
每個 visual region 必須可知:
D o m a i n I D , T e m p o r a l A n c h o r , B r a n c h I D , V e r s i o n . \boxed{
DomainID,
TemporalAnchor,
BranchID,
Version.
} D o main I D , T e m p or a l A n c h or , B r an c h I D , V er s i o n .
否則 multi-time projection 容易產生 semantic confusion。
103. Causal Canvas
Canvas edge 可以分:
containment;
causal;
resource;
ownership;
inferred;
speculative。
不能全部畫成同一種 arrow。
104. Causal Layer Toggle
Observer 可以:
L a y e r T o g g l e ( V e r i f i e d , I n f e r r e d , R e s o u r c e , C o n t a i n m e n t ) . \boxed{
LayerToggle(
Verified,
Inferred,
Resource,
Containment
).
} L a y er T o g g l e ( V er i f i e d , I n f er r e d , R eso u r ce , C o n t ainm e n t ) .
這是 observer operation,不改 source world。
105. Worldline View
Visual object 可展開:
γ x [ t a : t b ] . \boxed{
\gamma_x[t_a:t_b].
} γ x [ t a : t b ] .
使用:
path;
timeline;
state strip;
event graph;
animation。
106. Causal Cone View
對 error / result:
K − ( t a r g e t ) \boxed{
K^{-}(target)
} K − ( t a r g e t )
可直接在 Canvas 中高亮。
107. Branch Ensemble View
對:
B ( Σ t ) \mathcal B(\Sigma_t) B ( Σ t )
Canvas 可以一次顯示:
common prefix;
divergence;
each branch state;
policy differences;
cost;
risk;
commit status。
108. Critical Path View
對:
C r i t i c a l P a t h ( G C , w ) CriticalPath(G_C,w) C r i t i c a l P a t h ( G C , w )
可直接顯示真正 blocking chain,而不是只看 CPU utilization。
109. Antichain View
對最大 antichain:
A ⋆ A^\star A ⋆
可視化 structural concurrency opportunities。
但需標記:
Potential Concurrency ≠ Safe Concurrency . \boxed{
\text{Potential Concurrency}
\neq
\text{Safe Concurrency}.
} Potential Concurrency = Safe Concurrency .
110. Atomic Spacetime Reveal
PNCW Paper 01 的 Atomic Reveal 可提升成:
∅ → Y S T a u t h . \boxed{
\varnothing
\rightarrow
Y_{ST}^{auth}.
} ∅ → Y S T a u t h .
其中 Y S T a u t h Y_{ST}^{auth} Y S T a u t h 是完整 causal-temporal artifact。
111. Atomic Reveal 不等於全部 History Bytes Resident
仍然:
Logical Spacetime Availability ≠ Full Historical Residency . \boxed{
\text{Logical Spacetime Availability}
\neq
\text{Full Historical Residency}.
} Logical Spacetime Availability = Full Historical Residency .
112. Progressive Temporal Materialization
Reveal 後可按需:
expand older history;
load causal predecessors;
open replay;
materialize branch;
increase temporal resolution。
113. 一口氣看到的 Spacetime 版本
使用者可以一次看到:
Result
├─ current state
├─ causal predecessors
├─ parallel branches
├─ critical path
├─ speculative alternatives
├─ replay lineage
└─ unresolved causal edges
而不是先讀完一條 narrative 才理解整個 execution structure。
114. Sequential Narrative as Projection
自然語言仍然可以:
T e x t = Π n a r r a t i v e ( Y S T ) . \boxed{
Text
=
\Pi_{narrative}(Y_{ST}).
} T e x t = Π na r r a t i v e ( Y S T ) .
它是一種 projection,而不是 source ontology。
115. Narrative Order / Causal Order Non-Collapse
作者可以為可讀性先講 result、後講原因。
所以:
Narrative Order ≠ Causal Order . \boxed{
\text{Narrative Order}
\neq
\text{Causal Order}.
} Narrative Order = Causal Order .
