# Grassmannian Geometry of Quantum Causal Inference, Counterfactual Semantics, Constructor Theory, and the Mpemba / Entropy-Siphoning Laws in Lean 4 and Rust/WASM
**Author**: Wes Turner
**Companion Lean 4 Formalization**: [`src/signals/Signals/QuantumCausalInference.lean`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/QuantumCausalInference.lean), [`src/signals/Signals/Plabic.lean`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/Plabic.lean), [`src/signals/Signals/LayerCodes.lean`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/LayerCodes.lean), [`src/signals/Signals/Pending.lean`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/Pending.lean), [`src/signals/SignalsPending.lean`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/SignalsPending.lean)
**Companion Rust / WASM Implementation**: [`src/signals/numerical_adapter/src/quantum_causal_inference.rs`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/numerical_adapter/src/quantum_causal_inference.rs)
**Companion Papers**:
- [Unified Grassmannian Synthesis & Roadmap (`grassmannian-unification-applications-roadmap-paper.md`)](grassmannian-unification-applications-roadmap-paper.md)
- [Beyond $6{,}400\times$ Volumetric Efficiency: Jeffrey's Kinematic Engine, Normed Tensor Gaussian Splat Nanoimaging & Surgical Navigation, and Quantum Causal Semantic GraphRAG (`grassmannian-quantum-causal-splat-nanoimaging-surgical-graphrag-paper.md`)](grassmannian-quantum-causal-splat-nanoimaging-surgical-graphrag-paper.md)
- [Quantum Causal Grassmannian Transformers, JEV vs. $\operatorname{do}$-Calculus Decision Models, Tautological Fiber Bundles, & Spatiotemporal VideoGen (`grassmannian-quantum-causal-jev-bundle-transformer-videogen-paper.md`)](grassmannian-quantum-causal-jev-bundle-transformer-videogen-paper.md)
- [Complex Grassmannian $Gr_{\mathbb{C}}(2,4)$ Fubini–Study GPE, Dual-Schubert `iGPE`, & Homogeneous Plücker $\mathcal{Q}_4$ Madelung Hydrodynamics (`grassmannian-complex-gpe-igpe-plucker-madelung-paper.md`)](grassmannian-complex-gpe-igpe-plucker-madelung-paper.md)
- [Multi-Spectral & Multi-Array Grassmannian Imaging & 3D Layer-Coding `Plabic` Provenance (`grassmannian-multispectral-hubble-medical-nanoimaging-paper.md`)](grassmannian-multispectral-hubble-medical-nanoimaging-paper.md)
- [Causal Second-Sound Thermal Conduction, Inverse Compton, & Raman Thermometry (`grassmannian-thermal-compton-raman-paper.md`)](grassmannian-thermal-compton-raman-paper.md)
- [Huygens–Steiner Double-Slit, OAM, & Plabic LGV Networks (`grassmannian-huygens-double-slit-oam-paper.md`)](grassmannian-huygens-double-slit-oam-paper.md)
---
## Abstract
Does **Grassmannian geometry** govern causal relations, counterfactual interventions, and quantum causal inference? And how do the four thermodynamic and causal frameworks discussed in [`data/chats/_Mpemba Effect_ Landau vs. Quantum .md`](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Mpemba%20Effect_%20Landau%20vs.%20Quantum%20.md)—**(1) Landau Free-Energy Landscapes** {cite:p}`holtzman2022landau`, **(2) Quantum Resource Theory (QRT) & Thermo-Majorization** {cite:p}`summer2026resource`, **(3) Constructor Theory of Thermodynamics & Information** ({cite:p}`deutsch2013constructor`; {cite:p}`marletto2015info`), and **(4) Quantum Causal Models & Counterfactual Semantics** ({cite:p}`giarmatzi2023semantics`; {cite:p}`oreshkov2012quantumCorrelations`)—relate to conditional Landauer erasure {cite:p}`delRio2011thermodynamicMeaning` and the chat corpus's proposed **"entropy siphoning"**?
In this paper, we establish the exact mathematical and physical bridges and implement them in Lean 4 ([`Signals.QuantumCausalInference`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/QuantumCausalInference.lean), [`Signals.Pending`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/Pending.lean)) and zero-dependency Rust/WASM ([`quantum_causal_inference.rs`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/numerical_adapter/src/quantum_causal_inference.rs)):
1. **Five Grassmannian Structures in Classical and Quantum Causal Relations**:
- **Acyclic Causal Transfer & Plabic Positroid Cells ($\mathrm{Gr}_{\ge 0}(k, n)$)**: In any finite Directed Acyclic Graph (DAG) with nilpotent direct-effect matrix $W$ ($W^{h+1} = 0$), the total causal path transfer matrix is $T = \sum_{r=0}^h W^r = I + W T$ (`WeightedAcyclicNetwork.transfer_equation`). By the **Sullivant–Talaska–Draisma Trek Rule** {cite:p}`sullivant2010trek` and **Lindström–Gessel–Viennot (LGV) Lemma** (`lgv_two_by_two_det`, `planar_threeChannel_lgv_and_bottleneck_identities`), $k$ causes act on $k$ effects through vertex-disjoint (unconfounded) pathways iff the $k \times k$ Plücker minor $\det T_{A,B} > 0$ (interior of $\mathrm{Gr}_{>0}(k,n)$). Whenever $k$ causal pathways factor through $m < k$ shared confounders or mediators (**d-separation bottleneck**), the Plücker minor vanishes identically ($\det T_{A,B} = 0$, e.g. `singleHubRepresentative_det = 0` for $2\times 2$ through 1 confounder hub and `(A * B).det = 0` for $3\times 3$ through 2 mediators), placing the causal model on a boundary Schubert/positroid cell of $\mathrm{Gr}(k, n)$.
- **Pearl's Do-Calculus (`do(X := x)`) as Plabic Incoming-Edge Surgery & Subspace Projection**: In a confounded diamond ($U \to X, U \to Y, X \to Y$), observational transfer sums the direct causal path $d$ and the backdoor trek $b$ ($C_{\mathrm{obs}} = d + b$, matching `WeightedAcyclicNetwork.diamond_transfer`), whereas $\operatorname{do}(X := x)$ zeroes column $X$ of $W$ ($w_{U \to X} \mapsto 0$), isolating the pure causal effect $C_{\mathrm{do}} = d$ and exposing the confounding bias $C_{\mathrm{obs}} - C_{\mathrm{do}} = b$.
- **Lorentzian & DDF Acoustic Lightcone Causality on $\mathrm{Gr}_{\mathbb{C}}(2,4)$**: In Penrose twistor theory and our Dilatant Dark Fluid (`Signals.DDF`), Minkowski spacetime is the Klein quadric of $\mathrm{Gr}_{\mathbb{C}}(2,4)$, and subluminal causal propagation ($|\beta| = |v|/c < 1$) is equivalent to the strict positivity of the Plücker channel determinant product $\|w_A^{\mathbb{C}}(M_{\mathrm{cone}})\|^2 \|w_B^{\mathbb{C}}(M_{\mathrm{cone}})\|^2 = 1 - \beta^2 > 0$, vanishing on the null horizon ($|\beta| = 1$).
- **Choi–Jamiołkowski Quantum Causal Subspaces on $\mathrm{Gr}_{\mathbb{C}}(k, d_{\mathrm{in}} d_{\mathrm{out}})$**: A rank-$k$ quantum instrument or isometric channel $V : \mathcal{H}_{\mathrm{in}} \to \mathcal{H}_{\mathrm{out}}$ corresponds via the Choi–Jamiołkowski isomorphism to a $k$-dimensional subspace in $\mathrm{Gr}_{\mathbb{C}}(k, d_{\mathrm{in}} d_{\mathrm{out}})$ of real dimension $2 k (d_{\mathrm{in}} d_{\mathrm{out}} - k)$ ($\dim_{\mathbb{R}} \mathrm{Gr}_{\mathbb{C}}(4, 16) = 96$ for 2-qubit unitary causal channels).
- **Indefinite Causal Order (`Quantum Switch`) on $\mathrm{Gr}_{\mathbb{C}}(2,4)$**: Superposing causal orders $A \prec B$ ($U_B U_A$) and $B \prec A$ ($U_A U_B$) projects onto the symmetric anticommutator branch $E_+ = \frac{1}{4}\|\{U_A, U_B\}\psi\|^2$ and antisymmetric commutator branch $E_- = \frac{1}{4}\|[U_B, U_A]\psi\|^2$. Encoding the causal-order asymmetry $\delta_{\mathrm{switch}} = \frac{E_+ - E_-}{E_+ + E_-}$ into `ddfThermalGr24Matrix δ_switch θ` recovers the exact commutator-to-anticommutator causal witness $\frac{\|w_B^{\mathbb{C}}\|^2}{\|w_A^{\mathbb{C}}\|^2} = \frac{E_-}{E_+}$.
2. **Comparison of the Four Mpemba & Counterfactual Frameworks (`_Mpemba Effect_ Landau vs. Quantum .md`)**:
- We formalize and compare Landau quartic potentials and Strong Mpemba slow-mode orthogonal projection ($a_{2,\mathrm{hot}} = 0$ on the Grassmannian codimension-1 subspace $v_2^\perp$), QRT non-equilibrium free energy $\Delta F = k_B T \ln 2 \cdot D_{\mathrm{KL}}(\rho \| \gamma_\beta)$ and piecewise-linear Lorenz thermo-majorization ($\rho \succ_{\mathrm{th}} \sigma$), Constructor Theory task algebra (`ConstructorTask.compose`, side-information task unlocking), and Giarmatzi–Costa (`arXiv:2302.11783`) Abduction–Action–Prediction counterfactuals.
3. **Thermodynamic Critique & Numerical Audit of Chat "Entropy Siphoning"**:
- We prove the single-shot conditional Landauer formula $Q_{\mathrm{erasure}}(F) = k_B T \ln 2 \cdot (1 - 2F)$ {cite:p}`delRio2011thermodynamicMeaning` and its interventional counterfactual gain $\mathcal{Q}^* = Q(\operatorname{do}(F=1)) - Q(\operatorname{do}(F=0)) = -2 k_B T \ln 2 < 0$, while proving three critical physical and numerical corrections to the chat corpus:
1. **Closed-Loop Second-Law Compliance**: Including the observer memory reset ($Q_{\mathrm{reset}} \ge 2 F \, k_B T \ln 2$), the net cyclic heat is $Q_{\mathrm{closed}} = k_B T \ln 2 + Q_{\mathrm{overhead}} \ge k_B T \ln 2 > 0$.
2. **Isotropic Werner State Visibility ($p_* \approx 0.747614$) vs. Bell Singlet Fidelity ($F_* \approx 0.810710$) and Rank-2 Linear Conditional Entropy**: For an isotropic 2-qubit Werner state $\rho_W(p) = p|\Phi^+\rangle\langle\Phi^+| + (1-p)\frac{I_4}{4}$ with Bell singlet fidelity $F(p) = \frac{3p+1}{4}$, negative conditional entropy $S(A \mid O)_{\mathrm{Werner}} < 0$ requires Werner visibility $p > p_* \approx 0.747614$, which corresponds to Bell singlet fidelity **$F > F_* = \frac{3 p_* + 1}{4} \approx 0.810710$** (`werner_cooling_threshold_fidelity`). At the chat's claimed threshold $F = 1/2$ ($p = 1/3$), the exact conditional entropy is $S(A \mid O)_{\mathrm{Werner}}(1/2) = \frac{\ln 3}{2\ln 2} = \frac{1}{2}\log_2 3 \approx +0.7925\text{ bits} > \frac{1}{2} > 0$ (`wernerConditionalEntropyBitsAtHalf`), whereas the chat's linear formula $1 - 2F$ applies only to a rank-2 Bell mixture.
3. **$6.67 \times 10^7\times$ Numerical Audit of `lignolux_entropy_siphon_sim.py`**: In [`data/chats/_Quantum Processor and Soliton Discussions .md`, lines 4480–4525](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Quantum%20Processor%20and%20Soliton%20Discussions%20%20.md#L4480-L4525), the chat's own Python script (`T = 4.0 K`, `F = 0.99`, `n_ancillas = 2e8`, `qec_frequency = 1e8 Hz`) evaluates to **$P_{\mathrm{siphon}} = -7.50 \times 10^{-7}\text{ W} = -0.750\text{ }\mu\text{W}$** (and even at $1\text{ THz}$, $-7.50\text{ mW}$)—not the `-50 to -100 Watts` claimed in the chat prose!
4. **Beyond $6{,}400\times$ Volumetric Splat Efficiency, Jeffrey's Kinematic Engine, 3D Nanoimaging/Surgical Navigation, and Quantum Causal Semantic GraphRAG** (Section 5 & Master Theorem 7, [companion paper](grassmannian-quantum-causal-splat-nanoimaging-surgical-graphrag-paper.md)):
- We extend our Grassmannian quantum causal framework to **splat-sparse 3D volumetric mixing** (`splatSparseGrassmannVideoComplexity`, `volumetricAttentionEfficiencyRatio`), proving exact speedups of **$13{,}107.2\times$**, **$104{,}857.6\times$**, and **$1{,}310{,}720\times$** (all strictly $> 6{,}400\times$) over dense 3D spatiotemporal self-attention; formalize **Richard Jeffrey's Probability Kinematics (`Jeffrey's Kinematic Engine`)** with Pearl $\operatorname{do}$-deconfounding (`jeffreyKinematicUpdate`, `jeffreyKinematicShift`) and **Quantum Lüders–Plücker interference** (`quantumJeffreyPluckerKinematicUpdate`); prove **Jeffrey Geometric Rigidity**, $L^1$ mass conservation under adaptive major-axis splitting, and $\mathrm{Gr}_{\mathbb{C}}(2,4)$ depth variance recovery on **3D Normed Tensor Gaussian Splats (`NormedTensorGaussianSplat3D`)** for sub-nanometer nanoimaging and surgical navigation; and formulate **Quantum Causal Semantic GraphRAG** (`quantumCausalGraphRAGScore`) combining De Clercq tautological subspace filtering, LGV hub deconfounding, and Quantum Switch multi-hop ordering.
