# ⚡  L4 Solution — baseline:reference

**ID:** `L3-507-001-001-T1-reference-topup` · **Status:** ⊙ Testnet (genesis catalog)
**Parents:** ⬆ Benchmark [`L3-507-001-001-T1`](/artifact/l3/L3-507-001-001-T1)

> **🛠 Solver:** `baseline:reference`
> **📊 Performance:** **Q = `0.8`** · difficulty `0.15`
> **🔖 Status:** `testnet`

---

## 1. Solver

**`baseline:reference`**

## 2. Performance

| Property | Value |
|---|---|
| Difficulty score | 0.15 |
| Quality ratio | 0.8 |
| Reward pwm | 0 |

## 3. Operating Point (Ω_s)

| Parameter | Unit | Value |
|---|---|---|
| Snr db | dB | 20 |
| N dipoles | — | 8000 |
| T samples | — | 1000 |
| N eeg channels | — | 64 |
| N meg channels | — | 306 |
| Head layer count | — | 3 |
| Sampling rate hz | Hz | 1000 |
| Sensor position error | — | 0 |
| Head segmentation error | — | 0 |
| Conductivity uncertainty | — | 0 |
| Reference electrode drift | — | 0 |
| Source orientation assumption | — | cortically_constrained |
| Magnetic artifact contamination | — | 0 |

## 4. Certificate

**Hash:** `d66e59550eed433ea88f814dc3e1b912b46b6ea44f00dfda6e6564144c26dcb8`

**Status:** `testnet`

## 5. Code

Reference implementation — see [`code/`](/artifact/l4/L3-507-001-001-T1-reference-topup/raw/code/) when present.

## 6. Reproduction Report

Reconstruction results on the parent benchmark — see [`reports/`](/artifact/l4/L3-507-001-001-T1-reference-topup/raw/reports/) when present.

---

## File Mapping

This bundle consists of: `L3-507-001-001-T1-reference-topup.md`, `L3-507-001-001-T1-reference-topup.json`, `code/`, `reports/`.

| File | Role | How to regenerate |
|------|------|-------------------|
| `L3-507-001-001-T1-reference-topup.md` | Source of truth — edit this | Human or LLM |
| `L3-507-001-001-T1-reference-topup.json` | Structured metadata for the registry | LLM regenerates from the sections above |
| `code/solver.*` | Reference implementation | LLM regenerates from §1 Solver and §2 Operating point |
| `reports/results.md` | Reconstruction results | **Generated by running the solver** — do not fabricate |
| `reports/reproduce.md` | Step-by-step verification recipe | LLM regenerates as a recipe matching what was actually run |

**Prompt for your LLM after editing this Markdown:**

> Read the attached Markdown. Regenerate the sibling `.json` so every field matches.
> Also regenerate `code/solver.*` and `reports/reproduce.md` per the schema.
> Output each file in its own fenced code block tagged with the filename.
> Do NOT fabricate `reports/results.md` — it must come from actually running the solver.

_This Markdown was auto-synthesized from the catalog row for `L3-507-001-001-T1-reference-topup`._
_Edit it, regenerate the JSON, and submit at [/submit](/submit) to claim the artifact._