116. Text Can Preserve Causality Explicitly
本文不主張 sequence 必然丟失 causal information。
若 text 帶 explicit graph IDs / references,它可以描述 causal DAG。
因此更精確是:
Bare Linear Order does not itself encode full causal topology . \boxed{
\text{Bare Linear Order}
\text{ does not itself encode full causal topology}.
} Bare Linear Order does not itself encode full causal topology .
117. Presentation Mismatch
主要問題是:
Causal-Temporal World → Mandatory Bare Sequence \boxed{
\text{Causal-Temporal World}
\rightarrow
\text{Mandatory Bare Sequence}
} Causal-Temporal World → Mandatory Bare Sequence
可能使:
concurrency 變難看;
branch structure 變難導航;
replay/live 混淆;
critical path 不直觀;
temporal heterogeneity 被壓平。
118. Spacetime Observation Topologies
PNCW 現在可加入:
T O S T = { T i m e l i n e , D A G , W o r l d l i n e , B r a n c h T r e e , M u l t i T i m e C a n v a s , C a u s a l C o n e , H y b r i d } . \boxed{
\mathfrak T_O^{ST}
=
\{
Timeline,
DAG,
Worldline,
BranchTree,
MultiTimeCanvas,
CausalCone,
Hybrid
\}.
} T O S T = { T im e l in e , D A G , W or l d l in e , B r an c h T r ee , M u l t i T im e C an v a s , C a u s a l C o n e , H y b r i d } .
119. Timeline
適合 single-domain chronological view。
120. DAG
適合 causal dependencies。
121. Worldline
適合 tracking one entity across time。
122. Branch Tree
適合 replay / speculation / policy alternatives。
123. Multi-Time Canvas
適合同時比較多 domains 的不同 local times。
124. Hybrid
一個 artifact 可以同時:
DAG;
timeline;
worldline;
canvas;
text narrative。
125. Observer-Relative Spacetime
不同 observer 可選不同:
Δ T , K , B , Λ . \Delta T,
\mathcal K,
\mathcal B,
\Lambda. Δ T , K , B , Λ.
因此:
Y O 1 S T ≠ Y O 2 S T \boxed{
Y_{O_1}^{ST}
\neq
Y_{O_2}^{ST}
} Y O 1 S T = Y O 2 S T
完全合法。
126. State Equality / History Equality Non-Collapse
即使:
S ( t 1 ) = S ( t 2 ) , S(t_1)=S(t_2), S ( t 1 ) = S ( t 2 ) ,
仍可能:
H ( t 1 ) ≠ H ( t 2 ) . \boxed{
H(t_1)
\neq
H(t_2).
} H ( t 1 ) = H ( t 2 ) .
PNCW 不應只用 endpoint state 判斷相同 world history。
127. Same Endpoint / Different Worldlines
存在:
γ 1 ( t f ) = γ 2 ( t f ) \gamma_1(t_f)
=
\gamma_2(t_f) γ 1 ( t f ) = γ 2 ( t f )
但:
γ 1 ≠ γ 2 . \gamma_1
\neq
\gamma_2. γ 1 = γ 2 .
所以:
Same Endpoint ≠ Same Worldline . \boxed{
\text{Same Endpoint}
\neq
\text{Same Worldline}.
} Same Endpoint = Same Worldline .
128. Replay Equality / Live Equality Non-Collapse
Replay 可以重建相同 state:
S ^ t = S t \hat S_t=S_t S ^ t = S t
但 execution context / authority / external side effects 可能不同。
因此:
Replay State Equality ≠ Live World Identity . \boxed{
\text{Replay State Equality}
\neq
\text{Live World Identity}.
} Replay State Equality = Live World Identity .
129. Branch Equality / Commit Equality Non-Collapse
一個 speculative branch 可能與 live world state byte-identical,仍不代表已 committed。
130. Temporal Inactivity / Nonexistence Non-Collapse
Frozen domain 保留:
Temporal Inactivity ≠ Nonexistence . \boxed{
\text{Temporal Inactivity}
\neq
\text{Nonexistence}.