---
## Plain-Language Overview: How the Mpemba Effect, Phase-Diagram Complexity, and Entropy Siphoning Connect to Quantum Causal Inference (and Why They Share One Paper)
At first glance, **material phase diagrams and the Mpemba effect** (why hot water or a hot spin system can freeze/relax faster than a cold one), **entropy siphoning** (cooling a qubit during erasure using quantum entanglement), and **quantum causal inference** (Pearl's $\operatorname{do}$-calculus and counterfactuals on quantum networks) look like three separate fields. In plain terms, here is how they fit together—and when it makes sense to treat them in a single unified paper:
### 1. Why Material Phase-Diagram Complexity and the Mpemba Effect Are a Causal Inference Problem
- **Why simple materials cannot show a Mpemba effect**: Imagine a single-lane marble ramp representing temperature $T$. If a "hot" marble starts at $80^\circ\text{C}$ and a "cold" marble starts at $30^\circ\text{C}$ on the same single-lane track, the $80^\circ\text{C}$ marble *must* roll past $30^\circ\text{C}$ before reaching $0^\circ\text{C}$. It can never overtake the colder marble.
- **How phase-diagram complexity opens multi-lane shortcuts**: Real materials—such as water on a $3\text{D}$ Pressure–Volume–Temperature ($P\text{--}V\text{--}T$) surface with its anomalous ice-expansion hump and supercooled liquid region, polymers, vitrimers, spin glasses, or quantum condensates near a phase transition ([`_Mpemba Effect_ Landau vs. Quantum .md`, lines 127–288](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Mpemba%20Effect_%20Landau%20vs.%20Quantum%20.md#L127-L288))—are **not** a single-lane ramp. Near a phase transition, the Landau free-energy terrain $F(\eta, T)$ {cite:p}`holtzman2022landau` flattens and splits into **multiple valleys (metastable phases) separated by hills (activation barriers / bottlenecks)**:
- The **cold sample** starts concentrated inside a narrow metastable valley trapped behind a barrier. In linear-algebraic terms, its state has a large component along the **slowest relaxation eigenmode** $v_2$ (decaying slowly at rate $e^{-\lambda_2 t}$).
- The **hot sample** (or a quantum state prepared with a pre-cooling unitary pulse) starts spread across the terrain or rotated into a lane that **bypasses the metastable trap entirely**—its component along the slow mode is zero ($a_{2,\mathrm{hot}} = 0$), so it slides straight down the **fast relaxation mode** $v_3$ at rate $e^{-\lambda_3 t}$ ($\lambda_3 \gg \lambda_2$).
- **Where Quantum Causal Inference enters**:
1. **Correlation vs. Causation ($\operatorname{do}$-Calculus)**: When an experiment sees a hot sample freeze faster than a cold one, a skeptic immediately asks: *Did the initial state preparation cause the speedup, or was there a hidden confounder $U$ (evaporation mass loss, convection currents, container wall nucleation, or bath memory)?* Answering *"What would the relaxation time of this exact system have been if we had intervened to set $\operatorname{do}(\rho_0 := \rho_{\mathrm{cold}})$ instead of $\operatorname{do}(\rho_0 := \rho_{\mathrm{hot}})$ while holding the exogenous bath $U_E$ fixed?"* is a **Pearl Rung-3 counterfactual query** {cite:p}`giarmatzi2023semantics`.
2. **Metastable Traps Are Causal Bottlenecks**: In a causal graph, a metastable phase trap is a **rank-deficient bottleneck** where multiple pathways are forced through a single narrow mediator ($\det T_{A,B} = 0$ on the boundary of the Grassmannian). Preparing the **Strong Mpemba state** is a **causal intervention** ($\operatorname{do}(\text{Pre-Rotation})$) that applies the Grassmannian projector $P_{\perp v_2} \in \mathrm{Gr}_{\mathbb{R}}(N-1, N)$ to sever the causal edge into the slow metastable mode $v_2$.
### 2. Why "Entropy Siphoning" (Conditional Landauer Erasure) Is a Quantum Causal Intervention
- **Standard Landauer erasure is "blind"**: Resetting an unknown qubit $A$ to $|0\rangle$ without knowing its state compresses two possibilities into one, which must dump at least $+k_B T \ln 2$ of waste heat into the hardware bath.
- **Conditional Landauer erasure uses entanglement as a causal side-channel**: If the qubit $A$ being reset is quantum-mechanically entangled with an observer/controller memory $O$ {cite:p}`delRio2011thermodynamicMeaning`, the joint state $\rho_{AO}$ is pure ($S(A,O) = 0$) even though $A$ alone looks random ($S(A) = 1\text{ bit}$). The conditional entropy is negative ($S(A \mid O) < 0$). By conditioning the reset pulse on the quantum memory $O$, the controller coherently steers $A \to |0\rangle$ while **absorbing** up to $-k_B T \ln 2$ of heat from the bath in that single step.
- **Why Quantum Causal Inference is essential here**:
1. **Measuring the Causal Effect of Entanglement**: To prove that cooling was caused by quantum entanglement ($F$) rather than a classical calibration drift, an experiment must measure the **interventional counterfactual gain** ([`_Mpemba Effect_ Landau vs. Quantum .md`, lines 465–535](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Mpemba%20Effect_%20Landau%20vs.%20Quantum%20.md#L465-L535)):
$$
\mathcal{Q}^* = \mathbb{E}[Q \mid \operatorname{do}(F = 1)] - \mathbb{E}[Q \mid \operatorname{do}(F = 0)] = -2\,k_B T \ln 2,
$$
comparing the exact same physical erasure gate with the entanglement link intact ($\operatorname{do}(F = 1)$) versus surgically severed ($\operatorname{do}(F = 0)$).
2. **Catching the "Free-Lunch" Fallacy via the Full Causal Graph**: If you only look at node $A$ (the qubit), it looks like heat vanished ("entropy siphoning"). When you draw the **entire quantum causal graph** (data qubit $A$ + observer memory $O$ + heat bath $B$), causal accounting shows that erasing $A$ consumes the entanglement link and transfers the entropy into $O$. Resetting $O$ to run the next cycle costs at least $+2F\,k_B T \ln 2$ (`closedLoopErasureAndMemoryResetHeat`). Thus, causal inference proves that entropy siphoning is **spatial heat relocation** (moving heat dissipation from the cold data qubit $A$ to a remote ancilla reset sink $O$), not a perpetual-motion machine.
### 3. Does It Make Sense for All These Subjects to Be in the Same Paper?
**Yes—when the question is *how to actively control and verify fast, low-heat state reset in complex quantum/classical matter*, these four subjects are four parts of a single engineering and mathematical pipeline**:
| Question in a Quantum / Material Reset Cycle | Subject That Answers It | Shared Mathematical Object in `Signals` |
|---|---|---|
| **1. What does the material's terrain look like?** (Why are there slow metastable traps and fast lanes near phase transitions?) | **Phase-Diagram Complexity & Landau Mpemba Theory** {cite:p}`holtzman2022landau` | 3D $P\text{--}V\text{--}T$ van der Waals / water ice-hump (`vanDerWaalsPressure`, `waterAnomalousFreezingVolume`), Landau barrier $\Delta F_{\mathrm{barrier}}(T) = \frac{a_0^2(T-T_c)^2}{4b}$ (`landauDoubleWellBarrierHeight`), and projector $P_{\perp v_2} \in \mathrm{Gr}_{\mathbb{R}}(N-1, N)$ |
| **2. Which state shortcuts are allowed by the laws of thermodynamics?** | **Quantum Resource Theory {cite:p}`summer2026resource` & Constructor Theory {cite:p}`marletto2015info`** | Thermo-majorization preorder $\rho \succ_{\mathrm{th}} \sigma$ and task surplus $\Sigma_{\mathrm{task}} \ge 0$ |
| **3. How can quantum entanglement reduce or relocate the heat of a reset step?** | **Conditional Landauer Erasure / "Entropy Siphoning"** {cite:p}`delRio2011thermodynamicMeaning` | Single-shot $Q(F) = k_B T \ln 2\,(1-2F)$, Werner threshold $F_* \approx 0.8107$, and closed-loop reset bound |
| **4. How do we intervene to trigger the shortcut and prove it wasn't a confounding artifact?** | **Quantum Causal Inference & Counterfactual Semantics** ({cite:p}`giarmatzi2023semantics`; {cite:p}`pearl2009causality`) | Plabic/DAG transfer $T = (I-W)^{-1}$, $\operatorname{do}(X)$ surgery, Choi subspaces, $\mathrm{Gr}_{\mathbb{C}}(2,4)$ Quantum Switch, and Master Theorem 4 (`phase_diagram_complexity_mpemba_and_quantum_causal_synthesis_identities`) |
**When should they be treated separately?**
If a reader only needs **classical DAG causal discovery** (e.g., removing sensor confounding via `causal_dag_do_surgery_3x3`) or only needs **classical material phase relaxation** (without quantum entanglement), the modules can be used independently. They belong **together in this paper** because in quantum hardware design (such as QPU ancilla reset and non-equilibrium state preparation) and in the provenance chat corpus ([`_Mpemba Effect_ Landau vs. Quantum .md`](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Mpemba%20Effect_%20Landau%20vs.%20Quantum%20.md)), **bypassing slow phase-space bottlenecks (Mpemba)**, **relocating erasure heat via entanglement (conditional Landauer)**, and **proving causal attribution via $\operatorname{do}$-interventions** are the exact same physical operation viewed from dynamics, thermodynamics, and causal inference, all unified by **Grassmannian subspace projection**.
---
## 1. Is There Grassmannian Geometry in Causal Relations?
Yes—causal relations exhibit **five distinct, mathematically exact Grassmannian structures** spanning classical structural causal models (SCMs), relativistic spacetime causality, and quantum causal processes:
```{mermaid}
flowchart TD
subgraph DAG["1. Classical DAG & Plabic Causal Networks: Gr_>=0(k, n)"]
Transfer["Nilpotent Direct-Effect Matrix W (W^(h+1) = 0)\nTotal Causal Transfer T = (I - W)^(-1) = I + W T\nGuarded Mediator Subdivision: w = s * (w/s)"]
Trek["LGV & Trek Rule for k-Input -> k-Output Minors det T_{A,B}\n• Unconfounded Disjoint Paths: det T_{A,B} > 0 (Interior of Gr_>0(k,n))\n• Shared Confounder / Mediator Bottleneck (m < k): det T_{A,B} = 0\n (Boundary Schubert / Positroid Cell: singleHubRepresentative_det = 0)"]
DoCalc["Pearl's do(X := x) Intervention as Incoming-Edge Surgery\nC_obs = d + b (Direct + Backdoor Trek) -> C_do = d (Pure Causal Effect)"]
Transfer --> Trek --> DoCalc
end
subgraph Lightcone["2. Relativistic & DDF Lightcone Causality: Gr_C(2,4)"]
Cone["Penrose Klein Quadric & DDF Second-Sound Characteristic Cone\nM_cone = ddfThermalGr24Matrix(β, θ) with β = v / c\nPlücker Product |w_A^C|² |w_B^C|² = 1 - β²\n• Timelike Causal (|β| < 1): > 0\n• Null Horizon (|β| = 1): = 0\n• Spacelike Acausal (|β| > 1): < 0"]
end
subgraph QuantumCausal["3. Quantum Causal Models & Indefinite Causal Order"]
Choi["Choi-Jamiołkowski Process Subspaces\nRank-k Quantum Instrument V: H_in -> H_out\nSubspace in Gr_C(k, d_in * d_out)\ndim_R Gr_C(4, 16) = 96 (Two-Qubit Unitary Channels)"]
Switch["Quantum Switch (A ≺ B) + (B ≺ A) on Gr_C(2,4)\nE_+ = (1/4)||{U_A, U_B}ψ||² (Anticommutator)\nE_- = (1/4)||[U_B, U_A]ψ||² (Commutator)\nPlücker Ratio |w_B^C|² / |w_A^C|² = E_- / E_+"]
MpembaProj["Strong Mpemba & Counterfactual Abduction\nOrthogonal Projection P_{⊥ v_2} onto Fast Liouvillian Subspace\n(a_{2,hot} = 0 in Gr_R(N-1, N))"]
Choi --> Switch --> MpembaProj
end
DAG --> Lightcone --> QuantumCausal
```
### 1.1 Acyclic Structural Causal Networks, Trek / LGV Minors, and Pearl's `do(X)` Surgery (`Signals.Plabic` & `Signals.QuantumCausalInference`)
In Pearl's Structural Causal Models {cite:p}`pearl2009causality` and Wright's path analysis on a finite Directed Acyclic Graph (DAG) with $V$ vertices and height bound $h$, let $W_{ij}$ be the direct causal effect of node $i$ on node $j$. Because the graph is acyclic, $W^{h+1} = 0$ (`WeightedAcyclicNetwork.nilpotent`), and the total causal effect matrix summing all directed causal paths of length $0 \le r \le h$ is:
$$
T = \sum_{r=0}^h W^r = (I - W)^{-1}, \qquad T = I + W T
\quad (\text{proved in } \texttt{WeightedAcyclicNetwork.transfer\_equation}).
$$
How does the Grassmannian $\mathrm{Gr}(k, n)$ enter?
