Foundation Connection: From baseCamp to Institutional Adoption
How validated baseCamp science flows through NUCLEUS composition into foundation's institutional adoption pipeline.
The Bridge
projectFOUNDATION is the outward-facing project in sporeGarden. While projectNUCLEUS focuses inward (deploy, validate, compose on real hardware), projectFOUNDATION focuses outward: institutional relationships, metallic bonding with university HPC, public documentation, and the case for sovereign compute.
The boundary: projectNUCLEUS proves that the patterns work. projectFOUNDATION makes the case to institutions that the proof matters.
What baseCamp Provides to foundation
The 27 baseCamp papers are not academic papers submitted to journals. They are executable evidence — each paper is a binary you can run that reproduces published, peer-reviewed results. This is the raw material for institutional adoption:
| What foundation Needs | What baseCamp Provides |
|---|---|
| “Can this replace our Python stack?” | 16,000+ checks showing Rust parity at machine-epsilon |
| “Does it work on our GPUs?” | Cross-vendor validation: NVIDIA, AMD, Intel — same results |
| “Is the science real?” | Published papers reproduced: Sarkas, Bazavov, Dong, Fajgenbaum |
| “How much does it cost?” | $0.044 per MD simulation, $0.58 per QCD production — on consumer GPUs |
| “Can we trust the results?” | BLAKE3 content hashing, Merkle roots, ed25519-witnessed attribution braids |
| “Will it run on our HPC?” | Same binaries, same checks, any x86_64/aarch64 Linux |
The Adoption Pipeline
baseCamp science (28 papers, 16K+ checks)
↓ validated on active gate (projectNUCLEUS)
↓ 13 primals, full provenance, ABG workloads
primals.eco/lab (public evidence)
↓ rendered notebooks, reproduce-it-yourself guide
↓ PI reviews at primals.eco or shared workspace
foundation (institutional bridge)
↓ grant appendix (NIH, NSF, USDA mapped)
↓ MSU asset acceleration (ICER mapping)
↓ drug discovery pipeline (pharma/biotech)
metallic bonding (institutional HPC)
↓ university clusters join as compute pools
↓ same deploy graphs, same primals, same validationDomain-Specific Foundation Arguments
Bioinformatics (wetSpring → foundation)
The argument: ABG members run real 16S/scRNA-seq/metagenomics workloads through the NUCLEUS composition. 235+ checks pass. Python→Rust parity at tol=0.000000. Provenance chain tracks every FASTQ from NCBI download to final braid.
What the PI sees: “My student ran a 16S pipeline on commodity hardware, got the same results as QIIME2, and the whole pipeline has cryptographic provenance. Can we run this on ICER?”
Physics (hotSpring → foundation)
The argument: Lattice QCD production on a $500 RTX 3090 for $0.58. Same physics as million-dollar HPC clusters. guideStone artifact that any physicist can verify in 3 minutes.
What the PI sees: “This reproduces our HotQCD EOS tables on a gaming GPU. The deconfinement transition at β=5.69 matches. What happens if we run this on 50 nodes?”
Drug Discovery (healthSpring → foundation)
The argument: NONMEM + Monolix + WinNonlin costs $6,500/year. healthSpring replaces all three, runs 84× faster, and adds Anderson gut lattice modeling. 329/329 Fajgenbaum pathway checks pass.
What the PI sees: “Sirolimus is correctly ranked #1 by the sovereign pipeline. The PK/PD matches our commercial tools. And it’s AGPL.”
Agriculture (airSpring → foundation)
The argument: FAO-56 ET₀ at R²=0.97 from free, open APIs (Open-Meteo, NOAA). 57 papers reproduced. No weather station subscriptions, no licensed datasets.
What the PI sees: “Real Michigan data, 15,300 station-days, and it matches our institutional data pipeline. On a laptop.”
The Metallic Bonding Endgame
When foundation establishes institutional relationships, the bonding model shifts:
| Phase | Bond | What Happens |
|---|---|---|
| Now (Phase 1-2) | Covalent + Ionic | Covalent LAN cluster + ABG ionic sharing |
| Next (Phase 3) | + Weak public | primals.eco self-hosted, public guideStone |
| Then (Phase 4) | + Metallic | University HPC joins as compute pool |
Metallic bonding means delocalized capabilities — the pool has aggregate TFLOPS, not per-node assignments. biomeOS load-balances across the metallic pool. Individual node identity matters less; aggregate capacity and availability matter.
The same deploy graphs that run on the covalent cluster run on ICER’s rack-mounted servers. The same primals. The same validation. The same provenance. The only difference is the hardware under the composition.
What sporePrint Shows
The Lab section on primals.eco is the public window into this pipeline:
- Validation results: 235+ checks with full provenance chains
- Reproduce It Yourself: Clone, deploy, run, verify — anyone can do it
- Provenance Pipeline: How every computation becomes cryptographically witnessed
When the springs fully evolve to composition — when every baseCamp paper runs through ToadStool dispatch on a live NUCLEUS with full provenance — the science becomes visible, verifiable, and adoptable. That is the projectFOUNDATION endgame: not “trust us” but “verify it yourself, on your own hardware.”