} Temporal Inactivity = Nonexistence .
這與 PNCW dormant/materialized state完全相容。
131. Global Time / Local Time Non-Collapse
Reference Time ≠ Local Domain Time . \boxed{
\text{Reference Time}
\neq
\text{Local Domain Time}.
} Reference Time = Local Domain Time .
132. Local Time / Causal Order Non-Collapse
兩個 event timestamp:
t a < t b t_a<t_b t a < t b
不必自動推出:
a ≺ b . a\prec b. a ≺ b .
133. Temporal Distance / Causal Distance Non-Collapse
d t i m e ( a , b ) ≠ d c a u s a l ( a , b ) . \boxed{
d_{time}(a,b)
\neq
d_{causal}(a,b).
} d t im e ( a , b ) = d c a u s a l ( a , b ) .
時間很近的 events 可以因果無關;時間很遠的 events 可以有直接 dependency。
134. Spacetime Query Types
PNCW Runtime 應支援:
state-at-time;
worldline segment;
backward causal cone;
forward impact cone;
branch comparison;
replay/live diff;
critical path;
antichain / concurrency region;
cross-time snapshot diff;
multi-domain temporal alignment。
135. Query Example — Why?
「為什麼得到結果 (r)?」
Q u e r y w h y ( r ) → K − ( r ) . \boxed{
Query_{why}(r)
\rightarrow
K^{-}(r).
} Q u er y w h y ( r ) → K − ( r ) .
136. Query Example — What If?
「如果在 snapshot Σ t \Sigma_t Σ t 改 policy?」
Q u e r y w h a t i f → B ( Σ t ) . \boxed{
Query_{whatif}
\rightarrow
\mathcal B(\Sigma_t).
} Q u er y w ha t i f → B ( Σ t ) .
137. Query Example — What Changed?
Q u e r y c h a n g e → D i f f ( γ x [ t a : t b ] ) . \boxed{
Query_{change}
\rightarrow
Diff(
\gamma_x[t_a:t_b]
).
} Q u er y c han g e → D i f f ( γ x [ t a : t b ]) .
138. Query Example — What Is Blocking?
Q u e r y b l o c k → C r i t i c a l P a t h ( G C , w ) . \boxed{
Query_{block}
\rightarrow
CriticalPath(G_C,w).
} Q u er y b l oc k → C r i t i c a l P a t h ( G C , w ) .
139. Query Example — What Can Run Independently?
Q u e r y p a r a l l e l → A n t i c h a i n C a n d i d a t e s \boxed{
Query_{parallel}
\rightarrow
AntichainCandidates
} Q u er y p a r a l l e l → A n t i c hain C an d i d a t es
再經 resource / state compatibility filter。
140. Spacetime Projection Readiness
定義:
S T R e a d y = T e m p o r a l A n c h o r V a l i d ∧ B r a n c h A n c h o r V a l i d ∧ C a u s a l S c o p e S u f f i c i e n t ∧ V e r s i o n C o h e r e n t ∧ A u t h o r i t y V a l i d . \boxed{
\mathsf{STReady}
=
TemporalAnchorValid
\land
BranchAnchorValid
\land
CausalScopeSufficient
\land
VersionCoherent
\land
AuthorityValid.
} STReady = T e m p or a l A n c h or V a l i d ∧ B r an c h A n c h or V a l i d ∧ C a u s a l S co p e S u f f i c i e n t ∧ V er s i o n C o h er e n t ∧ A u t h or i t y V a l i d .
141. STReady / ContextReady Non-Collapse
S T R e a d y ≠ C o n t e x t R e a d y . \boxed{
\mathsf{STReady}
\neq
\mathsf{ContextReady}.
} STReady = ContextReady .
spacetime source scope ready 後,才建立 task-specific active context。
142. STReady / CarrierReady Non-Collapse
同樣:
S T R e a d y ≠ C a r r i e r R e a d y . \boxed{
\mathsf{STReady}
\neq
\mathsf{CarrierReady}.