- **The LGV / Trek Determinant Criterion for Causal Bottlenecks and Confounding ({cite:p}`sullivant2010trek`; {cite:p}`lindstrom1973lgv`; {cite:p}`gessel1985`)**:
Given $k$ intervention/source variables $A = \{a_1, \dots, a_k\}$ and $k$ outcome variables $B = \{b_1, \dots, b_k\}$, the $k \times k$ submatrix $T_{A,B}$ is a Plücker coordinate of the $k$-plane spanned by the causal response vectors in $\mathrm{Gr}(k, V)$:
1. **Unconfounded Vertex-Disjoint Causal Channels ($\det T_{A,B} > 0$)**: When $k$ causal pathways travel through $k$ independent, non-intersecting mediators (`planarThreeChannelLowerStep * planarThreeChannelUpperStep`), all $k!$ non-identity path permutations are blocked or positive, placing the causal system in the interior of the totally positive Grassmannian cell $\mathrm{Gr}_{>0}(k, n)$:
$$
\det\!\left(E_{\mathrm{lower}}^{(3)}(l_{10}, l_{21}) \, E_{\mathrm{upper}}^{(3)}(d_0, d_1, d_2, u_{01}, u_{12})\right) = d_0 d_1 d_2 > 0.
$$
2. **Shared Confounder / Mediator Bottleneck ($\det T_{A,B} = 0$ — d-Separation)**: If $k$ causes influence $k$ outcomes solely through $m < k$ intermediate mediators or a single common-cause confounder hub, the causal transfer matrix factors as $T_{A,B} = A_{k \times m} B_{m \times k}$ and its $k \times k$ Plücker minor vanishes identically:
$$
\det(\operatorname{singleHubRepresentative}(u, x, v, w_{\mathrm{out}})) = 0 \quad (k=2, m=1),
\qquad
\det(A_{3\times 2} B_{2\times 3}) = 0 \quad (k=3, m=2).
$$
Geometrically, **d-separation and unmeasured confounding correspond to degenerating from the open top cell $\mathrm{Gr}_{>0}(k, n)$ onto a lower-rank positroid / Schubert boundary stratum where $\Delta_I = 0$**!
- **Pearl's `do(X := x)` Intervention as Plabic Incoming-Edge Surgery (`causalObservationalTransfer`, `causalInterventionalDoTransfer`, `causal_dag_do_surgery_3x3`)**:
Consider a 3-node confounded DAG with unobserved confounder $U$ (node $0$), treatment $X$ (node $1$), and outcome $Y$ (node $2$), with edge weights $w_{01} = w_{U \to X}$, $w_{02} = w_{U \to Y}$, and $w_{12} = w_{X \to Y} = d$:
$$
W_{\mathrm{obs}} = \begin{pmatrix} 0 & w_{01} & w_{02} \\ 0 & 0 & w_{12} \\ 0 & 0 & 0 \end{pmatrix}
\implies
T_{\mathrm{obs}} = I + W_{\mathrm{obs}} + W_{\mathrm{obs}}^2 = \begin{pmatrix} 1 & w_{01} & w_{02} + w_{01} w_{12} \\ 0 & 1 & w_{12} \\ 0 & 0 & 1 \end{pmatrix}.
$$
In observational data, the total association between $X$ and $Y$ is contaminated by the backdoor confounding trek $b = w_{U \to X} w_{U \to Y}$ (`causalObservationalTransfer d b = d + b`, matching `WeightedAcyclicNetwork.diamond_transfer`). Pearl's intervention $\operatorname{do}(X := x)$ performs **graph surgery** by zeroing column $1$ of $W$ ($w_{01} \mapsto 0$), yielding:
$$
W_{\mathrm{do}(X)} = \begin{pmatrix} 0 & 0 & w_{02} \\ 0 & 0 & w_{12} \\ 0 & 0 & 0 \end{pmatrix}
\implies
C_{\mathrm{do}} = d = w_{12}, \qquad
C_{\mathrm{obs}} - C_{\mathrm{do}} = b.
$$
### 1.2 Relativistic and DDF Lightcone Causality on $\mathrm{Gr}_{\mathbb{C}}(2,4)$ (`causalLightconeInterval`)
In relativistic physics {cite:p}`penrose1986spinors` and our Dilatant Dark Fluid (`Signals.DDF`, [grassmannian-ddf-moving-sheath-paper.md](grassmannian-ddf-moving-sheath-paper.md)), complexified Minkowski spacetime is the Klein quadric of 2-planes in twistor space $\mathbb{C}^4$, i.e. $\mathrm{Gr}_{\mathbb{C}}(2,4)$. For a signal or thermal second-sound wave propagating at normalized velocity $\beta = v / c$, the $\mathrm{Gr}_{\mathbb{C}}(2,4)$ twistor matrix $M_{\mathrm{cone}} = \operatorname{ddfThermalGr24Matrix}(\beta, \theta)$ has complex Plücker channel amplitudes satisfying:
$$
\|w_A^{\mathbb{C}}(M_{\mathrm{cone}})\|^2 = 1 + \beta, \qquad
\|w_B^{\mathbb{C}}(M_{\mathrm{cone}})\|^2 = 1 - \beta, \qquad
\|w_A^{\mathbb{C}}(M_{\mathrm{cone}})\|^2 \, \|w_B^{\mathbb{C}}(M_{\mathrm{cone}})\|^2 = 1 - \beta^2 = \frac{d\tau^2}{dt^2}.
$$
Thus:
- **Timelike Causal Interior ($|\beta| < 1$)**: Both Plücker channel norms are strictly positive and $\|w_A^{\mathbb{C}}\|^2 \|w_B^{\mathbb{C}}\|^2 = 1 - \beta^2 > 0$.
- **Null Lightcone / Horizon Boundary ($|\beta| = 1$)**: One Plücker channel vanishes ($\|w_B^{\mathbb{C}}\|^2 = 0$), degenerating onto a rank-deficient boundary of the positive Grassmannian (`causalLightconeInterval 1 = 0`).
- **Spacelike Acausal Exterior ($|\beta| > 1$)**: The interval $1 - \beta^2 < 0$ flips sign, violating positivity.
---
## 2. Quantum Causal Models, Counterfactual Semantics (`arXiv:2302.11783`), and Indefinite Causal Order (`Quantum Switch`)
### 2.1 Quantum Causal Models and Giarmatzi–Costa Counterfactual Semantics (`arXiv:2302.11783`)
In classical Pearl SCMs {cite:p}`pearl2009causality`, a counterfactual query *"Given that we observed $(X = x_{\mathrm{obs}}, Y = y_{\mathrm{obs}})$, what would $Y$ have been had we intervened $\operatorname{do}(X := x_{\mathrm{cf}})$?"* (Rung 3 of Pearl's Ladder of Causation) is evaluated by the three-step **Abduction–Action–Prediction** algorithm:
1. **Abduction**: Update the distribution over the exogenous background variables $U_E$ from the factual evidence $(x_{\mathrm{obs}}, y_{\mathrm{obs}})$: for $Y = \alpha X + U_E$, $u_E = y_{\mathrm{obs}} - \alpha x_{\mathrm{obs}}$ (`abducedExogenousState`).
2. **Action**: Replace the structural mechanism for $X$ by the intervention $\operatorname{do}(X := x_{\mathrm{cf}})$.
3. **Prediction**: Compute the counterfactual outcome in the modified model using the abduced $u_E$:
$$
Y_{x_{\mathrm{cf}}}(u_E) = \alpha x_{\mathrm{cf}} + (y_{\mathrm{obs}} - \alpha x_{\mathrm{obs}}) = y_{\mathrm{obs}} + \alpha(x_{\mathrm{cf}} - x_{\mathrm{obs}}), \qquad
\Delta Y_{\mathrm{cf}} = Y_{x_{\mathrm{cf}}} - Y_{x_{\mathrm{obs}}} = \alpha(x_{\mathrm{cf}} - x_{\mathrm{obs}}).
$$
Why is extending counterfactuals to quantum mechanics subtle? By Bell's theorem and the Kochen–Specker theorem, quantum systems do not possess pre-existing deterministic values for unperformed measurements. In **Giarmatzi & Costa (2023)** ([`arXiv:2302.11783`](#giarmatzi2023semantics)) and **Allen, Barrett, Horsman, Lee, & Spekkens (2017)** {cite:p}`allen2017qcm`, a **Quantum Causal Model (QCM)** resolves this by shifting causal determinism from hidden outcome values to **unitary channels and purified exogenous environment ancillas**:
- Every quantum node $A$ is governed by a unitary interaction $U_A : \mathcal{H}_{A_{\mathrm{in}}} \otimes \mathcal{H}_{E_A} \to \mathcal{H}_{A_{\mathrm{out}}} \otimes \mathcal{H}_{F_A}$ with an exogenous auxiliary state $\rho_{E_A}$.
- Under the **Choi–Jamiołkowski isomorphism**, a rank-$k$ quantum instrument or isometric channel between $\mathcal{H}_{\mathrm{in}}$ ($\dim = d_{\mathrm{in}}$) and $\mathcal{H}_{\mathrm{out}}$ ($\dim = d_{\mathrm{out}}$) is a $k$-dimensional subspace in $\mathrm{Gr}_{\mathbb{C}}(k, d_{\mathrm{in}} d_{\mathrm{out}})$, whose real manifold dimension is (`quantumChoiGrassmannianRealDim`):
$$
\dim_{\mathbb{R}} \mathrm{Gr}_{\mathbb{C}}(k, d_{\mathrm{in}} d_{\mathrm{out}}) = 2 k (d_{\mathrm{in}} d_{\mathrm{out}} - k)
\qquad (\dim_{\mathbb{R}} \mathrm{Gr}_{\mathbb{C}}(4, 16) = 96 \text{ for } k = d_{\mathrm{in}} = d_{\mathrm{out}} = 4).
$$
- In a quantum counterfactual {cite:p}`giarmatzi2023semantics`, **Abduction** projects the exogenous environment state $\rho_E$ onto the subspace compatible with the factual instrument outcome via a Grassmannian projector $P_E \in \mathrm{Gr}_{\mathbb{C}}(k_E, \dim \mathcal{H}_E)$, **Action** replaces the channel at node $X$ with a counterfactual intervention channel $\operatorname{do}(X := \mathcal{I}_{x_{\mathrm{cf}}})$, and **Prediction** contracts the updated exogenous state through the modified quantum comb / process matrix!
### 2.2 Indefinite Causal Order and the Quantum Switch on $\mathrm{Gr}_{\mathbb{C}}(2,4)$ (`quantumSwitchCausalAsymmetry`, `evaluate_quantum_switch_2x2`)
In standard causal DAGs, either event $A$ precedes $B$ ($A \prec B$) or $B$ precedes $A$ ($B \prec A$). In quantum mechanics, **Oreshkov, Costa, & Brukner (2012)** {cite:p}`oreshkov2012quantumCorrelations` and **Chiribella, D'Ariano, Perinotti, & Valiron (2013)** {cite:p}`chiribella2013switch` showed that quantum operations $U_A, U_B$ can be placed in a coherent superposition of causal orders via the **Quantum Switch**:
$$
\mathcal{S}(U_A, U_B)\!\left(\frac{|0\rangle_c + |1\rangle_c}{\sqrt{2}} \otimes |\psi\rangle\right)
= \frac{1}{\sqrt{2}}|0\rangle_c (U_B U_A|\psi\rangle) + \frac{1}{\sqrt{2}}|1\rangle_c (U_A U_B|\psi\rangle)
= |+\rangle_c \frac{\{U_A, U_B\}}{2}|\psi\rangle + |-\rangle_c \frac{[U_B, U_A]}{2}|\psi\rangle.
$$
Measuring the control qubit in the $|\pm\rangle_c$ basis projects onto two orthogonal causal interference branches with probabilities:
$$
E_+ = \frac{1}{4}\bigl\|\{U_A, U_B\}|\psi\rangle\bigr\|^2, \qquad
E_- = \frac{1}{4}\bigl\|[U_B, U_A]|\psi\rangle\bigr\|^2, \qquad
E_+ + E_- = 1.
$$
In Theorem `grassmannian_causal_relations_do_calculus_and_quantum_switch_identities` ([`QuantumCausalInference.lean`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/QuantumCausalInference.lean#L230-L395)) and `evaluate_quantum_switch_2x2` ([`quantum_causal_inference.rs`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/numerical_adapter/src/quantum_causal_inference.rs#L130-L185)), we prove that encoding the causal-order asymmetry $\delta_{\mathrm{switch}} = \frac{E_+ - E_-}{E_+ + E_-}$ into the complex Grassmannian $\mathrm{Gr}_{\mathbb{C}}(2,4)$ twistor matrix $M_{\mathrm{sw}} = \operatorname{ddfThermalGr24Matrix}(\delta_{\mathrm{switch}}, \theta)$ yields:
$$
\|w_A^{\mathbb{C}}(M_{\mathrm{sw}})\|^2 = 1 + \delta_{\mathrm{switch}} = \frac{2 E_+}{E_+ + E_-}, \qquad
\|w_B^{\mathbb{C}}(M_{\mathrm{sw}})\|^2 = 1 - \delta_{\mathrm{switch}} = \frac{2 E_-}{E_+ + E_-}, \qquad
\frac{\|w_B^{\mathbb{C}}(M_{\mathrm{sw}})\|^2}{\|w_A^{\mathbb{C}}(M_{\mathrm{sw}})\|^2} = \frac{E_-}{E_+}.
$$
- **Compatible / Commuting Causal Operations ($[U_A, U_B] = 0$, e.g. $U_A = I, U_B = Z$)**: $E_- = 0, E_+ = 1 \implies \delta_{\mathrm{switch}} = +1$ and $\|w_B^{\mathbb{C}}\|^2 = 0$.
