Add Makefile targets for null-results and report; update README/docs for null-results, script paths, formatting, and status wording; make examples a package and fix imports in examples/refined_feasibility; ensure scripts add repo root to PYTHONPATH and add timing to benchmark output; add benchmark_results.json and optimized_vs_baseline figure

Author: arcticoder

Makefile

@@ -11,7 +11,7 @@ help:
 	@echo "  make domain-sweep  - Domain padding sensitivity study"
 	@echo ""
 	@echo "Analysis:"
-	@echo "  make analysis      - Run interactive analysis menu"
+	@echo "  make analysis      - Run analysis CLI (prints usage; see --help for subcommands)"
 	@echo "  make optimize      - Run geometry optimization"
 	@echo ""
 	@echo "Code quality:"
@@ -92,6 +92,15 @@ paper:
 	cd papers && pdflatex coherence_gravity_coupling.tex && bibtex coherence_gravity_coupling \
 	  && pdflatex coherence_gravity_coupling.tex && pdflatex coherence_gravity_coupling.tex
 
+# Build null_results PDF (curvature–EM coupling null-results paper)
+null-results:
+	cd papers && pdflatex null_results.tex && bibtex null_results \
+	  || true; pdflatex null_results.tex && pdflatex null_results.tex
+
+# Generate consolidated analysis tables (CSV/Markdown/LaTeX)
+report:
+	python scripts/generate_report.py --all
+
 manifest:
 	python scripts/generate_manifest.py --output data_manifest.csv --roots results papers/figures
 

README.md

@@ -29,13 +29,22 @@ pdflatex coherence_gravity_coupling.tex && bibtex coherence_gravity_coupling \
    && pdflatex coherence_gravity_coupling.tex && pdflatex coherence_gravity_coupling.tex
 ```
 
+```bash
+# Also build the null-results paper (curvature–EM coupling constraints)
+cd papers
+pdflatex null_results.tex && bibtex null_results \
+   && pdflatex null_results.tex && pdflatex null_results.tex
+```
+
+Why multiple pdflatex runs? LaTeX generally requires 2–3 passes to resolve cross-references, citations, and the bibliography. The sequence above (pdflatex → bibtex → pdflatex → pdflatex) is the standard pattern to ensure all references are correct for both papers.
+
 Expected outputs:
 - `papers/figures/convergence_analysis.pdf` (Fig 1)
 - `papers/figures/material_comparison.pdf` (Fig 2)
 - `papers/figures/landscape_YBCO_z_slice.pdf` (Fig 3)
 - `papers/coherence_gravity_coupling.pdf` (5 pages)
 
-Runtime guidance (Intel i7-10700K, 32GB RAM): 41³ ~ 3–5s/solve; 61³ ~ 5–8s/solve; 81³ ~ 20–30s; 101³ ~ 1–2min. Full convergence (61/81/101) ~ 1–8 hours.
+Runtime guidance (Intel i7-10700K, 32GB RAM): 41³ ~ 3–5s/solve; 61³ ~ 5–8s/solve; 81³ ~ 20–30s; 101³ ~ 1–2min. Full convergence (61/81/101) ~ 1–8 hours. Note: These timings apply to the 3D solver runs (examples/analysis). pdflatex and figure generation complete in seconds.
 
