A native, formally verified signal processing engine for astronomical baseband recordings.
New to TurboKain? Read the complete, hands-on User Guide (
docs/USER_GUIDE.md) for plain-English walk-throughs of all 14 instruments, configuration presets, Python automation, and pipeline recipes.
Classical Search for Extraterrestrial Intelligence (SETI) historically presupposes intentional, high-power narrowband isotropic beacons directed at the Solar System. Modern communication theory and orbital link budgets dictate that advanced intelligences communicating across interstellar baselines will optimize strictly for channel capacity and energy efficiency (
- Point-to-Point and Off-Axis: Directed between non-terrestrial nodes; observable from Earth only when our line of sight intercepts the transmission beam, its forward sidelobes, or interstellar scattering volume.
- Noise-Matched: Modulated using high-order constellations, spread-spectrum coding, and forward error correction (FEC). Under power-detection methods, optimal transmissions are mathematically indistinguishable from Gaussian thermal noise.
- Continuous and Persistent: Operating over decade- to millennial-scale baselines, exhibiting cyclic framing, telemetry synchronization, and packetized framing.
TurboKain is designed for the Bystander Mission: extracting structural signatures that thermal noise cannot produce from raw dual-polarization radio telescope baseband recordings.
The pipeline processes baseband voltage streams down to phase coherence, cyclostationary spectral correlation densities, dispersed pulse trains, chirped Doppler carriers, microsecond-scale autocorrelation lattices, and post-Shannon symbolic machine execution.
TurboKain applies the SQLite / BusyBox doctrine to high-performance astrophysics pipelines: source code is maintained in strictly decoupled, domain-isolated modules during development, then fused into a single whole-program translation unit for native compilation.
kain/core/*.kn ──► kain amalgamate --raw kain/core -o kain/core.kn ──► kain build kain/core.kn ──► core.exe (~1.2 MB)
(14 modules) (whole-program LLVM) │
├── core <tool> [args...]
├── core help <tool>
├── core prove
└── core sweep <f32>
- Whole-Program Optimization (WPO): Amalgamating into core.knexposes the entire call graph to LLVM. The optimizer performs aggressive inter-procedural inlining, dead-code elimination, and loop vectorization across instrument boundaries.
- Hermetic Portability: Compiles into a single self-contained executable (core.exe, ~1.2 MB) linked directly against operating system system-call boundaries (kernel32). It requires zero runtime dependencies, interpreters, or shared libraries.
- Direct Memory Handoffs: Instruments share contiguous memory arenas (ByteandFloatmemory regions) without disk roundtrips.
- Dual Invocation Model: dispatch.kncaptures invocation context viaGetCommandLineA(). It operates as a subcommand suite (core <tool> [args...]) or as a multi-call binary (copying or linkingcore.exeto<tool>.exeexecutes that instrument directly).
The core engine comprises 14 specialized instruments spanning the complete RF analysis chain:
For
Phase-modulated digital communications (BPSK, QPSK, FSK) exhibit non-zero spectral correlation at cyclic frequency
Cold plasma dispersion delays lower frequencies according to the dispersion measure (
Demodulated bitstreams are evaluated for computational density and non-random state transitions:
- Linear Complexity: Evaluated via the Berlekamp-Massey algorithm to determine the shortest linear feedback shift register (LFSR) capable of generating the sequence.
- Universal Computation: Bit sequences are seeded as execution memory in single-instruction computing engines (One-Instruction Set Computer / Subleq) and 1D Rule 110 cellular automata to detect self-propagating structures and halting properties.