- **Anticommuting Pauli Operations ($\{U_A, U_B\} = 0$, e.g. $U_A = X, U_B = Z$)**: $E_+ = 0, E_- = 1 \implies \delta_{\mathrm{switch}} = -1$ and $\|w_A^{\mathbb{C}}\|^2 = 0$, providing a $100\%$ deterministic witness of indefinite causal order!
- **Complex Parallelogram Conservation & Oreshkov–Costa–Brukner (OCB) Causal Inequality Violation (`quantumSwitchSymBranchWeight`, `ocbQuantumIndefiniteBound`)**:
In Theorem `grassmannian_mpemba_projector_werner_threshold_and_ocb_causal_witness_identities`, we further prove that for any complex causal-order path amplitudes $z_{AB}, z_{BA} \in \mathbb{C}$, the symmetric and antisymmetric branch weights obey the exact parallelogram identity $E_+(z_{AB}, z_{BA}) + E_-(z_{AB}, z_{BA}) = \frac{1}{2}(\|z_{AB}\|^2 + \|z_{BA}\|^2)$, and that indefinite causal order violates the Oreshkov–Costa–Brukner (2012) causal inequality:
$$
p_{\mathrm{OCB}}^{\mathrm{causal}} = \frac{3}{4} = 0.75 \;<\; p_{\mathrm{OCB}}^{\mathrm{quantum}} = \frac{2 + \sqrt{2}}{4} \approx 0.85355339, \qquad
\bigl(4\,p_{\mathrm{OCB}}^{\mathrm{quantum}} - 2\bigr)^2 = 2.
$$
---
## 3. Review and Comparison of the Four Mpemba & Counterfactual Models (`_Mpemba Effect_ Landau vs. Quantum .md`)
In [`data/chats/_Mpemba Effect_ Landau vs. Quantum .md`](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Mpemba%20Effect_%20Landau%20vs.%20Quantum%20.md#L40-L495), four frameworks are compared for modeling non-equilibrium relaxation, the Mpemba effect, and counterfactual reasoning. Below is our formal mathematical synthesis and Lean 4 / Rust comparison across all four frameworks:
| Dimension | 1. Landau Free-Energy & Markovian Modes {cite:p}`holtzman2022landau` | 2. Quantum Resource Theory (QRT) of Athermality {cite:p}`summer2026resource` | 3. Constructor Theory of Thermodynamics ({cite:p}`marletto2015info`; [CT FAQ Q3](#deutschmarletto2021constructor)) | 4. Quantum Causal Models (QCMs) {cite:p}`giarmatzi2023semantics` |
|---|---|---|---|---|
| **Fundamental Ontology** | Order-parameter landscape $F(\eta, T)$ & Fokker–Planck / Liouvillian eigenmodes $\{v_k\}$ | Density states $\rho$, Gibbs state $\gamma_\beta$, & Thermal Operations ($\mathcal{E}(\gamma_\beta) = \gamma_\beta$) | Substrate-independent **Tasks** $\mathcal{T} = \{i \to o\}$ & cyclic **Constructors** $C$ | Directed unitary networks $U_A$ / process matrices $W$ & purified exogenous ancillas $\rho_E$ |
| **Mathematical Representation of Free Energy / Resource** | Quartic Landau functional $F(\eta) = \frac{1}{2}a\eta^2 + \frac{1}{4}b\eta^4 - h\eta$; double-well minimum $F(\pm\sqrt{-a/b}) = -\frac{a^2}{4b} < 0$ (`landauFreeEnergy`) | Rényi / KL non-equilibrium free energy $\Delta F = k_B T \ln 2 \cdot D_{\mathrm{KL}}(\rho \| \gamma_\beta)$ & piecewise-linear Lorenz curve (`qrtNonEquilibriumFreeEnergy`, `thermo_majorization_dominates`) | Task thermodynamic surplus $\Sigma_{\mathrm{task}} = W_{\mathrm{sup}} + k_B T \ln 2\,(\Delta S_{\mathrm{sys}} + I_{\mathrm{side}}) \ge 0$ (`ConstructorTask.thermodynamicSurplus`) | Interventional causal capacity $\Delta\mathcal{Q}^* = \mathbb{E}[Q \mid \operatorname{do}(K=1)] - \mathbb{E}[Q \mid \operatorname{do}(K=0)]$ (`entropySiphonCounterfactualGain`) |
| **Mechanism of the Mpemba Effect** | **Slow-mode overlap suppression**: $\|a_2^{\mathrm{hot}}\| < \|a_2^{\mathrm{cold}}\|$ (Weak Mpemba) or $a_2^{\mathrm{hot}} = 0$ (Strong Mpemba exponential speedup at rate $\lambda_3 > \lambda_2$) | **Majorization / Resource Shortcut**: Hotter state (or state after unitary pre-rotation) thermo-majorizes $\gamma_\beta$ along a faster trajectory ($D_{\mathrm{siphon}} < D_{\mathrm{alone}}$) | **Task Possibility Asymmetry**: Task $\mathcal{T}_{\mathrm{cool}}(\tau)$ is possible ($\mathcal{T}^{\checkmark}$) from prepared hot/correlated state within time $\tau$, but impossible ($\mathcal{T}^{\times}$) from trapped cold state | **Causal Path Selection**: $\operatorname{do}(\text{Pre-Rotation / Entanglement})$ severs the slow metastable causal bottleneck (`singleHubRepresentative` $\to$ fast channel) |
| **Semantics of Counterfactuals** | **Alternative Trajectory**: Compare $p_{\mathrm{hot}}(t)$ vs. $p_{\mathrm{cold}}(t)$ on the same potential $F(\eta)$ (`mpembaTwoModeDistance`) | **State-Space Preorder**: $\rho \succ_{\mathrm{th}} \sigma$ checks whether transformation $\rho \to \sigma$ is permitted under Thermal Operations | **First-Class Primitive**: Physical laws are statements of which tasks are Possible ($\mathcal{T}^{\checkmark}$) vs. Impossible ($\mathcal{T}^{\times}$) (`ConstructorTask.isPossible`) | **Abduction–Action–Prediction**: Infer exogenous $\rho_E$ from observation, apply $\operatorname{do}(X := x_{\mathrm{cf}})$, and propagate forward (`counterfactualStructuralOutcome`) |
| **Grassmannian Realization in `Signals`** | Strong Mpemba $a_2^{\mathrm{hot}} = 0$ is an orthogonal projection $P_{\perp v_2}$ onto a codimension-1 subspace in $\mathrm{Gr}_{\mathbb{R}}(N-1, N)$ | Unitary pre-cooling rotates the eigenframe of $\rho_0$ on the complex flag / Grassmannian manifold $\mathrm{Gr}_{\mathbb{C}}(k, N)$ | Parallel/serial task composition corresponds to disjoint vs. serial Plabic network concatenation in `Signals.Plabic` | Choi subspaces in $\mathrm{Gr}_{\mathbb{C}}(k, d_{\mathrm{in}}d_{\mathrm{out}})$ and Quantum Switch Plücker ratios on $\mathrm{Gr}_{\mathbb{C}}(2,4)$ |
### 3.1 Strong Mpemba Crossing Proof (`mpembaTwoModeDistance`, `strong_mpemba_crossing_time`)
Consider a two-eigenmode relaxation process with slow decay rate $\lambda_2 > 0$ and fast decay rate $\lambda_3 > \lambda_2$, so that $w_2(t) = e^{-\lambda_2 t}$ and $w_3(t) = e^{-\lambda_3 t}$ satisfy $0 < w_3(t) < w_2(t) < 1$ for all $t > 0$. In `mpemba_constructor_theory_and_entropy_siphoning_identities` ([`QuantumCausalInference.lean`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/QuantumCausalInference.lean#L430-L515)):
- Let the **cold state** have mode overlaps $(a_{2,\mathrm{cold}}, a_{3,\mathrm{cold}}) = (1, 1)$, and let the **hot state** be prepared in the **Strong Mpemba Grassmannian subspace** orthogonal to the slow mode, $(a_{2,\mathrm{hot}}, a_{3,\mathrm{hot}}) = (0, 4)$:
- At $t = 0$ ($w_2 = w_3 = 1$), the hot state is twice as far from equilibrium as the cold state:
$$
D_{\mathrm{cold}}(0) = 1 \cdot 1 + 1 \cdot 1 = 2 \;<\; D_{\mathrm{hot}}(0) = 0 \cdot 1 + 4 \cdot 1 = 4.
$$
- At a later time $t > 0$ where $w_2(t) = 1/2$ and $w_3(t) = 1/16$ (for $\lambda_3 = 4\lambda_2$), the hot state has bypassed the slow bottleneck and overtaken the cold state:
$$
D_{\mathrm{hot}}(t) = 0 \cdot \frac{1}{2} + 4 \cdot \frac{1}{16} = \frac{1}{4} = \frac{4}{16}
\;<\;
D_{\mathrm{cold}}(t) = 1 \cdot \frac{1}{2} + 1 \cdot \frac{1}{16} = \frac{9}{16}.
$$
- More generally, for a pure slow-mode cold state $(a_{2,\mathrm{cold}}, 0)$ and Strong Mpemba hot state $(0, a_{3,\mathrm{hot}})$ with $a_{3,\mathrm{hot}} > a_{2,\mathrm{cold}} > 0$, the exact trajectory crossing $D_{\mathrm{hot}}(t_{\mathrm{cross}}) = D_{\mathrm{cold}}(t_{\mathrm{cross}})$ occurs at (`strongMpembaCrossingTime`, `strong_mpemba_crossing_time`):
$$
t_{\mathrm{cross}} = \frac{\ln(a_{3,\mathrm{hot}} / a_{2,\mathrm{cold}})}{\lambda_3 - \lambda_2} > 0.
$$
- **Explicit $\mathrm{Gr}_{\mathbb{R}}(1, 2)$ Orthogonal Complement Projector (`mpembaProjector00`, `mpembaProjectedSlowOverlap`, `apply_strong_mpemba_gr12_projector`)**: Given any normalized slow Liouvillian eigenmode $v_{\mathrm{slow}} = (c, s)^{\mathsf{T}}$ with $c^2 + s^2 = 1$, preparing the Strong Mpemba state applies the rank-1 orthogonal complement projector
$$
P_{\perp v_2}(c, s) = I_2 - v_{\mathrm{slow}} v_{\mathrm{slow}}^{\mathsf{T}} = \begin{pmatrix} s^2 & -c s \\ -c s & c^2 \end{pmatrix} \in \mathrm{Gr}_{\mathbb{R}}(1, 2),
$$
which satisfies $\operatorname{Tr} P_{\perp v_2} = s^2 + c^2 = 1$, $\det P_{\perp v_2} = s^2 c^2 - (-c s)^2 = 0$, $P_{\perp v_2}^2 = P_{\perp v_2}$, and $\langle v_{\mathrm{slow}}, P_{\perp v_2} x \rangle = 0$ identically for all initial state vectors $x = (x_0, x_1)^{\mathsf{T}} \in \mathbb{R}^2$.
---
## 4. Review, Critique, and Thermodynamic Audit of "Entropy Siphoning" (`del Rio et al. 2011` vs. Chat Claims)
### 4.1 What Is Mathematically and Physically True: Single-Shot Conditional Landauer Erasure {cite:p}`delRio2011thermodynamicMeaning`
In classical thermodynamics, Landauer's principle {cite:p}`landauer1961irreversibility` states that erasing 1 bit of information at temperature $T$ dissipates at least $+k_B T \ln 2$ of heat into the environment. In **del Rio, Åberg, Renner, Dahlsten, & Vedral (2011)** ([*Nature* 474, 61–63](#delrio2011negative)), this bound is generalized to the quantum regime where the system $A$ being erased is correlated or entangled with a quantum memory $O$ held by the observer:
$$
Q_{\mathrm{erasure}}(A \mid O) = k_B T \ln 2 \cdot S(A \mid O), \qquad S(A \mid O) = S(\rho_{AO}) - S(\rho_O).
$$
In `delRioConditionalLandauerHeat` ([`QuantumCausalInference.lean`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/QuantumCausalInference.lean#L185-L225)), using the linear Bell-mixture parameterization $S_{\mathrm{lin}}(A \mid O) = 1 - 2F$ for Bell fidelity $F \in [0, 1]$, we prove the three canonical regimes and the counterfactual $\operatorname{do}(F)$ intervention gain:
1. **Control ($\operatorname{do}(F = 0)$, Unentangled Classical Bit)**: $S(A \mid O) = +1 \implies Q_{\mathrm{erasure}}(0) = +k_B T \ln 2 > 0$ (standard Landauer heating).
2. **Reversible Threshold ($F = 1/2$, Classical Correlation)**: $S(A \mid O) = 0 \implies Q_{\mathrm{erasure}}(1/2) = 0$ (zero-heat reversible reset).
