"The architecture is not the design of space, it is the design of the biological response to the environment."
MARS-OS (Martian Adaptive Rhythm Synchronization — Operations Support) is a computational framework that quantifies biological time drift in human crews during Mars surface operations, models its cognitive performance impact, and generates photobiological intervention schedules to maintain operational readiness.
The core problem: Mars rotates in 24h 37m. The human circadian system is calibrated to 24h. That 617-second daily mismatch is silent, inevitable, and (without active management) mission-critical. By Sol 10 without intervention, accumulated drift exceeds 6 hours: the documented threshold for mission-critical cognitive impairment.
This is Level 3 of a five-level research architecture on Human Performance and Circadian Systems in Constrained Environments. See RESEARCH_LINE.md.
mars-os-circadian-kernel/
│
├── mars_os/ # Python package
│ ├── __init__.py
│ └── circadian_kernel.py # Core computational engine
│
├── tests/ # pytest unit suite
│ ├── __init__.py
│ └── test_kernel.py # 63 tests — 100% pass rate
│
├── db/
│ └── schema.sql # PostgreSQL / SQLite schema
│ # missions · crew_profiles · sol_states
│ # interventions · simulation_runs · views
│
├── docs/
│ └── METHODOLOGY.md # Model derivation, validation, references
│
├── joss/
│ ├── paper.md # JOSS submission paper
│ └── paper.bib
│
├── .github/workflows/
│ └── ci.yml # CI: pytest × Python 3.10 / 3.11 / 3.12
│
├── app.py # Streamlit dashboard
├── CITATION.cff # Software citation (renders on GitHub)
├── RESEARCH_LINE.md # Full 5-level research architecture
├── LICENSE # MIT
├── .gitignore
├── README.md
└── requirements.txt
| Parameter | Earth | Mars | Delta |
|---|---|---|---|
| Rotation period | 24.000 h | 24.617 h | +37 min/sol |
| Drift @ Sol 10 | 0.0 h | 6.17 h | Critical threshold |
| Drift @ Sol 30 | 0.0 h | 18.5 h | Severe desynchrony |
| Primary zeitgeber | Solar (24h) | Solar (24.617h) | Insufficient recalibration |
Uncorrected circadian desynchrony produces: 20–40% cognitive throughput reduction, increased EVA decision latency, REM degradation, immune suppression. Documented in NASA HI-SEAS isolation studies and polar overwinter research.
┌────────────────────────────────────────────────────────────┐
│ MARS-OS KERNEL │
│ │
│ INPUT: Sol counter · Chronotype · Correction factor │
│ │ │
│ VAN DER POL SCN MODEL (Kronauer 1999) │
│ dX/dt = Y │
│ dY/dt = μ(1−X²)Y − ω²X + F(t) │
│ F(t) = Mars light-dark forcing at 24.617h │
│ │ │
│ DRIFT ENGINE │
│ D_eff = D_raw × (1 − c × 0.72) [max 72%, Lockley 2003]│
│ │ │
│ PERFORMANCE MODEL │
│ PI = 1 / (1 + e^(0.8 × (d mod 6 − 3))) │
│ Calibrated to NASA astronaut fatigue data │
│ │ │
│ OUTPUT: Alert (GREEN/YELLOW/RED) · Light Rx (480nm PRC) │
└────────────────────────────────────────────────────────────┘
│
▼
┌──────────────────────────┐
│ db/schema.sql │
│ PostgreSQL / SQLite │
│ missions │
│ crew_profiles │
│ sol_states ← primary │
│ interventions │
│ simulation_runs │
└──────────────────────────┘
The suprachiasmatic nucleus (SCN) is a self-sustaining limit-cycle oscillator. Its behavior is captured by the Van der Pol equation — a nonlinear ODE where the oscillation amplitude self-regulates and external forcing (zeitgebers) can phase-shift the rhythm. This model was validated for mammalian circadian systems by Kronauer et al. (1999) and remains standard in quantitative chronobiology.
The key physical insight: the SCN is not a simple 24h timer that can be trivially re-set. It is a nonlinear oscillator with a bounded daily entrainment capacity (~1–2h/day maximum phase shift). Mars creates a forcing mismatch that, without intervention, exceeds the system's natural correction bandwidth.
Melanopsin-expressing ipRGCs (intrinsically photosensitive retinal ganglion cells) project directly to the SCN via the retinohypothalamic tract. These cells peak in sensitivity at 480nm blue light. Lockley et al. (2003) demonstrated that narrow-band 480nm exposure produces phase shifts 2.5× more effective than broadband white light at equivalent photon densities, establishing the PRC-based intervention protocol implemented here.