TurboKain compiles directly from source through the native Kain compiler:
# 1. Synthesize the amalgamated single-file core
kain amalgamate --raw kain/core -o kain/core.kn
# 2. Compile to native executable
kain build kain/core.kn --target llvm -o core.exeEvery instrument contains mathematical self-tests verifying analytical bounds against synthetic Gaussian noise and injected reference signals. Run the full battery natively:
tkc prove
# or: core proveVerification output demonstrates zero-divergence against analytical ground truths:
================================================================================
TurboKain Core Suite — Unified Native Prove Battery (9 instruments)
================================================================================
[1/9] bitslice --prove -> receipt=PASS prove=4/4
[2/9] boxcar_bank --prove -> receipt=PASS prove=4/4
[3/9] config --prove -> receipt=PASS prove=6/6
[4/9] drift_hunt --prove -> receipt=PASS prove=4/4
[5/9] fil_reader --prove -> receipt=PASS prove=4/4
[6/9] frame_hunt --prove -> receipt=PASS prove=9/9
[7/9] lag_hunt --prove -> receipt=PASS prove=9/9
[8/9] xeno_scan --selftest -> [selftest] ALL PASS
[9/9] xvm_sandbox --selftest-> receipt=PASS selftest=24/24
================================================================================
Core Battery Receipt: ALL 9 PROVE BATTERIES PASSED (receipt=PASS)
================================================================================
The binary is aliased as tkc (TurboKain Core), core, and turbokain_core. Tool shorthands (fam, lag, boxcar, drift, frame, sk, xeno, xvm, bits, cad, cfg, fil) are supported out of the box.
# Master directory of all 14 tools and data flows
tkc help
# Detailed mathematical parameters, flags, and contracts for an instrument
tkc help slice
tkc help fam
tkc help boxcar
tkc help xvmExecute the complete 7-stage screening and detection battery on a voltage slice in a single pass:
tkc sweep <path_to_voltage.f32> --out-dir reports/target_sweep/ --fs 2929687.5This executes in sequence:
- sk_gate(Spectral kurtosis RFI screening)
- xeno_scan(Statistical anomaly lattice)
- boxcar_bank(Transient dispersed pulse detection)
- drift_hunt(Chirped carrier dedoppler extraction)
- frame_hunt(Harmonic comb and periodicity identification)
- lag_hunt(Direct time-domain autocorrelation lattice)
- fam_god(Cyclostationary spectral correlation density mapping)
Individual instruments execute directly with explicit argument contracts:
# Ingest 128 blocks of channel 44 from a raw GUPPI baseband file
tkc slice /data/raw/blc00_guppi.raw 44 /data/slices/ch44.f32 128 --pol 0
# Run multi-decade cyclostationary baud rate estimation
tkc fam --in /data/slices/ch44.f32 --fs 2929687.5 --segbank --out reports/fam.md
# Decimate and slice bits at detected baud rate
tkc bits --in /data/slices/ch44.f32 --alpha 11090.0 --out /data/bits/ch44_b11090
# Evaluate computational complexity in the symbolic execution sandbox
tkc xvm --in /data/bits/ch44_b11090.head.sign.bin --out reports/xvm.md
# Gate spatial persistence against an off-target reference observation
tkc cad --on reports/on_target.md --off reports/off_target.md --out reports/cadence.md- Noise-Matched Auditability: A non-detection is scientifically valid only when accompanied by explicit noise floor sensitivity measurements. Unsubstantiated negative results are prohibited.
- Immutable Ledgers: Every modification, build artifact, and observational verdict is logged sequentially in memory.tsv(change log) andcatalog.tsv(instrument ledger).
- Thresholds as Telemetry: Algorithmic thresholds must not be hardcoded in pipeline logic. All operating bounds, filter dimensions, and significance levels must derive from command-line arguments or formal configuration records (config.kn).
- Independent Veto Invariants: Candidate dispositions (CLEAN,WATCH,COMMON,CANDIDATE) are governed by formal logical invariants (lawblocks in Kain). Automated tools generate candidate metrics and evidence receipts; promotion to interstellar candidate status requires multi-epoch verification and human analyst adjudication.
TurboKain ingests raw baseband recordings from major radio observatories:
- Green Bank Telescope (GBT): GUPPI baseband format (2-bit and 8-bit complex voltage streams).
- Parkes Observatory (Murriyang): Multibeam baseband and filterbank archives.
- MeerKAT: High-density array voltage records.
All signal processing algorithms are validated against Python and C oracle benchmarks (SetiYeti) and verified across sky observations including FRB 121102, TRAPPIST-1, Sagittarius B2, and interstellar interloper 1I/'Oumuamua.