3. **Maximal Quantum Entanglement ($\operatorname{do}(F = 1)$, Pure Bell Pair $|\Phi^+\rangle_{AO}$)**: $S(\rho_{AO}) = 0$ while $S(\rho_O) = 1$, so $S(A \mid O) = -1 < 0$ and:
$$
Q_{\mathrm{erasure}}(1) = -k_B T \ln 2 < 0, \qquad
\mathcal{Q}^* = Q(\operatorname{do}(F = 1)) - Q(\operatorname{do}(F = 0)) = -2\,k_B T \ln 2 < 0.
$$
In Constructor Theory (`ConstructorTask` in [`QuantumCausalInference.lean`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/QuantumCausalInference.lean#L150-L182)), zero-work 1-bit erasure ($\Delta S_{\mathrm{sys}} = -1, W_{\mathrm{supplied}} = 0$) is **impossible** with $I_{\mathrm{side}} = 0$ ($\Sigma_{\mathrm{task}} = -k_B T \ln 2 < 0$), becomes **reversibly possible** at $I_{\mathrm{side}} = 1$ ($\Sigma_{\mathrm{task}} = 0$), and achieves a **positive cooling surplus** at $I_{\mathrm{side}} = 2$ ($\Sigma_{\mathrm{task}} = +k_B T \ln 2 > 0$).
### 4.2 Three Critical Flaws in the Chat Corpus's "Entropy Siphoning" Claims (and Their Formal Resolution)
Following our research-evidence triage rules ([AGENTS.md](https://github.com/westurner/sustainablefactory/blob/main/AGENTS.md); [`docs/chat-physics-catalog.md`, lines 16, 137](https://github.com/westurner/sustainablefactory/blob/main/docs/chat-physics-catalog.md#L16)), we audit the chat claims in [`_Mpemba Effect_ Landau vs. Quantum .md`, lines 298–495](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Mpemba%20Effect_%20Landau%20vs.%20Quantum%20.md#L298-L495) and [`_Quantum Processor and Soliton Discussions .md`, lines 4470–4700](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Quantum%20Processor%20and%20Soliton%20Discussions%20%20.md#L4470-L4700):
1. **Critique 1 — Closed-Loop Second-Law Accounting (`closedLoopErasureAndMemoryResetHeat`)**:
Negative conditional entropy $S(A \mid O) < 0$ is **not** a free perpetual-motion refrigerant: erasing $A$ consumes the entanglement between $A$ and $O$ and leaves the observer memory $O$ in a mixed state of entropy $\Delta S_O = +2F$ bits (for the linear model). To operate a continuous cyclic QEC loop, the memory $O$ must itself be reset (or fresh entanglement must be generated by external laser/microwave work $W_{\mathrm{pump}} \ge 2 F \, k_B T \ln 2 + Q_{\mathrm{overhead}}$). Summing the single-shot siphon heat and the memory reset cost gives:
$$
Q_{\mathrm{closed}}(F) = \underbrace{k_B T \ln 2 \cdot (1 - 2F)}_{Q_{\mathrm{erasure}}(A \mid O) \,<\, 0} + \underbrace{2F\,k_B T \ln 2 + Q_{\mathrm{overhead}}}_{Q_{\mathrm{reset}}(O) \,>\, 0} = k_B T \ln 2 + Q_{\mathrm{overhead}} \ge k_B T \ln 2 > 0.
$$
Thus, entangled erasure can **spatially relocate** heat away from the data qubits (if the ancilla/memory is physically transported or optically reset at a remote heat sink), but it **cannot** reduce the total closed-loop entropy production below Landauer's $+k_B T \ln 2$ per erased bit!
2. **Critique 2 — Isotropic 2-Qubit Werner State Visibility ($p_* \approx 0.747614$) vs. Bell Singlet Fidelity ($F_* \approx 0.810710$) and Linearized Conditional Entropy (`wernerConditionalEntropyBitsAtHalf`, `werner_cooling_threshold_fidelity`)**:
In `lignolux_entropy_siphon_sim.py` (line 4504), the chat writes `heat_per_reset = k_B * T_ambient * np.log(2) * (1 - 2 * fidelity)` and claims that cooling begins whenever `fidelity > 0.5`. However, if hardware depolarizing noise produces an isotropic 2-qubit Werner state
$$
\rho_W(p) = p\,|\Phi^+\rangle\langle\Phi^+| + (1 - p)\frac{I_4}{4}
= F\,|\Phi^+\rangle\langle\Phi^+| + \frac{1 - F}{3}\bigl(I_4 - |\Phi^+\rangle\langle\Phi^+|\bigr),
\qquad F(p) = \frac{3p + 1}{4} \iff p(F) = \frac{4F - 1}{3},
$$
its four eigenvalues are $(F, \frac{1-F}{3}, \frac{1-F}{3}, \frac{1-F}{3})$ (`wernerTripletEigenvalue`) and the exact conditional von Neumann entropy (in bits) is:
$$
S(A \mid O)_{\mathrm{Werner}}(F) = -F\log_2 F - (1 - F)\log_2\!\left(\frac{1 - F}{3}\right) - 1.
$$
As machine-checked in Lean 4 (`wernerConditionalEntropyBitsAtHalf` in [`QuantumCausalInference.lean`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/QuantumCausalInference.lean#L555-L665)) and Rust (`exact_werner_conditional_entropy_bits`, `werner_cooling_threshold_fidelity` in [`quantum_causal_inference.rs`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/numerical_adapter/src/quantum_causal_inference.rs#L400-L460)):
- At the Peres–Horodecki entanglement threshold $p = 1/3 \iff F = 1/2$, the chat's linear proxy vanishes ($1 - 2F = 0$), whereas the isotropic Werner state has strictly positive conditional entropy:
$$
S(A \mid O)_{\mathrm{Werner}}(1/2) = \frac{\ln 3}{2 \ln 2} = \frac{1}{2}\log_2 3 \approx +0.79248125\text{ bits} > \frac{1}{2} > 0
\quad (\text{net Landauer heating!}).
$$
- Furthermore, the true zero-cooling threshold $S(A \mid O)_{\mathrm{Werner}}(F_*) = 0$ occurs at Werner visibility $p_* \approx 0.74761383$, which corresponds via $F(p) = \frac{3p+1}{4}$ to a **Bell singlet fidelity threshold of $F_* \approx 0.81071038$ ($81.07\%$ Bell fidelity)**—strictly higher than both $0.5$ and $p_*$!
3. **Critique 3 — $6.67 \times 10^7\times$ Numerical Error in `lignolux_entropy_siphon_sim.py` (`-0.750 μW` vs. `-50 to -100 Watts`)**:
In [`data/chats/_Quantum Processor and Soliton Discussions .md`, lines 4487–4525](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Quantum%20Processor%20and%20Soliton%20Discussions%20%20.md#L4487-L4525), the chat specifies:
- $k_B = 1.38 \times 10^{-23}\text{ J/K}$, $T_{\mathrm{ambient}} = 4.0\text{ K} \implies k_B T \ln 2 \approx 3.82617 \times 10^{-23}\text{ J}$,
- $N_{\mathrm{ancillas}} = 2 \times 10^8$, $f_{\mathrm{QEC}} = 10^8\text{ Hz}$, and $F = 0.99 \implies 1 - 2F = -0.98$.
Evaluating the chat's own line 4507 (`total_power_watts = heat_per_reset * n_ancillas * qec_frequency`) in Lean 4 (`lignoluxTapeSiphonPowerWatts`) and Rust (`lignolux_tape_siphon_power_watts`) yields:
$$
P_{\mathrm{siphon}}(4\text{ K}, 100\text{ MHz}) = (3.82617 \times 10^{-23}\text{ J}) \times (-0.98) \times (2 \times 10^{16}\text{ s}^{-1}) = -7.4993 \times 10^{-7}\text{ W} = -0.750\text{ }\mu\text{W},
$$
which is **$6.67 \times 10^7\times$ smaller** than the `-50 to -100 Watts` claimed in the chat prose at line 4520! Even if one replaced `qec_frequency = 1e8` by `clock_rate = 1e12` ($1\text{ THz}$), the ideal single-shot cooling power at $4\text{ K}$ would be $-7.50\text{ mW}$, still four orders of magnitude below $-50\text{ W}$. Accordingly, hardware claims of net QPU refrigeration are isolated in `QuantumCausalMpembaSynthesisContract` ([`Signals/Pending.lean`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/Pending.lean#L5395-L5510)), which enforces closed-loop Second-Law accounting, unentangled $\operatorname{do}(F=0)$ control subtraction, and N-LIG TES bolometer calibration.
---
## 5. Beyond $6{,}400\times$ Volumetric Efficiency, Jeffrey's Kinematic Engine, 3D Splat Nanoimaging/Surgical Navigation, and Quantum Causal Semantic GraphRAG
In Section 11 of [`src/signals/Signals/QuantumCausalInference.lean`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/QuantumCausalInference.lean#L1680-L2164) and our companion paper ([`grassmannian-quantum-causal-splat-nanoimaging-surgical-graphrag-paper.md`](grassmannian-quantum-causal-splat-nanoimaging-surgical-graphrag-paper.md)), we extend the Grassmannian quantum causal framework developed in Sections 1–4 to three real-time volumetric and semantic inference domains:
1. **Beyond $6{,}400\times$ Volumetric Attention Efficiency via Splat-Sparse Grassmannian Mixing (`splatSparseGrassmannVideoComplexity`, `volumetricAttentionEfficiencyRatio`, `effectiveVolumetricSpeedupWithOverhead`)**:
Replacing dense 3D spatiotemporal self-attention $\mathcal{C}_{\mathrm{attn}} = (T H W)^2 d_{\mathrm{head}}$ by multi-lag Plücker subspace mixing over $N_{\mathrm{splats}}$ active 3D Normed Tensor Gaussian Splats ($\mathcal{C}_{\mathrm{splat\text{-}grass}} = N_{\mathrm{splats}} \cdot m_{\mathrm{lags}} \cdot d_{\mathrm{pl\ddot{u}cker}}$) yields exact machine-checked speedups of:
- **$13{,}107.2\times > 6{,}400\times$** on a dense $16 \times 16 \times 16$ patch volume ($N_{\mathrm{splats}} = 4{,}096, d_{\mathrm{head}} = 64, m_{\mathrm{lags}} = 2, d_{\mathrm{pl\ddot{u}cker}} = 10$),
- **$104{,}857.6\times > 6{,}400\times$** on a dense $32 \times 32 \times 32$ surgical volume ($N_{\mathrm{splats}} = 32{,}768$) or a $16^3$ volume with $8\times$ splat sparsity ($N_{\mathrm{splats}} = 512$),
- **$1{,}310{,}720\times > 6{,}400\times$** on a $32 \times 32 \times 40$ endoscopic/OCT surgical slab ($N_{\mathrm{vox}} = 40{,}960$) with $10\times$ adaptive splat sparsity ($N_{\mathrm{splats}} = 4{,}096$), and
- **$6{,}553.6\times > 6{,}400\times$** (`effectiveVolumetricSpeedupWithOverhead`) even after including $C_{\mathrm{overhead}} = 16{,}302{,}080$ FLOPs ($199\times$ the bare Plücker wedge FLOPs) of per-splat linear projection and 3D spatial radix-sorting overhead.
2. **Jeffrey's Kinematic Engine (`jeffreyKinematicUpdate`, `jeffreyKinematicShift`, `quantumJeffreyPluckerKinematicUpdate`, `partitionWeightedQuantumJeffreyUpdate`, `sequentialJeffreyPool`, `diaconisZabellLogOddsUpdate`)**:
When noisy sensor telemetry or soft semantic queries update partition probabilities $P(E_k) = p_k \mapsto Q(E_k) = q_k \in (0, 1)$ without making any cell strictly certain ($Q(E_k) < 1$), **Richard Jeffrey's Probability Kinematics** {cite:p}`jeffrey1965logicOfDecision` preserves the conditional fiber mechanism $c_k^{\operatorname{do}} = P(S \mid \operatorname{do}(E_k))$ (`Jeffrey Rigidity`), removes backdoor sensor confounding $q_1 b_1 + q_2 b_2$ via Pearl's $\operatorname{do}(E_k)$ surgery, incorporates coherent $\mathrm{Gr}_{\mathbb{C}}(2,4)$ Plücker cross-channel interference (`quantumJeffreyPluckerKinematicUpdate`, `partitionWeightedQuantumJeffreyUpdate` satisfying crisp boundary rigidity at $(1,0)$ and $(0,1)$, {cite:p}`qci-luders1950`), and resolves the Diaconis–Zabell sequential order-asymmetry $\lambda^2(q_2 - q_1)$ via additive log-odds commutativity (`diaconisZabellLogOddsUpdate`).
3. **3D Volumetric Nanoimaging & Surgical Navigation (`NormedTensorGaussianSplat3D.jeffreyKinematicWeightUpdate`)**:
Updating a 3D Normed Tensor Gaussian Splat (`NormedTensorGaussianSplat3D`) via Jeffrey's Kinematic Engine preserves $\det\Sigma_{\mathrm{3D}} = \sigma_1^2 \sigma_2^2 \sigma_3^2$ and $\operatorname{Tr}\Sigma_{\mathrm{2D}} = \sigma_1^2 + \sigma_2^2$ (**Jeffrey Geometric Rigidity**), simultaneously conserves 0th moment ($w_+ + w_- = q_1 c_1^{\operatorname{do}} + q_2 c_2^{\operatorname{do}}$), 1st moment ($\frac{1}{2}(c_+.\mathrm{centerX} + c_-.\mathrm{centerX}) = \mathrm{splat}_j.\mathrm{centerX}$), and 2nd central moment ($\frac{1}{2}[(\sigma_{1,+}^2 + \delta_1^2) + (\sigma_{1,-}^2 + \delta_1^2)] = \sigma_1^2$ at $\delta_1 = \frac{\sqrt{3}}{2}\sigma_1$) while sharpening $\det\Sigma_{\mathrm{3D},\pm} = \frac{1}{4}\det\Sigma_{\mathrm{3D}}$ under adaptive major-axis splitting at surgical margins or nanoscale defects, and recovers the hidden axial depth variance $\sigma_3^2 = \operatorname{Tr}\Sigma_{\mathrm{2D}} / R_{\mathrm{depth}}$ from the holographic $\mathrm{Gr}_{\mathbb{C}}(2,4)$ Plücker channel ratio.