} STReady = CarrierReady .
source causal-temporal scope 可合法,但還未 Freeze / spatialize。
143. PNCW Pipeline 升級
Paper 06:
W t → P P l a n → C q a c t i v e → E k → P k → V . W_t
\rightarrow
PPlan
\rightarrow
C_q^{active}
\rightarrow
\mathcal E_k
\rightarrow
P_k
\rightarrow
V. W t → P P l an → C q a c t i v e → E k → P k → V .
Paper 07 升級為:
S → Ω q , O S T → C q a c t i v e → E k → P k → V q , k . \boxed{
\mathfrak S
\rightarrow
\Omega_{q,O}^{ST}
\rightarrow
C_q^{active}
\rightarrow
\mathcal E_k
\rightarrow
P_k
\rightarrow
V_{q,k}.
} S → Ω q , O S T → C q a c t i v e → E k → P k → V q , k .
144. Spacetime Projection Scope
定義:
Ω q , O S T = ( σ , Δ T , K , B , Λ , A n c h o r ) . \boxed{
\Omega_{q,O}^{ST}
=
(
\sigma,
\Delta T,
\mathcal K,
\mathcal B,
\Lambda,
Anchor
).
} Ω q , O S T = ( σ , Δ T , K , B , Λ , A n c h or ) .
這是進入 Context Projection 前的 source-selection layer。
145. Projection-Native Software Spacetime
本文正式定義:
P N S S = ⟨ S , Π S T , T O S T , S T R e a d y , L S T ⟩ . \boxed{
\mathsf{PNSS}
=
\left\langle
\mathfrak S,
\Pi^{ST},
\mathfrak T_O^{ST},
\mathsf{STReady},
\mathcal L_{ST}
\right\rangle.
} PNSS = ⟨ S , Π S T , T O S T , STReady , L S T ⟩ .
其中:
S \mathfrak S S :software spacetime;
Π S T \Pi^{ST} Π S T :spacetime projection operators;
T O S T \mathfrak T_O^{ST} T O S T :observation topologies;
S T R e a d y \mathsf{STReady} STReady :projection readiness;
L S T \mathcal L_{ST} L S T :temporal / causal / branch ledger。
146. PNCW Paper 07 規範 v0.1
PNCW-ST1 — World / Snapshot Separation
World 不得與單一 state snapshot 塌縮。
PNCW-ST2 — Software / Physical Spacetime Separation
Software spacetime 不得被宣稱為 physical spacetime。
PNCW-ST3 — State / Worldline Separation
單點 state 不等於完整 worldline。
PNCW-ST4 — Global / Local Time Separation
不同 domains 可以有不同 local temporal functions。
PNCW-ST5 — Temporal / Compute Rate Separation
logical time rate 不得偷換成 hardware speedup。
PNCW-ST6 — Serialization / Causality Separation
ledger / token / UI order 不得被默認成必要 causal order。
PNCW-ST7 — Containment / Causality Separation
ownership / UI tree 不得冒充 causal DAG。
PNCW-ST8 — Missing Edge / Independence Separation
未觀測 edge 不代表已證明 independent。
PNCW-ST9 — Candidate Branch / Commit Separation
speculative / replay branch 不得冒充 live committed world。
PNCW-ST10 — Temporal Inactivity / Deletion Separation
Frozen domain 不得被當不存在。
PNCW-ST11 — Multi-Resolution Spacetime Projection
state / temporal / causal / branch / observe / render resolution 必須可分離。
PNCW-ST12 — Temporal / Branch Anchoring
任何 authoritative spacetime projection 必須綁定 temporal anchor、branch identity 與 causal scope。
147. Proposition 1 — State Projection Is a Special Case of Spacetime Projection
如果:
Δ T = { t ⋆ } , \Delta T=\{t^\star\}, Δ T = { t ⋆ } ,
B = { b ⋆ } , \mathcal B=\{b^\star\}, B = { b ⋆ } ,
且 K \mathcal K K 僅保留當前 state projection 必要 relations,
則:
Π S T ( S ) = Π ( W t ⋆ ) \boxed{
\Pi^{ST}(\mathfrak S)
=
\Pi(W_{t^\star})
} Π S T ( S ) = Π ( W t ⋆ )
在指定 scope 下成立。
148. Proposition 2 — Same Causal Structure Admits Multiple Valid Linearizations
若:
P = ( E , ≺ ) \mathcal P=(E,\prec) P = ( E , ≺ )
不是 total order,
則存在多個:
L i ( P ) L_i(\mathcal P) L i ( P )
尊重所有 hard causal edges。
因此:
Causal Structure ⇏ Unique Presentation Sequence . \boxed{
\text{Causal Structure}
\not\Rightarrow
\text{Unique Presentation Sequence}.