 ### Using Make (recommended)
 
@@ -47,7 +56,7 @@ make test          # Run full test suite (23 tests, ~90s)
 make quick-bench   # Quick benchmark at 41³ (~30s)
 make figures       # Generate manuscript figures
 make paper         # Build coherence_gravity_coupling.pdf
-make analysis      # Interactive analysis menu
+make analysis      # Analysis CLI (prints usage; see --help for subcommands)
 make optimize      # Run geometry optimization
 make cache-info    # Show cache statistics
 make cache-clean   # Clear all cached results
@@ -60,7 +69,7 @@ This repo now includes a module and CLI to derive exclusion limits on a curvatur
 - Implementation: `src/field_equations/curvature_coupling.py`
 - Plots: `src/visualization/plot_utils.py::plot_exclusion_limits`
 - Preprint: `papers/null_results.tex`
-- Reports: auto-generated CSV/Markdown/LaTeX in `results/reports/` via `python generate_report.py --all`
+- Reports: auto-generated CSV/Markdown/LaTeX in `results/reports/` via `python scripts/generate_report.py --all` or `make report`
 
 ### CLI examples
 
@@ -449,21 +458,22 @@ where $`\tilde{T}_{\mu\nu} = T_{\mu\nu}^{\text{matter}} + T_{\mu\nu}^{\Phi}`$ in
 
 4. **Run geometric Cavendish simulation**:
 ```bash
-   # Single configuration with volume-averaged force (recommended)
-   python -c "
-   import numpy as np
-   from examples.geometric_cavendish import sweep_coherent_position
-   
-   result = sweep_coherent_position(
-       y_range=np.linspace(0.0, 0.05, 3),
-       z_range=np.linspace(-0.12, -0.04, 3),
-       xi=100.0,
-       Phi0=6.67e8,  # YBCO
-       verbose=True
-   )
-   print(f'\\nOptimal position: {result[\"optimal\"][\"position\"]}')
-   print(f'Delta tau: {result[\"optimal\"][\"delta_tau\"]:.3e} N·m')
-   "
+# Single configuration with volume-averaged force (recommended)
+python -c "
+import numpy as np
+from examples.geometric_cavendish import sweep_coherent_position
+
+result = sweep_coherent_position(
+   y_range=np.linspace(0.0, 0.05, 3),
+   z_range=np.linspace(-0.12, -0.04, 3),
+   base_geom_params={},
+   xi=100.0,
+   Phi0=6.67e8,  # YBCO
+   verbose=True
+)
+print(f'\nOptimal position: {result[\"optimal\"][\"position\"]}')
+print(f'Delta tau: {result[\"optimal\"][\"delta_tau\"]:.3e} N·m')
+"
 ```
 
 4. **Run convergence test**:
@@ -472,10 +482,10 @@ python -c "
 from examples.geometric_cavendish import convergence_test
 
 convergence_test(
-      grid_resolutions=[41, 61],
-      xi=100.0,
-      Phi0=6.67e8,
-      verbose=True
+   grid_resolutions=[41, 61],
+   xi=100.0,
+   Phi0=6.67e8,
+   verbose=True
 )
 "
 ```
@@ -597,7 +607,7 @@ coherence-gravity-coupling/
 