4. **Quantum Causal Semantic GraphRAG (`quantumCausalGraphRAGScore`)**:
Combining De Clercq tautological subspace filtering ($\Pi_{Z_q} v^\perp = 0$), LGV disjoint-trek rank provenance (`singleHubRepresentative_det_zero` flagging rank-1 hub confounders), Pearl $\operatorname{do}$-deconfounded Jeffrey belief propagation, and Quantum Switch multi-hop ordering (`quantumSwitchSymBranchWeight`, `quantumSwitchAntiBranchWeight`) eliminates hub bias and contradictory multi-hop hallucinations in GraphRAG {cite:p}`edge2024graphrag`.
5. **Optional Complex Vector-Relation Configurations (`VRC`) vs. Plabic Rational Charts in Grassmannian Quantum Causal Inference (`GrassmannianCausalBackend`, `qciRecommendBackend`, `qciGr24PluckerMinor`, `qciOptionalVrcJeffreyKinematicUpdate`, `qciOptionalVrcGraphRAGScore`)**:
In Sections 13 and 14 of [`src/signals/Signals/QuantumCausalInference.lean`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/Signals/QuantumCausalInference.lean#L2110-L2305) (Master Theorem 9 `qci_optional_vrc_vs_plabic_complex_grassmannian_identities`, [`quantum_causal_inference.rs`](https://github.com/westurner/sustainablefactory/blob/main/src/signals/numerical_adapter/src/quantum_causal_inference.rs#L1180-L1330), and [`sustainablefactory/qci_grassmannian_dft.py`](../sustainablefactory/qci_grassmannian_dft.py#L1680-L1800); see also [Paper 22 (`grassmannian-kohn-sham-dft-complex-plabic-vrc-paper.md`)](grassmannian-kohn-sham-dft-complex-plabic-vrc-paper.md) and [Paper 23 (`grassmannian-speed-accuracy-barrier-quantum-chemistry-paper.md`)](grassmannian-speed-accuracy-barrier-quantum-chemistry-paper.md)), we extend Grassmannian Quantum Causal Inference (`QCI`) with:
- **QCI Directed Conditional Mutual Information, Phase Velocity, & Spin-Channel Mutual Information (`qciDirectedConditionalMutualInfo`, `qciCausalPhaseVelocity`, `qciConfinedPlasmonProcessWeight`, `qciSpinChannelMutualInformation`)**: Formalizes directed conditional mutual information $I(X \to Y \mid Z_{\mathrm{past}}) = H(Y_t \mid Y_{\mathrm{past}}, Z_{\mathrm{past}}) - H(Y_t \mid Y_{\mathrm{past}}, X_{t-\Delta t}, Z_{\mathrm{past}})$, plasmonic phase velocity $v_p = \Delta x / \Delta t_{\mathrm{causal}}$, confined process-matrix causal superposition $W_{\mathrm{conf}} = \alpha W_{A \to B \to C} + \beta W_{C \to B \to A}$, and unrestricted spin-channel mutual information $I(\alpha;\beta) = S(P_\alpha) + S(P_\beta) - S(P_{\alpha\beta})$.
- **Optional Complex Homogeneous VRC Parameterization (`GrassmannianCausalBackend.complexHomogeneousVRC` vs. `.plabicRationalChart`)**: Standard Postnikov Plabic boundary measurement maps (`gr24RationalChartMatrix`) use rational face weights $1/x_X$ and $x_1' = (x_1 x_X + x_2)/(x_X + 1)$, which are well-defined on the totally positive real Grassmannian $\mathrm{Gr}_{>0}(k,n)$ ($x_f > 0$) but diverge on $\mathrm{Gr}_{\mathbb{C}}(k,n)$ at the destructive-interference / on-shell polariton resonance pole $x_X = -1$ and chart boundary $x_X = 0$. The selector `qciRecommendBackend isComplex hasResonance` automatically switches to **Complex Homogeneous Vector-Relation Configurations (`VRC`, `(V, K)` with $V K^\top = 0$)** (`gr24HomogeneousVRC`, `gr24MutatedHomogeneousVRC`) whenever complex phases or resonances are present, proving:
1. *Away from chart poles ($x_X \ne 0$)*: Clearing the $1/x_X$ denominator at white vertex $w_2$ scales Plücker minors involving column $2$ ($\Delta_{02}, \Delta_{12}, \Delta_{23}$) by $x_X$ while leaving $\Delta_{01}, \Delta_{03}, \Delta_{13}$ unchanged, so both uncrossed causal channels scale by the exact same projective factor $x_X$ (`qciGr24UncrossedChannelA .complexHomogeneousVRC = x_X * qciGr24UncrossedChannelA .plabicRationalChart`, `qciGr24UncrossedChannelB .complexHomogeneousVRC = x_X * qciGr24UncrossedChannelB .plabicRationalChart`) and represent the identical projective subspace in $\mathrm{Gr}_{\mathbb{C}}(2,4)$.
2. *At the on-shell destructive-interference / polariton resonance ($x_X = -1$)*: The mutated Complex VRC (`mutated = true`) remains completely pole-free with finite Plücker minors ($\Delta_{01} = 1, \Delta_{02} = 0, \Delta_{03} = x_2, \Delta_{12} = x_2 - x_1, \Delta_{13} = 1, \Delta_{23} = x_1 x_2 - x_2^2$) satisfying the exact 3-term Plücker relation.
3. *Under Pearl $\operatorname{do}$-surgical backdoor severing (`doSeverBackdoor = true`, setting $x_2 \mapsto 0$)*: The backdoor Plücker channel vanishes identically ($w_B^{\mathbb{C}} = 0$) in both `plabicRationalChart` and `complexHomogeneousVRC`, collapsing `qciOptionalVrcJeffreyKinematicUpdate` and `qciOptionalVrcGraphRAGScore` to the exact deconfounded classical Jeffrey update $q_1 c_1^{\operatorname{do}} + q_2 c_2^{\operatorname{do}}$.
---
## 6. Evidence Triage Table (`research-evidence-triage`)
| Claim / Model Component | Source Citation | Evidence Status | Confidence | Owning Lean 4 / Rust Symbol | Missing Control / Next Check |
|---|---|---|---|---|---|
| **Plabic / LGV Causal Transfer & d-Separation Bottlenecks ($\det T_{A,B} = 0$ vs. $> 0$)** | {cite:p}`sullivant2010trek`; {cite:p}`postnikov2006` | `scholarly theory` (verified math) | High | `grassmannian_causal_relations_do_calculus_and_quantum_switch_identities`, `causal_dag_do_surgery_3x3` | Apply to empirical multi-node sensor DAG covariance matrices |
| **Quantum Causal Counterfactuals (`Abduction–Action–Prediction`) & Quantum Switch on $\mathrm{Gr}_{\mathbb{C}}(2,4)$** | [`_Mpemba Effect_ Landau vs. Quantum .md`, lines 390–455](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Mpemba%20Effect_%20Landau%20vs.%20Quantum%20.md#L390-L455); {cite:p}`giarmatzi2023semantics`; {cite:p}`chiribella2013switch` | `scholarly theory` (verified math) | High | `counterfactualStructuralOutcome`, `quantumSwitchCausalAsymmetry`, `evaluate_quantum_switch_2x2` | Interferometric visibility calibration on photonic Quantum Switch bench |
| **Landau Quartic Potential & Strong Mpemba Slow-Mode Orthogonal Projection ($a_{2,\mathrm{hot}} = 0$)** | [`_Mpemba Effect_ Landau vs. Quantum .md`, lines 45–75, 240–272](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Mpemba%20Effect_%20Landau%20vs.%20Quantum%20.md#L45-L75); {cite:p}`holtzman2022landau` | `scholarly theory` (verified math) | High | `landauFreeEnergy`, `mpembaTwoModeDistance`, `strongMpembaCrossingTime` | Measure Liouvillian decay rates $\lambda_2, \lambda_3$ in target spin/phonon bath |
| **Quantum Resource Theory (QRT) Thermo-Majorization & Constructor Theory Task Algebra** | [`_Mpemba Effect_ Landau vs. Quantum .md`, lines 78–110, 355–435](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Mpemba%20Effect_%20Landau%20vs.%20Quantum%20.md#L78-L110); {cite:p}`summer2026resource`; {cite:p}`marletto2015info` | `scholarly theory` (verified math) | High | `qrtNonEquilibriumFreeEnergy`, `ConstructorTask`, `thermo_majorization_dominates` | Tomographic state reconstruction of $\rho(t)$ relative to $\gamma_\beta$ |
| **Single-Shot Conditional Landauer Cooling ($S(A\|O) < 0 \implies Q < 0$) vs. Closed-Loop Reset** | [`_Mpemba Effect_ Landau vs. Quantum .md`, lines 298–325, 440–480](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Mpemba%20Effect_%20Landau%20vs.%20Quantum%20.md#L298-L325); {cite:p}`delRio2011thermodynamicMeaning` | `scholarly theory` (single-shot) / `calibration-required` (cyclic) | High | `delRioConditionalLandauerHeat`, `closedLoopErasureAndMemoryResetHeat`, `wernerConditionalEntropyBitsAtHalf`, `werner_cooling_threshold_fidelity` | Include ancilla memory reset heat $Q_{\mathrm{reset}}$ and Werner threshold $F_* \approx 0.8107$ ($p_* \approx 0.7476$) |
| **Lignolux Tape `-50 to -100 W` Entropy-Siphon Refrigeration Claim** | [`_Quantum Processor and Soliton Discussions .md`, lines 4480–4525](https://github.com/westurner/sustainablefactory/blob/main/data/chats/_Quantum%20Processor%20and%20Soliton%20Discussions%20%20.md#L4480-L4525) | `unsupported` ($-0.750\text{ }\mu\text{W}$ formula mismatch; `Pending` contract only) | High (audit) | `lignoluxTapeSiphonPowerWatts`, `QuantumCausalMpembaSynthesisContract` (`Signals.Pending`) | Keep in `Signals.Pending`; require $\operatorname{do}(F=0)$ control subtraction & TES calibration |
| **$>6{,}400\times$ Splat-Grassmann Volumetric Efficiency, Jeffrey's Kinematic Engine, & Quantum Causal Semantic GraphRAG** | {cite:p}`jeffrey1965logicOfDecision`; {cite:p}`qci-luders1950`; {cite:p}`declercq2026bundleTransformer`; {cite:p}`edge2024graphrag` | `scholarly theory` (verified math) / `calibration-required` (clinical margin & GraphRAG recall) | High | `volumetricAttentionEfficiencyRatio`, `jeffreyKinematicUpdate`, `quantumCausalGraphRAGScore`, `QuantumCausalSurgicalSplatGraphRAGHypothesis` | Phantom registration, motion-gating calibration, and multi-hop GraphRAG benchmark evaluation |
| **Optional Complex Vector-Relation Configurations (`VRC`) vs. Plabic Rational Charts & QCI Spin-Channel Mutual Information** | [`Google-Search_Density-Functional-Theory_20261105-1512.md`, lines 669–765, 1453–1479](https://github.com/westurner/sustainablefactory/blob/main/data/chats/Google-Search_Density-Functional-Theory_20261105-1512.md#L669-L765); [Paper 22](grassmannian-kohn-sham-dft-complex-plabic-vrc-paper.md) | `scholarly theory` (verified math) | High | `GrassmannianCausalBackend`, `qci_optional_vrc_vs_plabic_complex_grassmannian_identities`, `qciSpinChannelMutualInformation` | Experimental polariton / spin-wave interferometry at destructive resonance $x_X = -1$ |
---
## 7. Self-Contained Machine-Checked Lean 4 Verification Block
The following Lean 4 verification block compiles directly against `Signals` and `SignalsPending` (`cd src/signals && lake env lean`):
```lean
import Signals
import SignalsPending
namespace QuantumCausalInferenceMpembaPaperVerification
open Signals
open Signals.DDF
open Signals.Geometry
open Signals.Plabic
open Signals.QuantumCausalInference
open Signals.Pending
-- 1. Verification of Grassmannian causal transfer (W^3 = 0, T = 1 + W * T),
-- guarded mediator subdivision, shared-hub confounder bottleneck (det = 0),
-- 3-through-2 mediator bottleneck (det(A * B) = 0) vs. 3-channel disjoint positivity (det > 0),
-- Pearl's do(X) incoming-edge surgery, Gr_C(2,4) lightcone causality (1 - β² > 0),