} Causal Structure ⇒ Unique Presentation Sequence .
149. Proposition 3 — Same Endpoint Does Not Determine Worldline
存在:
γ 1 ( t f ) = γ 2 ( t f ) \gamma_1(t_f)
=
\gamma_2(t_f) γ 1 ( t f ) = γ 2 ( t f )
但:
γ 1 ≠ γ 2 . \gamma_1
\neq
\gamma_2. γ 1 = γ 2 .
所以 endpoint-only projection 不能完整代表 execution history。
150. Proposition 4 — Multi-Temporal Projection Does Not Require Temporal Synchrony
只要 cross-domain synchronization contracts 在 projection scope 中被滿足,就可以同時投影:
D i ( t i ) D_i(t_i) D i ( t i )
與:
D j ( t j ) D_j(t_j) D j ( t j )
即使:
t i ≠ t j . t_i\neq t_j. t i = t j .
151. Proposition 5 — Logical Spacetime Availability Does Not Require Full Historical Materialization
若:
temporal anchors stable;
branch lineage known;
causal scope index available;
required regions materialized;
則:
L o g i c a l V i s i b l e ( Y S T ) = 1 \boxed{
\mathsf{LogicalVisible}(Y^{ST})=1
} LogicalVisible ( Y S T ) = 1
可以同時:
ρ h i s t o r y < 1. \rho_{history}<1. ρ hi s t or y < 1.
152. Proposition 6 — Causal Incomparability Does Not Guarantee Safe Parallelism
若:
u ∥ C v , u\parallel_C v, u ∥ C v ,
仍需:
C o m p a t i b l e r e s o u r c e ∧ C o m p a t i b l e s t a t e . Compatible_{resource}
\land
Compatible_{state}. C o m p a t ib l e r eso u r ce ∧ C o m p a t ib l e s t a t e .
因此 antichain 只能作 potential concurrency signal。
153. 對「一口氣看到」的最終修正
Paper 00–06 的「一口氣看到」主要是:
一次取得完整 structured artifact / visual world。
Paper 07 加上:
一次取得完整 causal-temporal structure 。
154. 一口氣看到不只是一張大圖
真正可以一次取得:
State + Worldlines + Causal DAG + Local Times + Branches + Replay + Critical Path . \boxed{
\text{State}
+
\text{Worldlines}
+
\text{Causal DAG}
+
\text{Local Times}
+
\text{Branches}
+
\text{Replay}
+
\text{Critical Path}.
} State + Worldlines + Causal DAG + Local Times + Branches + Replay + Critical Path .
155. 從 Output 到 World Inspection
因此 PNCW 不再只回答:
AI 最後輸出什麼?
而開始回答:
observer 要如何查看 AI / software world 正在怎麼演化、為什麼演化、還有哪些可選分支?
156. High-Dimensional 不再是唯一重點
原先問題:
H i g h D → S e q u e n c e . HighD
\rightarrow
Sequence. H i g h D → S e q u e n ce .
現在更完整:
H i g h D i m e n s i o n a l + M u l t i T e m p o r a l + C a u s a l + B r a n c h e d + C r o s s D o m a i n → Observer Projection . \boxed{
\begin{aligned}
&HighDimensional\\
+&MultiTemporal\\
+&Causal\\
+&Branched\\
+&CrossDomain
\\
&\rightarrow
\text{Observer Projection}.