Maximum achievable correction: 72% of raw drift per cycle.
The Performance Index (PI) follows a sigmoid degradation function calibrated against NASA astronaut cognitive performance data. Drift is taken modulo CRITICAL_DRIFT_H (6h) — reflecting the periodic nature of sleep-wake cycles. The mission-critical threshold occurs at approximately Sol 9.7 without intervention.
git clone https://github.com/[your-username]/mars-os-circadian-kernel
cd mars-os-circadian-kernel
pip install -r requirements.txt
# Streamlit dashboard
streamlit run app.py
# CLI
python -m mars_os.circadian_kernel
# Tests
pytest tests/ -vUse as a module:
from mars_os import MissionConfig, run_mission, to_dataframe
config = MissionConfig(
duration_sols = 180,
chronotype = "intermediate",
correction_factor = 0.60,
base_wake_time_h = 7.0,
)
states = run_mission(config)
df = to_dataframe(states)
print(df[["sol", "drift_h", "performance_index", "alert_level"]].head(15))Database:
# SQLite
sqlite3 mars_os.db < db/schema.sql
# PostgreSQL
psql -d your_database -f db/schema.sql| Parameter | Default | Description |
|---|---|---|
duration_sols |
180 |
Mission length in sols |
chronotype |
"intermediate" |
"morning" / "intermediate" / "evening" |
correction_factor |
0.0 |
Phototherapy efficacy [0.0–1.0] |
base_wake_time_h |
7.0 |
Crew wake time (Mars local, h) |
| Function | Returns | Description |
|---|---|---|
run_mission(config) |
List[CircadianState] |
Full mission assessment |
to_dataframe(states) |
pd.DataFrame |
Analysis-ready export |
compare_scenarios(sols) |
pd.DataFrame |
4-scenario comparison |
compute_drift(sol, cf, chrono) |
float |
Effective drift (hours) |
performance_index(drift_h) |
float [0,1] |
Cognitive PI |
alert_level(pi) |
str |
GREEN / YELLOW / RED |
compute_intervention(drift_h) |
dict |
480nm Rx schedule |
simulate_scn(sols, mu, offset) |
(t, x) |
Van der Pol integration |
The schema converts single-run outputs into a persistent, queryable mission data system.
| Table | Purpose |
|---|---|
missions |
Mission registry |
crew_profiles |
Chronobiological parameters per crew member |
sol_states |
Per-sol PI and drift time series |
interventions |
Administered phototherapy with compliance tracking |
simulation_runs |
Reproducibility metadata |
Pre-built views: v_mission_summary · v_crew_alert_timeline · v_intervention_compliance.
| Level | Project | Environment | Status |
|---|---|---|---|
| 1 | Circadian Productivity Analyzer | Earth | Planned |
| 2 | Circadian Disruption Simulator | Aviation / shift work | Planned |
| 3 | MARS-OS: Circadian-Kernel | Mars (24.617h) | Active — v1.0 |
| 4 | Circadian Evolution Scenarios | Theoretical | Future |
| 5 | Cross-Body Comparative Model | Earth / Moon / Mars | Future |
Overarching Question: How does human performance adapt when biological time decouples from environmental time — and how do we engineer the bridge?
- v1.0 — Van der Pol SCN model · drift engine · sigmoid PI model · 480nm Rx · Streamlit dashboard · PostgreSQL schema · 63-test suite · CI/CD
- v1.1 — Multi-crew synchrony divergence
- v1.2 — Melatonin pharmacokinetics
- v2.0 — ML-based PI prediction from HRV / actigraphy
- v3.0 — Lunar non-24h extension (~708h cycle)
@software{palencia_robles_2026_marsos,
author = {Palencia Robles, Diego José},
title = {MARS-OS: Circadian-Kernel},
version = {1.0.0},
year = {2026},
url = {https://github.com/[your-username]/mars-os-circadian-kernel},
license = {MIT}
}Diego José Palencia Robles PhD Researcher | Systems Architect Human Performance & Circadian Systems in Constrained Environments Computational Physics · NLP · High-Performance ML
MIT — See LICENSE. Scientific use encouraged with attribution.
- Kronauer, R.E. et al. (1999). J Biol Rhythms 14(6).
- Lockley, S.W. et al. (2003). J Clin Endocrinol Metab 88(9).
- Barger, L.K. et al. (2014). Lancet Neurol 13(9).
- NASA Human Research Program (2019). Sleep, Circadian Rhythms, and Fatigue — Evidence Report.
- Monk, T.H. et al. (2001). Chronobiol Int 18(6).