-- Giarmatzi-Costa counterfactual abduction-action-prediction, and the Quantum Switch on Gr_C(2,4)
example (theta : ℝ) (A : Matrix (Fin 3) (Fin 2) ℝ) (B : Matrix (Fin 2) (Fin 3) ℝ) :
let net := WeightedAcyclicNetwork.diamond (2 : NNReal) (3 : NNReal) (4 : NNReal) (5 : NNReal)
let M_cone := ddfThermalGr24Matrix (3 / 5) theta
let delta_sw := quantumSwitchCausalAsymmetry 3 1
let M_sw := ddfThermalGr24Matrix delta_sw theta
net.weight ^ 3 = 0 ∧
net.transfer = 1 + net.weight * net.transfer ∧
(WeightedAcyclicNetwork.twoEdgePath (2 : NNReal) ((6 : NNReal) / 2)).transfer 0 2 =
(WeightedAcyclicNetwork.oneEdgePath (6 : NNReal)).transfer 0 1 ∧
(singleHubRepresentative (2 : NNReal) (3 : NNReal) (4 : NNReal) (5 : NNReal)).mat.det = 0 ∧
(A * B).det = 0 ∧
0 < (planarThreeChannelLowerStep 1 1 1 5 7 *
planarThreeChannelUpperStep 2 3 4 6 8).det ∧
causalObservationalTransfer 5 2 = 5 + 2 ∧
causalInterventionalDoTransfer 5 2 = 5 ∧
causalConfoundingBias 5 2 = 2 ∧
0 < Complex.normSq (complexPluckerSinkAmplitudeA M_cone) *
Complex.normSq (complexPluckerSinkAmplitudeB M_cone) ∧
causalLightconeInterval 1 = 0 ∧
counterfactualStructuralOutcome 2 3 10 3 = 10 ∧
counterfactualInterventionalDifference 2 3 10 5 = 2 * (5 - 3) ∧
Complex.normSq (complexPluckerSinkAmplitudeB M_sw) /
Complex.normSq (complexPluckerSinkAmplitudeA M_sw) = (1 : ℝ) / 3 ∧
quantumChoiGrassmannianRealDim 4 4 4 = 96 := by
intro net M_cone delta_sw M_sw
have hc := grassmannian_causal_relations_do_calculus_and_quantum_switch_identities
net 6 2 2 3 4 5 (by norm_num) 2 3 4 6 8 5 7 (by norm_num) (by norm_num) (by norm_num)
A B 5 2 2 3 10 5 (3 / 5) 3 1 theta
(by norm_num [abs_of_pos]) (by norm_num) (by norm_num)
exact ⟨hc.1, hc.2.1, hc.2.2.1, hc.2.2.2.1, hc.2.2.2.2.1, hc.2.2.2.2.2.1,
hc.2.2.2.2.2.2.1, hc.2.2.2.2.2.2.2.1, hc.2.2.2.2.2.2.2.2.1,
hc.2.2.2.2.2.2.2.2.2.2.1, hc.2.2.2.2.2.2.2.2.2.2.2.1,
hc.2.2.2.2.2.2.2.2.2.2.2.2.1, hc.2.2.2.2.2.2.2.2.2.2.2.2.2.1,
hc.2.2.2.2.2.2.2.2.2.2.2.2.2.2.1, hc.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2⟩
-- 2. Verification of Landau double-well barrier, Strong Mpemba trajectory crossing,
-- QRT thermo-majorization shortcut, Constructor Theory task composition & side-information unlocking,
-- del Rio conditional Landauer cooling, Second-Law closed-loop reset, and Lignolux -0.750 μW audit
example (contract : QuantumCausalMpembaSynthesisContract) :
let task_no_info : ConstructorTask := ⟨-1, 0, 0⟩
let task_classical_info : ConstructorTask := ⟨-1, 0, 1⟩
let task_quantum_entangled : ConstructorTask := ⟨-1, 0, 2⟩
landauFreeEnergy 1 (-2) 2 0 < landauFreeEnergy 0 (-2) 2 0 ∧
mpembaTwoModeDistance 1 1 1 1 < mpembaTwoModeDistance 0 4 1 1 ∧
mpembaTwoModeDistance 0 4 (1 / 2) (1 / 16) <
mpembaTwoModeDistance 1 1 (1 / 2) (1 / 16) ∧
¬task_no_info.isPossible 1 ∧
task_classical_info.isPossible 1 ∧
task_quantum_entangled.isPossible 1 ∧
delRioConditionalLandauerHeat 1 0 = 1 ∧
delRioConditionalLandauerHeat 1 (1 / 2) = 0 ∧
delRioConditionalLandauerHeat 1 1 = -1 ∧
entropySiphonCounterfactualGain 1 1 0 = -2 ∧
(1 : ℝ) ≤ closedLoopErasureAndMemoryResetHeat 1 (99 / 100) (1 / 10) ∧
|lignoluxTapeSiphonPowerWatts (3826e-26) (99 / 100) (2e8) (1e8)| < 1e-6 ∧
66000000 * |lignoluxTapeSiphonPowerWatts (3826e-26) (99 / 100) (2e8) (1e8)| < 50 ∧
contract.layerCode.topologicalDimension = 3 ∧
delRioConditionalLandauerHeat contract.thermalScaleJoules contract.bellFidelity < 0 ∧
contract.thermalScaleJoules ≤
closedLoopErasureAndMemoryResetHeat
contract.thermalScaleJoules contract.bellFidelity
contract.opticalAndMemoryOverheadJoules := by
intro task_no_info task_classical_info task_quantum_entangled
let t0 : ConstructorTask := ⟨0, 0, 0⟩
have ht0 : t0.isPossible 1 := by
unfold ConstructorTask.isPossible ConstructorTask.thermodynamicSurplus
norm_num [t0]
have hm := mpemba_constructor_theory_and_entropy_siphoning_identities
1 (-2) 2 1 4 1 (99 / 100) (1 / 10)
(by norm_num) (by norm_num) (by norm_num) (by norm_num) (by norm_num) (by norm_num)
t0 t0 ht0 ht0
have hp := SignalsPending.quantum_causal_mpemba_and_entropy_siphon_bounds contract
exact ⟨hm.2.2.1, hm.2.2.2.1, hm.2.2.2.2.1,
hm.2.2.2.2.2.2.2.2.2.1, hm.2.2.2.2.2.2.2.2.2.2.2.1,
hm.2.2.2.2.2.2.2.2.2.2.2.2.2.1,
by unfold delRioConditionalLandauerHeat conditionalEntropyLinearBits; norm_num,
hm.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.1,
by unfold delRioConditionalLandauerHeat conditionalEntropyLinearBits; norm_num,
by unfold entropySiphonCounterfactualGain delRioConditionalLandauerHeat conditionalEntropyLinearBits; norm_num,
hm.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.1,
hm.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.1,
hm.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2,
hp.1, hp.2.2.2.1, hp.2.2.2.2.2.1⟩
-- 3. Verification of 3-node confounded DAG do(X) surgery, Strong Mpemba Gr_R(1,2)
-- orthogonal complement projector (trace = 1, det = 0, P^2 = P, slow-mode annihilation = 0),
-- Quantum Switch complex parallelogram law & OCB causal inequality violation (3/4 < (2+√2)/4),
-- and isotropic 2-qubit Werner visibility-to-fidelity bijection & F = 1/2 positive heating
-- (0 = S_lin(1/2) < 1/2 < S_Werner(1/2) = ln 3 / (2 ln 2))
example (x0 x1 fidelity : ℝ) (zAB zBA z : ℂ) :
confoundedDagTransfer02 2 3 4 - confoundedDagTransfer02 0 3 4 = 8 ∧
confoundedDagTransfer12 0 3 4 = 4 ∧
mpembaProjector00 (3 / 5) (4 / 5) + mpembaProjector11 (3 / 5) (4 / 5) = 1 ∧
mpembaProjector00 (3 / 5) (4 / 5) * mpembaProjector11 (3 / 5) (4 / 5) -
(mpembaProjector01 (3 / 5) (4 / 5)) ^ 2 = 0 ∧
mpembaProjectedSlowOverlap (3 / 5) (4 / 5) x0 x1 = 0 ∧
quantumSwitchSymBranchWeight zAB zBA + quantumSwitchAntiBranchWeight zAB zBA =
(1 / 2 : ℝ) * (Complex.normSq zAB + Complex.normSq zBA) ∧
quantumSwitchAntiBranchWeight z z = 0 ∧
quantumSwitchSymBranchWeight z (-z) = 0 ∧
ocbCausalSeparableBound < ocbQuantumIndefiniteBound ∧
(4 * ocbQuantumIndefiniteBound - 2) ^ 2 = 2 ∧
fidelity + 3 * wernerTripletEigenvalue fidelity = 1 ∧
conditionalEntropyLinearBits (1 / 2) = 0 ∧
(1 / 2 : ℝ) < wernerConditionalEntropyBitsAtHalf ∧
0 < wernerConditionalEntropyBitsAtHalf ∧
wernerVisibilityFromFidelity (wernerFidelityFromVisibility fidelity) = fidelity ∧
wernerFidelityFromVisibility (1 / 3) = 1 / 2 := by
have h :=
grassmannian_mpemba_projector_werner_threshold_and_ocb_causal_witness_identities
2 3 4 (3 / 5) (4 / 5) x0 x1 fidelity (by norm_num) zAB zBA z
refine ⟨by unfold confoundedDagTransfer02; norm_num,
h.2.1, h.2.2.1, h.2.2.2.1, h.2.2.2.2.2.2.1, h.2.2.2.2.2.2.2.1,
h.2.2.2.2.2.2.2.2.1, h.2.2.2.2.2.2.2.2.2.1, h.2.2.2.2.2.2.2.2.2.2.1,
h.2.2.2.2.2.2.2.2.2.2.2.1, h.2.2.2.2.2.2.2.2.2.2.2.2.1,
h.2.2.2.2.2.2.2.2.2.2.2.2.2.1, h.2.2.2.2.2.2.2.2.2.2.2.2.2.2.1,
h.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.1,
h.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.1,
h.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2⟩
-- 4. Verification of 3D P-V-T van der Waals surface, water ice-hump volume expansion,
-- Landau T > T_c vs. T < T_c phase-transition barrier dichotomy, Strong Mpemba Gr_R(1,2)
-- slow-mode annihilation, do-surgery deconfounding, and conditional Landauer / Second-Law synthesis
example (x0 x1 fidelity : ℝ) :
0 < vanDerWaalsPressure (1 / 10) 80 2 (1 / 2) 1 -
vanDerWaalsPressure (1 / 10) 30 2 (1 / 2) 1 ∧
waterAnomalousFreezingVolume 18 (3 / 2) 280 273 = 18 ∧
waterAnomalousFreezingVolume 18 (3 / 2) 265 273 = 39 / 2 ∧
waterAnomalousFreezingVolume 18 (3 / 2) 280 273 <
waterAnomalousFreezingVolume 18 (3 / 2) 265 273 ∧
0 < landauOrderCoeff 1 300 273 ∧
0 < landauFreeEnergy 1 (landauOrderCoeff 1 300 273) 2 0 ∧
landauOrderCoeff 1 271 273 < 0 ∧
landauFreeEnergy 0 (landauOrderCoeff 1 271 273) 2 0 -
landauFreeEnergy 1 (landauOrderCoeff 1 271 273) 2 0 =
landauDoubleWellBarrierHeight 1 271 273 2 ∧
0 < landauDoubleWellBarrierHeight 1 271 273 2 ∧
mpembaProjectedSlowOverlap (3 / 5) (4 / 5) x0 x1 = 0 ∧
causalObservationalTransfer 5 2 - causalInterventionalDoTransfer 5 2 = 2 ∧
entropySiphonCounterfactualGain 1 1 0 = -2 ∧
(1 / 2 : ℝ) < wernerConditionalEntropyBitsAtHalf ∧
1 ≤ closedLoopErasureAndMemoryResetHeat 1 fidelity (1 / 10) := by
have h :=
phase_diagram_complexity_mpemba_and_quantum_causal_synthesis_identities
(1 / 10) 30 80 2 (1 / 2) 1 18 (3 / 2) 265 273 280
1 273 300 271 2 1 1 (3 / 5) (4 / 5) x0 x1 5 2 1 fidelity (1 / 10)
(by norm_num) (by norm_num) (by norm_num)
(by norm_num) (by norm_num) (by norm_num)
(by norm_num) (by norm_num) (by norm_num) (by norm_num)
(by norm_num) (by unfold landauOrderCoeff; norm_num)
(by norm_num) (by norm_num) (by norm_num)
refine ⟨h.2.1, h.2.2.1, by linarith [h.2.2.2.1], h.2.2.2.2.1,
h.2.2.2.2.2.1, h.2.2.2.2.2.2.1, h.2.2.2.2.2.2.2.1,
h.2.2.2.2.2.2.2.2.1, h.2.2.2.2.2.2.2.2.2.1, h.2.2.2.2.2.2.2.2.2.2.1,
h.2.2.2.2.2.2.2.2.2.2.2.1, by linarith [h.2.2.2.2.2.2.2.2.2.2.2.2.1],
h.2.2.2.2.2.2.2.2.2.2.2.2.2.1, h.2.2.2.2.2.2.2.2.2.2.2.2.2.2⟩
-- 5. Verification of >6,400x splat-sparse Grassmannian volumetric efficiency
-- (13,107.2x, 104,857.6x, and 1,310,720x), Jeffrey's Kinematic Engine with Pearl do-deconfounding
-- and Quantum Lüders-Plücker interference, 3D Normed Tensor Gaussian Splat volumetric recovery,
-- and Quantum Causal Semantic GraphRAG
example (q1 c1_do c2_do b1 b2 gamma_coh wA_re wA_im wB_re wB_im s_taut : ℝ) :
volumetricAttentionEfficiencyRatio 16 16 16 64 4096 2 10 = 13107.2 ∧
volumetricAttentionEfficiencyRatio 16 16 16 64 4096 2 10 > 6400 ∧
volumetricAttentionEfficiencyRatio 32 32 32 64 32768 2 10 = 104857.6 ∧
volumetricAttentionEfficiencyRatio 32 32 40 64 4096 2 10 = 1310720 ∧
volumetricAttentionEfficiencyRatio 32 32 40 64 4096 2 10 > 6400 ∧
jeffreyKinematicShift (3 / 10) (7 / 10) q1 (1 - q1) c1_do c2_do =
(q1 - 3 / 10) * (c1_do - c2_do) ∧
quantumCausalGraphRAGScore s_taut q1 (1 - q1) (c1_do + b1) (c2_do + b2)
gamma_coh wA_re wA_im wB_re wB_im -
quantumCausalGraphRAGScore s_taut q1 (1 - q1) c1_do c2_do
gamma_coh wA_re wA_im wB_re wB_im =
jeffreyKinematicUpdate q1 (1 - q1) b1 b2 ∧
(quantumCausalGraphRAGScore s_taut q1 (1 - q1) c1_do c2_do
gamma_coh wA_re wA_im wB_re wB_im +
quantumCausalGraphRAGScore s_taut q1 (1 - q1) c1_do c2_do
gamma_coh wA_re wA_im (-wB_re) (-wB_im)) / 2 =
s_taut + jeffreyKinematicUpdate q1 (1 - q1) c1_do c2_do := by
refine ⟨?_, ?_, ?_, ?_, ?_, ?_, ?_, ?_⟩
· norm_num [volumetricAttentionEfficiencyRatio,