\end{aligned}
} + + + + H i g h D im e n s i o na l M u l t i T e m p or a l C a u s a l B r an c h e d C r ossD o main → Observer Projection .
157. 真正的 Information-Geometry Mismatch
當 source 是:
( S , G C , B ) , (\mathfrak S,G_C,\mathcal B), ( S , G C , B ) ,
而唯一 presentation 是:
( y 1 , … , y n ) , (y_1,\ldots,y_n), ( y 1 , … , y n ) ,
問題不是 sequence「錯」,而是它只是 source geometry 的一種低自由度 projection。
158. Sequence Remains Valid
文本、speech、logs 仍非常重要。
所以:
Sequence remains a first-class spacetime projection . \boxed{
\text{Sequence}
\text{ remains a first-class spacetime projection}.
} Sequence remains a first-class spacetime projection .
159. 但 Sequence 不再是唯一視角
Observer 應能選:
narrative;
causal DAG;
timeline;
branch tree;
worldline;
multi-time canvas;
hybrid。
160. PNCW Runtime Architecture v2 Form
Paper 07 後,完整 read-side pipeline 可寫:
S → S p a c e t i m e S c o p e Ω q , O S T → C o n t e x t P r o j e c t i o n C q a c t i v e → S P E T F r e e z e E k → H D S R C P k → M R M I C / N V C L V q , k S T → V i s i b i l i t y C o m m i t U q , k S T . \boxed{
\begin{aligned}
\mathfrak S
&\xrightarrow{\mathsf{SpacetimeScope}}
\Omega_{q,O}^{ST}\\
&\xrightarrow{\mathsf{ContextProjection}}
C_{q}^{active}\\
&\xrightarrow{\mathsf{SPETFreeze}}
\mathcal E_k\\
&\xrightarrow{\mathsf{HDSRC}}
P_k\\
&\xrightarrow{\mathsf{MRMIC/NVCL}}
V_{q,k}^{ST}\\
&\xrightarrow{\mathsf{VisibilityCommit}}
U_{q,k}^{ST}.
\end{aligned}
} S SpacetimeScope Ω q , O S T ContextProjection C q a c t i v e SPETFreeze E k HDSRC P k MRMIC/NVCL V q , k S T VisibilityCommit U q , k S T .
161. Vertical Slice — Paper 07 Extension
在原 PNCW MVP 上新增:
至少 3 個 local temporal domains;
live / replay / speculative temporal classes;
causal DAG;
one critical path;
one antichain region;
snapshot + branch;
multi-time Canvas;
backward causal cone query;
branch comparison;
atomic spacetime reveal。
162. Benchmark 1 — Linear Narrative vs Causal DAG
比較使用者理解:
dependency;
parallelism;
blockers;
branch points。
163. Benchmark 2 — Single-Time vs Multi-Time View
測試多 domain 不同 local times 是否:
降低 context switching;
提升 anomaly understanding;
降低時間混淆。
164. Benchmark 3 — Full History vs Causal Cone
比較:
FullReplayMaterialization \text{FullReplayMaterialization} FullReplayMaterialization
與:
K − ( t a r g e t ) K^{-}(target) K − ( t a r g e t )
在:
bytes;
latency;
comprehension;
diagnosis accuracy;
上的差異。
165. Benchmark 4 — Sequential Branch Description vs Branch Canvas
比較:
A narrative
B narrative
C narrative
與:
B r a n c h E n s e m b l e V i e w . BranchEnsembleView. B r an c h E n se mb l e V i e w .