 ### Solver Performance
 
-**Recent Improvements** (January 2025): Implemented performance optimizations for 3D Poisson solver:
+**Recent Improvements** (October 2025): Implemented performance optimizations for 3D Poisson solver:
 
 #### Key Features
 - **Diagonal (Jacobi) preconditioner**: Fast preconditioning with O(N) setup cost
@@ -641,7 +651,7 @@ For details, see [`SOLVER_PERFORMANCE_IMPROVEMENTS.md`](SOLVER_PERFORMANCE_IMPRO
 
 #### Result Caching
 
-**New Feature** (January 2025): Optional result caching for parameter sweeps.
+**New Feature** (October 2025): Optional result caching for parameter sweeps.
 
 ```python
 # Enable caching to skip repeated expensive calculations
@@ -669,7 +679,7 @@ make cache-clean  # Clear all cached results
 
 #### Domain Size and Boundary Conditions
 
-**Recommendation** (January 2025): Use `domain_size ≥ 2.5× minimum_enclosing_size`.
+**Recommendation** (October 2025): Use `domain_size ≥ 2.5× minimum_enclosing_size`.
 
 **Domain Study Results** (61³ resolution, xi=100, YBCO):
 - **Padding 2.0×**: 7.1% variation in Δτ
@@ -811,7 +821,7 @@ Improvement: 1.00× (already optimal)
 - ⏳ Energy cost of warp drops to laboratory scale (~MJ instead of Earth mass) (not yet validated for full warp metric)
 - ⏳ Path to engineering curvature becomes plausible (testable with proposed experiment)
 
-**Phase D Status:** 🎯 **BREAKTHROUGH ACHIEVED** - Experiment is **trivially feasible**
+**Phase D Status:** 🎯 **BREAKTHROUGH ACHIEVED** - Experiment is **feasible but challenging** (requires cryogenics and precision isolation)
 
 ### Latest Updates (Phase D+)
 
@@ -880,10 +890,12 @@ Improvement: 1.00× (already optimal)
 
 ### Grid Convergence
 - **41³ grid**: Fast (~3s/solve), sufficient for parameter scans
-- **61³ grid**: Moderate (~10s/solve), shows ~220% Δτ change from 41³ (not fully converged)
+- **61³ grid**: Moderate (~10s/solve), often shows O(10²%) Δτ change vs 41³ (not fully converged)
 - **81³+ grid**: Recommended for quantitative predictions; requires volume averaging for stability
 - **Volume averaging**: Reduces aliasing by integrating ∇φ over test mass volume; use for convergence studies
 
+Note: A message like "Not fully converged (>62.49% change)" on 41³→61³ is expected for point-sample torque. Enable volume averaging (use_volume_average=True) and/or include 81³ to see the relative change drop below ~1–5%.
+
 ### Experimental Challenges
 1. **Cryogenic operation required**: Room temperature noise floor too high (0/18 configs <24hr feasible)
 2. **Integration times**: 0.7-24 hours for SNR=5 (not milliseconds as initially estimated)

benchmark_results.json

@@ -0,0 +1,74 @@
+{
+  "resolution": 41,
+  "n_runs": 2,
+  "materials": [
+    "rb87_bec"
+  ],
+  "results": {
+    "rb87_bec": {
+      "cg+none": {
+        "mean_time": 3.213704396504909,
+        "std_time": 0.008873526501702145,
+        "min_time": 3.2048308700032067,
+        "max_time": 3.222577923006611,
+        "mean_residual": 0.0,
+        "mean_torque": -5.105970735621204e-13,
+        "solver_method": "cg",
+        "preconditioner": "none",
+        "speedup": 1.0
+      },
+      "cg+diagonal": {
+        "mean_time": 3.120515021495521,
+        "std_time": 0.06067115650512278,
+        "min_time": 3.0598438649903983,
+        "max_time": 3.181186178000644,
+        "mean_residual": 0.0,
+        "mean_torque": -4.99161046947598e-13,
+        "solver_method": "cg",
+        "preconditioner": "diagonal",
+        "speedup": 1.0298634598351417
+      },
+      "bicgstab+none": {
+        "mean_time": 3.839985500992043,
+        "std_time": 0.10384562800754793,
+        "min_time": 3.736139872984495,
+        "max_time": 3.943831128999591,
+        "mean_residual": 0.0,
+        "mean_torque": -4.961596603107724e-13,
+        "solver_method": "bicgstab",
+        "preconditioner": "none",
+        "speedup": 0.8369053465630698
+      },
+      "bicgstab+diagonal": {
+        "mean_time": 3.181468226001016,
+        "std_time": 0.0014283429918577895,
+        "min_time": 3.180039883009158,
+        "max_time": 3.1828965689928737,
+        "mean_residual": 0.0,
+        "mean_torque": -4.991611064723489e-13,
+        "solver_method": "bicgstab",
+        "preconditioner": "diagonal",
+        "speedup": 1.0101324823050055
+      },
+      "cg+amg": {
+        "mean_time": 3.3534587640024256,
+        "std_time": 0.009777534985914826,
+        "min_time": 3.3436812290165108,
+        "max_time": 3.3632362989883404,
+        "mean_residual": 0.0,
+        "mean_torque": -4.991610680311029e-13,
+        "solver_method": "cg",
+        "preconditioner": "amg",
+        "speedup": 0.9583253060995696
+      }
+    }
+  },
+  "best_configs": {
+    "rb87_bec": {
+      "config": "cg+diagonal",
+      "time": 3.120515021495521,
+      "speedup": 1.0298634598351417
+    }
+  },
+  "target_speedup": 2.0
+}

docs/SOLVER_PERFORMANCE_IMPROVEMENTS.md

@@ -94,11 +94,15 @@ def run_geometric_cavendish(
 Created `benchmark_solver.py` to systematically test performance:
 