spatiotemporalFullAttentionComplexity, splatSparseGrassmannVideoComplexity]
· norm_num [volumetricAttentionEfficiencyRatio,
spatiotemporalFullAttentionComplexity, splatSparseGrassmannVideoComplexity]
· norm_num [volumetricAttentionEfficiencyRatio,
spatiotemporalFullAttentionComplexity, splatSparseGrassmannVideoComplexity]
· norm_num [volumetricAttentionEfficiencyRatio,
spatiotemporalFullAttentionComplexity, splatSparseGrassmannVideoComplexity]
· norm_num [volumetricAttentionEfficiencyRatio,
spatiotemporalFullAttentionComplexity, splatSparseGrassmannVideoComplexity]
· unfold jeffreyKinematicShift; ring
· unfold quantumCausalGraphRAGScore quantumJeffreyPluckerKinematicUpdate
jeffreyKinematicUpdate
ring
· unfold quantumCausalGraphRAGScore quantumJeffreyPluckerKinematicUpdate
jeffreyKinematicUpdate qcdtPluckerInterferenceTerm
ring
-- 6. Verification of Optional Complex Vector-Relation Configuration (VRC) vs. Plabic
-- Rational Chart parameterization on Gr_C(2,4) (Master Theorem 9):
-- automatic backend recommendation, Plücker quadratic relation across all backends,
-- projective x_X scaling away from poles, pole-free regularity at x_X = -1, and
-- Pearl do-surgical backdoor severing (doSeverBackdoor = true -> w_B = 0)
example (backend : GrassmannianCausalBackend) (mutated isRes doSever : Bool)
(x1 x2 xX : ℂ) (hxX : xX ≠ 0) (s_taut q1 q2 c1_do c2_do gamma_coh : ℝ) :
qciRecommendBackend true isRes = .complexHomogeneousVRC ∧
qciRecommendBackend false true = .complexHomogeneousVRC ∧
qciRecommendBackend false false = .plabicRationalChart ∧
pluckerQuadraticRelation2x4 (qciGr24BoundaryMatrix backend mutated x1 x2 xX doSever) = 0 ∧
qciGr24UncrossedChannelA .complexHomogeneousVRC false x1 x2 xX false =
xX * qciGr24UncrossedChannelA .plabicRationalChart false x1 x2 xX false ∧
qciGr24UncrossedChannelB .complexHomogeneousVRC false x1 x2 xX false =
xX * qciGr24UncrossedChannelB .plabicRationalChart false x1 x2 xX false ∧
qciGr24PluckerMinor .complexHomogeneousVRC true x1 x2 (-1) false 1 2 = x2 - x1 ∧
qciGr24UncrossedChannelB .complexHomogeneousVRC mutated x1 x2 xX true = 0 ∧
qciOptionalVrcJeffreyKinematicUpdate .complexHomogeneousVRC mutated
q1 q2 c1_do c2_do gamma_coh x1 x2 xX true =
jeffreyKinematicUpdate q1 q2 c1_do c2_do ∧
qciOptionalVrcGraphRAGScore .complexHomogeneousVRC mutated
s_taut q1 q2 c1_do c2_do gamma_coh x1 x2 xX true =
s_taut + jeffreyKinematicUpdate q1 q2 c1_do c2_do := by
have h9 := qci_optional_vrc_vs_plabic_complex_grassmannian_identities
backend mutated isRes doSever x1 x2 xX hxX s_taut q1 q2 c1_do c2_do gamma_coh
exact ⟨h9.1, h9.2.1, h9.2.2.1, h9.2.2.2.1,
h9.2.2.2.2.2.2.2.2.2.2.1, h9.2.2.2.2.2.2.2.2.2.2.2.1,
h9.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.1,
h9.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.1,
h9.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.1,
h9.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2.2⟩
end QuantumCausalInferenceMpembaPaperVerification
```
---
### Scholarly References
(qci-jeffrey1965)=
0a. Jeffrey, R. C. (1965). *The Logic of Decision*. McGraw-Hill, New York. (`jeffrey1965logicOfDecision`)
(qci-luders1950)=
0b. Lüders, G. (1950). Über die Zustandsänderung durch den Meßprozeß. *Annalen der Physik*, 443(5–8), 322–328. (`luders1950zustandsanderung`)
(qci-declercq2026)=
0c. De Clercq, C. (2026). *Bundle-Transformer: A Geometric Position Paper*. Université Sorbonne Paris Nord, LAGA. (`declercq2026bundleTransformer`)
(qci-edge2024)=
0d. Edge, D., et al. (2024). From Local to Global: A Graph RAG Approach to Query-Focused Summarization. `arXiv:2404.16130`. (`edge2024graphrag`)
(holtzman2022landau)=
1. Holtzman, R., & Raz, O. (2022). Landau theory for the Mpemba effect through phase transitions. *Communications Physics*, 5(1), 280. DOI: `10.1038/s42005-022-01063-2`. [https://doi.org/10.1038/s42005-022-01063-2](https://doi.org/10.1038/s42005-022-01063-2).
(summer2026resource)=
2. Summer, A., Moroder, M., Bettmann, L. P., Turkeshi, X., Marvian, I., & Goold, J. (2026). Resource-Theoretical Unification of Mpemba Effects: Classical and Quantum. *Physical Review X*, 16, 011050. DOI: `10.1103/rbt4-psfd`. [https://doi.org/10.1103/rbt4-psfd](https://doi.org/10.1103/rbt4-psfd).
(giarmatzi2023semantics)=
3. Giarmatzi, C., & Costa, F. (2023). A Semantics for Counterfactuals in Quantum Causal Models. `arXiv:2302.11783`. [https://arxiv.org/abs/2302.11783](https://arxiv.org/abs/2302.11783).
(delrio2011negative)=
4. del Rio, L., Åberg, J., Renner, R., Dahlsten, O., & Vedral, V. (2011). The thermodynamic meaning of negative entropy. *Nature*, 474(7349), 61–63. DOI: `10.1038/nature10123`. [https://doi.org/10.1038/nature10123](https://doi.org/10.1038/nature10123).
(deutsch2013constructor)=
5. Deutsch, D. (2013). Constructor theory. *Synthese*, 190(18), 4331–4359. DOI: `10.1007/s11229-013-0279-z`.
(marletto2015info)=
6. Deutsch, D., & Marletto, C. (2015). Constructor theory of information. *Proceedings of the Royal Society A*, 471(2174), 20140540. DOI: `10.1098/rspa.2014.0540`; and Marletto, C. (2016). Constructor Theory of Thermodynamics. `arXiv:1608.02625`.
(deutschmarletto2021constructor)=
7. Deutsch, D., & Marletto, C. (2021). *Constructor Theory Frequently Asked Questions (Q3: How does Constructor Theory differ from the framework of Resource Theories?)*. [https://www.constructortheory.org/faqs/](https://www.constructortheory.org/faqs/).
(oreshkov2012process)=
8. Oreshkov, O., Costa, F., & Brukner, Č. (2012). Quantum correlations with no causal order. *Nature Communications*, 3, 1092. DOI: `10.1038/ncomms2076`.
(chiribella2013switch)=
9. Chiribella, G., D'Ariano, G. M., Perinotti, P., & Valiron, B. (2013). Quantum computations without definite causal structure. *Physical Review A*, 88(2), 022318. DOI: `10.1103/PhysRevA.88.022318`.
(allen2017qcm)=
10. Allen, J.-M. A., Barrett, J., Horsman, D. C., Lee, C. M., & Spekkens, R. W. (2017). Quantum Common Causes and Quantum Causal Models. *Physical Review X*, 7(3), 031021. DOI: `10.1103/PhysRevX.7.031021`.
(sullivant2010trek)=
11. Sullivant, S., Talaska, K., & Draisma, J. (2010). Trek separation for Gaussian graphical models. *The Annals of Statistics*, 38(3), 1665–1685. DOI: `10.1214/09-AOS760`.
(pearl2009causality)=
12. Pearl, J. (2009). *Causality: Models, Reasoning, and Inference* (2nd ed.). Cambridge University Press. DOI: `10.1017/CBO9780511803161`.
(landauer1961irreversibility)=
13. Landauer, R. (1961). Irreversibility and Heat Generation in the Computing Process. *IBM Journal of Research and Development*, 5(3), 183–191. DOI: `10.1147/rd.53.0183`.
(penrose1986spinors)=
14. Penrose, R., & Rindler, W. (1986). *Spinors and Space-Time, Vol. 2: Spinor and Twistor Methods in Space-Time Geometry*. Cambridge University Press.
(postnikov2006plabic)=
15. Postnikov, A. (2006). Total positivity, Grassmannians, and networks. `arXiv:math/0609764`.
(lindstrom1973lgv)=
16. Lindström, B. (1973). On the vector representations of induced matroids. *Bulletin of the London Mathematical Society*, 5(1), 85–90; and Gessel, I., & Viennot, G. (1985). Binomial determinants, paths, and hook length formulae. *Advances in Mathematics*, 58(3), 300–321.
(mpemba-chat2026)=
17. Turner, Wesley and Google Geminis (2026). *Mpemba Effect: Landau vs. Quantum Resource Theory, Constructor Theory, Counterfactual Semantics in Quantum Causal Models, and Entropy Siphoning*. Local corpus: {download}`data/chats/_Mpemba Effect_ Landau vs. Quantum .md, lines 1–555 <../data/chats/_Mpemba Effect_ Landau vs. Quantum .md>` and {download}`data/chats/_Quantum Processor and Soliton Discussions .md, lines 4470–4700 <../data/chats/_Quantum Processor and Soliton Discussions .md>`.
```bibtex
@article{holtzman2022landau,
title = {Landau theory for the Mpemba effect through phase transitions},
author = {Holtzman, Roi and Raz, Oren},
journal = {Communications Physics},
volume = {5},
number = {1},
pages = {280},
year = {2022},
doi = {10.1038/s42005-022-01063-2}
}
@article{summer2026resource,
title = {Resource-Theoretical Unification of Mpemba Effects: Classical and Quantum},
author = {Summer, Alessandro and Moroder, Mattia and Bettmann, Laetitia P. and Turkeshi, Xhek and Marvian, Iman and Goold, John},
journal = {Physical Review X},
volume = {16},
year = {2026},
doi = {10.1103/rbt4-psfd}
}
@article{giarmatzi2023semantics,
title = {A Semantics for Counterfactuals in Quantum Causal Models},
author = {Giarmatzi, Christina and Costa, Fabio},
journal = {arXiv preprint arXiv:2302.11783},
year = {2023},
url = {https://arxiv.org/abs/2302.11783}
}
@article{delrio2011negative,
title = {The thermodynamic meaning of negative entropy},
author = {del Rio, L{\'\i}dia and {\AA}berg, Johan and Renner, Renato and Dahlsten, Oscar and Vedral, Vlatko},
journal = {Nature},
volume = {474},
number = {7349},
pages = {61--63},
year = {2011},
doi = {10.1038/nature10123}
}
@misc{chat_mpemba_landau_quantum_2026,
author = {{Google Gemini / User}},
title = {Mpemba Effect: Landau vs. Quantum, Constructor Theory, Counterfactuals in Quantum Causal Models, and Entropy Siphoning},
year = {2026},
howpublished = {AI chat log},
note = {Source: data/chats/_Mpemba Effect_ Landau vs. Quantum .md, lines 1--555; data/chats/_Quantum Processor and Soliton Discussions .md, lines 4470--4700}
}
```
```json
{
"@context": "https://schema.org",
"@type": "ScholarlyArticle",
"headline": "Grassmannian Geometry of Quantum Causal Inference, Counterfactual Semantics, Constructor Theory, and the Mpemba / Entropy-Siphoning Laws in Lean 4 and Rust/WASM",
"author": {
"@type": "Person",
"name": "Wes Turner"
},
"dateModified": "2026-11-03",
"isBasedOn": [
"data/chats/_Mpemba Effect_ Landau vs. Quantum .md",
"data/chats/_Quantum Processor and Soliton Discussions .md",
"src/signals/Signals/QuantumCausalInference.lean",
"src/signals/numerical_adapter/src/quantum_causal_inference.rs"
],
"about": [
"Grassmannian Geometry of Causal Networks and d-Separation",
"Quantum Causal Models and Interventional Counterfactual Semantics",
"Indefinite Causal Order and the Quantum Switch on Gr_C(2,4)",
"Landau Free-Energy and Quantum Resource Theory Mpemba Effects",
"Constructor Theory of Thermodynamics and Information",
"Conditional Landauer Erasure and Entropy Siphoning Audit"
]
}
```