166. Metrics
定義:
M S T = ⟨ T e m p o r a l C o v e r a g e , C a u s a l R e c a l l , C a u s a l P r e c i s i o n , B r a n c h C o v e r a g e , H i s t o r y M a t e r i a l i z e d F r a c t i o n , T i m e T o C a u s a l U n d e r s t a n d i n g , T i m e T o B r a n c h C o m p a r i s o n , C r i t i c a l P a t h A c c u r a c y , T e m p o r a l C o n f u s i o n R a t e ⟩ . \boxed{
\mathbf M_{ST}
=
\left\langle
TemporalCoverage,
CausalRecall,
CausalPrecision,
BranchCoverage,
HistoryMaterializedFraction,
TimeToCausalUnderstanding,
TimeToBranchComparison,
CriticalPathAccuracy,
TemporalConfusionRate
\right\rangle.
} M S T = ⟨ T e m p or a l C o v er a g e , C a u s a l R ec a l l , C a u s a l P r ec i s i o n , B r an c h C o v er a g e , H i s t or y M a t er ia l i z e d F r a c t i o n , T im e T o C a u s a l U n d er s t an d in g , T im e T o B r an c h C o m p a r i so n , C r i t i c a l P a t h A cc u r a cy , T e m p or a l C o n f u s i o n R a t e ⟩ .
167. Failure Condition — Temporal Collapse
若所有 domains 最後仍被強迫:
t i = τ , t_i=\tau, t i = τ ,
則 multi-temporal projection價值消失。
168. Failure Condition — Causal Collapse
若 UI 只按 timestamp 排序,不保留 causal edge:
Causal Projection Failed . \boxed{
\text{Causal Projection Failed}.
} Causal Projection Failed .
169. Failure Condition — Branch Collapse
若 speculative / replay branch 與 live world 無 identity boundary:
Branch Governance Failed . \boxed{
\text{Branch Governance Failed}.
} Branch Governance Failed .
170. Failure Condition — Full-History Materialization
若每次 causal query 都 load 全 history:
Selective Spacetime Materialization Failed . \boxed{
\text{Selective Spacetime Materialization Failed}.
} Selective Spacetime Materialization Failed .
171. Failure Condition — Temporal Label Loss
若 Canvas panel 無法辨識 local time / branch / replay status,multi-time view 可能反而造成錯誤理解。
172. Failure Condition — Causal Overclaim
AI inferred edge 若未標記 evidence class,可能把 hypothesis 當 hard dependency。
因此:
Causal Visualization requires epistemic labeling . \boxed{
\text{Causal Visualization}
\text{ requires epistemic labeling}.
} Causal Visualization requires epistemic labeling .
173. 與 Paper 08 的接口
Paper 07 完成:
What kind of world is being projected? \boxed{
\text{What kind of world is being projected?}
} What kind of world is being projected?
答案:
一個 multi-temporal、causal、branched software spacetime。
Paper 08 接下來才處理:
How is that world perceived, remembered, attended to, acted upon, and verified? \boxed{
\text{How is that world perceived, remembered, attended to, acted upon, and verified?}
} How is that world perceived, remembered, attended to, acted upon, and verified?
174. Paper 08 的預留鏈
下一篇預留:
S → Machine Observation → Cross-Structural Perceptual State → Memory → Attention → Cognition → Projection → Action → S ′ . \boxed{
\mathfrak S
\rightarrow
\text{Machine Observation}
\rightarrow
\text{Cross-Structural Perceptual State}
\rightarrow
\text{Memory}
\rightarrow
\text{Attention}
\rightarrow
\text{Cognition}
\rightarrow
\text{Projection}
\rightarrow
\text{Action}
\rightarrow
\mathfrak S'.
} S → Machine Observation → Cross-Structural Perceptual State → Memory → Attention → Cognition → Projection → Action → S ′ .
但本文不提前形式化其 actuation semantics。
175. 系列位置更新
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 P 07 : Projection-Native Software Spacetime P 08 : Projection-Native Perception–Action Loop \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}\\
P07 &: \text{Projection-Native Software Spacetime}\\
P08 &: \text{Projection-Native Perception–Action Loop}
\end{aligned}
} P 00 P 01 P 02 P 03 P 04 P 05 P 06 P 07 P 08 : 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 : Projection-Native Software Spacetime : Projection-Native Perception–Action Loop
176. 結論
PNCW Paper 00–06 將「AI 結果」從 token stream 提升為:
structured projected computational world . \boxed{
\text{structured projected computational world}.