 ```bash
-# Test at 61³ resolution
-python benchmark_solver.py --resolution 61 --runs 3 --materials rb87_bec --xi 100
+# Test at 61³ resolution (from repo root)
+python scripts/benchmark_solver.py --resolution 61 --runs 3 --materials rb87_bec --xi 100
 
 # Test multiple materials
-python benchmark_solver.py --resolution 81 --runs 2 --materials rb87_bec nb_cavity al_film
+python scripts/benchmark_solver.py --resolution 81 --runs 2 --materials rb87_bec nb_cavity al_film
+
+# Or use the Makefile presets
+make quick-bench   # 41³, fast sanity check
+make bench         # 61³, standard benchmark
 ```
 
 Output includes:
@@ -253,15 +257,19 @@ COO format advantages:
 Run benchmarks:
 ```bash
 # Quick test (41³)
-python benchmark_solver.py --resolution 41 --runs 2
+python scripts/benchmark_solver.py --resolution 41 --runs 2
 
 # Standard test (61³)
-python benchmark_solver.py --resolution 61 --runs 3
+python scripts/benchmark_solver.py --resolution 61 --runs 3
 
 # Comprehensive test (multiple resolutions)
 for res in 41 61 81; do
-    python benchmark_solver.py --resolution $res --runs 3 --output benchmark_${res}.json
+    python scripts/benchmark_solver.py --resolution $res --runs 3 --output benchmark_${res}.json
 done
+
+# Makefile alternatives
+make quick-bench
+make bench
 ```
 
 Verify correctness:

examples/__init__.py

@@ -0,0 +1 @@
+# Make `examples` a package so absolute imports like `from examples.geometric_cavendish import ...` work

examples/figures/optimized_vs_baseline.png

@@ -12,8 +12,13 @@
 import numpy as np
 import json
 from pathlib import Path
+import sys
 import matplotlib.pyplot as plt
 from scipy.constants import k as k_B, hbar, c, G as G_newton
+# Ensure repository root is on sys.path so `import examples...` works when running this file directly
+ROOT = Path(__file__).resolve().parents[1]
+if str(ROOT) not in sys.path:
+    sys.path.insert(0, str(ROOT))
 
 # Physical constants (baseline geometry)
 m_test_baseline = 10e-3  # Test mass 10 mg (typical for micro-torsion balance)

examples/refined_feasibility.py

@@ -10,7 +10,11 @@
 
 import sys
 from pathlib import Path
-sys.path.insert(0, str(Path(__file__).parent))
+
+# Ensure repository root is on sys.path so `examples` and `src` are importable
+ROOT = Path(__file__).resolve().parents[1]
+if str(ROOT) not in sys.path:
+    sys.path.insert(0, str(ROOT))
 
 from examples.geometric_cavendish import run_geometric_cavendish
 from src.analysis.phi_calibration import get_all_calibrations
@@ -41,6 +45,7 @@ def benchmark_configuration(
     print(f"\n   Testing: {solver_method} + {preconditioner} (n={n_runs})")
 
     for i in range(n_runs):
+        t0 = time.perf_counter()
         result = run_geometric_cavendish(
             xi=xi,
             Phi0=Phi0,
@@ -49,12 +54,11 @@ def benchmark_configuration(
             solver_method=solver_method,
             preconditioner=preconditioner
         )
-        
-        sys.path.insert(0, str(Path(__file__).parent.parent))
+        dt = time.perf_counter() - t0
+        times.append(dt)
         residuals.append(result.get('residual_coherent', 0.0))
         torques.append(result['tau_coherent'])
-        
-        print(f"      Run {i+1}: {times[-1]:.2f} s, residual: {residuals[-1]:.2e}")
+        print(f"      Run {i+1}: {dt:.2f} s, residual: {residuals[-1]:.2e}")
 
     return {
         'mean_time': float(np.mean(times)),

scripts/benchmark_solver.py

@@ -35,6 +35,11 @@
 import subprocess
 from importlib import metadata as _metadata
 
+# Ensure repository root on sys.path so `examples`/`src` are importable when run via Makefile
+ROOT = Path(__file__).resolve().parents[1]
+if str(ROOT) not in sys.path:
+    sys.path.insert(0, str(ROOT))
+
 from examples.geometric_cavendish import run_geometric_cavendish
 
 # Output directory