} structured projected computational world .
本文再往下一層追問:
那個被投影的 world 究竟是什麼?
答案不應只是一個:
W t . W_t. W t .
更一般地,它是:
S = States + Local Times + Events + Causality + Worldlines + Branches + Observers + Resources + History . \boxed{
\mathfrak S
=
\text{States}
+
\text{Local Times}
+
\text{Events}
+
\text{Causality}
+
\text{Worldlines}
+
\text{Branches}
+
\text{Observers}
+
\text{Resources}
+
\text{History}.
} S = States + Local Times + Events + Causality + Worldlines + Branches + Observers + Resources + History .
因此:
State Projection ⊂ Spacetime Projection . \boxed{
\text{State Projection}
\subset
\text{Spacetime Projection}.
} State Projection ⊂ Spacetime Projection .
一個 snapshot view 只是最簡單的 projection。
Observer 還可以要求:
一段 worldline;
一個 causal cone;
一條 critical path;
一組 parallel antichain events;
replay / live comparison;
speculative branch ensemble;
多個 local temporal domains 的同時狀態。
本文同時確立:
Serialization Order ≠ Necessary Causal Order . \boxed{
\text{Serialization Order}
\neq
\text{Necessary Causal Order}.
} Serialization Order = Necessary Causal Order .
所以一條 token stream 可以是合法 narrative projection,但它不是世界因果結構本身。
真正的 PNCW 因此不是:
High-D State → Big Image . \text{High-D State}
\rightarrow
\text{Big Image}. High-D State → Big Image .
而是:
High-Dimensional + Multi-Temporal + Causal + Branched + Cross-Domain Software Spacetime → Task-Relative Projected World . \boxed{
\begin{aligned}
&\text{High-Dimensional}\\
+&\text{Multi-Temporal}\\
+&\text{Causal}\\
+&\text{Branched}\\
+&\text{Cross-Domain Software Spacetime}
\\
&\rightarrow
\text{Task-Relative Projected World}.
\end{aligned}
} + + + + High-Dimensional Multi-Temporal Causal Branched Cross-Domain Software Spacetime → Task-Relative Projected World .
這也重新定義了「一口氣看到」。
它不只是一次看到一份完整文件。
它可以是:
一次看見目前世界、它如何形成、哪些路徑互相獨立、真正 blocking 的 critical chain、哪些 speculative futures 尚未 commit,以及不同 software domains 此刻各自位於哪一個 local time。
因此 Paper 07 的核心母命題是:
A computational world need not be projected as a single state slice or a single temporal sequence. \boxed{
\text{A computational world need not be projected
as a single state slice
or a single temporal sequence.}
} A computational world need not be projected as a single state slice or a single temporal sequence.
以及:
Projection-Native Computation must eventually become Projection-Native Spacetime . \boxed{
\text{Projection-Native Computation}
\text{ must eventually become }
\text{Projection-Native Spacetime}.
} Projection-Native Computation must eventually become Projection-Native Spacetime .
這是 PNCW 從非序列 output architecture 進一步走向完整 software-world observation architecture 的關鍵一步。
內部理論與工程血統
本文主要承接:
PNCW Paper 00–06;
Software Spacetime: From Single Execution Time to Multi-Temporal Computing;
Multi-Temporal Computing and Adaptive Time;
Software Causal Topology and Computational Efficiency;
Fractal AI Spacetime Governance;
PHOSPHOR Spacetime architecture lineage;
CTCL causal / temporal ledger concepts;
GCM computation / observation / materialization separation;
SPET stable projection epoch;
HDSRC projected-native carrier;
MRMIC/NVCL recursive visual computational world。
本文保留 Software Spacetime 原系列的 claim boundary:Software Spacetime 是工程抽象而非物理時空同一性;multi-temporal speed control 不等於硬體性能憑空放大;causal incomparability 不等於安全平行;speculative branch 不等於 world commit;AI policy 不等於無界 authority。