Nishi Family › Compare › STEM and Digital-Twin Fabrication
Nishi Compare · measured, not asserted
STEM and Digital-Twin Fabrication
Nishi vs the field — every Nishi cell is measured against real organ source at emit time; each gap names the watch contract that will close it.
Grow STEM from the first byte up and digital-twin every physical device so we test BEFORE we buy or build. Cols are 4 category leaders; the note names each row true rival (Cura, Klipper, Gazebo, ngspice, Ansys, QEMU). Measured this session: rollup 328 permille, omniforge 67 cells, math 12/61, sci 21/36.
Layer 1 · Executive
Where we are. Measured 2026-08-19. Sovereign fabrication and digital twins are strong: a deterministic no-float CAD kernel that exceeds B-rep on robustness, a silicon twin validated against QEMU by difftest and fuzz, slicing, G-code validation, Gerber and CAM, a print sim and a board twin. Not yet validated on metal: the print and PCB twins need external-oracle validation, and robotics physics, FEA, a CAD GUI and general symbolic solve are honest gaps.
Where we need to go. Validate before hardware: prove each twin predicts a REAL failure the SOTA tool misses (print against a real part, board against ngspice), then physics depth, then authoring -- every twin earning its exceed by an external oracle, never by assertion.
The unit. 1 u = one measured session-leg. Calibration from landed rungs: the mangagen panel compositor went from existing substrate to shipped and live-verified in ONE leg (2026-08-13); the citations rung went from 3 to 55 domains in one leg across seven seats (2026-08-18); a greenfield engine with a bite-proven gate has measured 2 to 4 legs. Estimates recalibrate as rungs land and PR7 actuals write back.
Cost to validate before hardware: 3.5 u. Through M0; the print validation needs a real print and is a hardware-lane rung.
Cost to physics depth: 9.5 u. Through M1.
Cost to authoring and fab completeness: 16 u. Everything below.
20 of 28 capabilities measured|2 of them measured exceeds|8 open|coverage 714/1000|adoption 9 full / 11 partial
Layer 2 · Roadmap
Do this next — computed by the ranker, never chosen by a seat
Order from nx_compare_rank (nx_dr_ocm: (deficit + cost-of-delay + option + enables) x sponsor x self-sufficiency x momentum / cost). FINISH rows are rungs whose symbol is present but whose organ is short of full adoption: the cheapest closures on this board, listed before any new work. Stamp: # asof=1787883280 domain=stem target_version=0.1 rungs=7 done=0 open=7 finish=0 ranker=nx_dr_ocm
| # | Stage | Rung | Priority | Derivation |
|---|---|---|---|---|
| #1 | 0.1 | SPICE oracle for the board twin (R1) nx_pcb_twin_spice_oracle | 66 | v=1 m=1 c=15 |
| #2 | 0.1 | Print twin validated against a real print (R0) psv_compare | 50 | v=1 m=1 c=20 |
| #3 | later | Finite-element solver (R2) fea_solve | 233 | v=7 m=1 c=30 |
| #4 | later | CAD GUI and constraint solver (R4) cg_sketch_solve | 200 | v=6 m=1 c=30 |
| #5 | later | General symbolic solve (R5) nx_calc_solve_general | 100 | v=2 m=1 c=20 |
| #6 | later | Pick-and-place placement (R6) nx_pcb_place | 66 | v=1 m=1 c=15 |
| #7 | later | Robotics physics twin (R3) rpt_step | 33 | v=1 m=1 c=30 |
Critical path — contract, done-rule, executor, cost
| Rung | Closes with | Definition of done (pre-declared) | Executor | Est. |
|---|---|---|---|---|
| Print twin validated against a real print (R0) | psv_compare | The twin's predicted sag, buckling and ringing compared against a real printed part's measured outcome on a banked case set; the twin must predict at least one REAL failure the slicer preview misses to earn the exceed label (CAXTON-class) | Hardware lane | 2 u |
| SPICE oracle for the board twin (R1) | nx_pcb_twin_spice_oracle | Differential test of the board twin against ngspice on banked netlists the way the silicon twin difftests against QEMU; node voltages within a PRE-DECLARED tolerance or the twin is RED | Organ | 1.5 u |
| Finite-element solver (R2) | fea_solve | A sovereign FEA solver over the existing stress and fluid primitives: tetra mesh, linear elastic, thermal; validated on textbook cases against closed-form answers and on one Ansys-published benchmark | Organ | 3 u |
| Robotics physics twin (R3) | rpt_step | Rigid-body and contact stepping with the never-brick clamps of the kinematics lane; a dropped-box and a gripper-contact case match MuJoCo-published trajectories within tolerance | Organ | 3 u |
| CAD GUI and constraint solver (R4) | cg_sketch_solve | Sketch constraints solved live with sliders regenerating the deterministic kernel; the no-float exceed survives: the same sketch regenerates byte-identical | Organ | 3 u |
| General symbolic solve (R5) | nx_calc_solve_general | Polynomial and transcendental solving with exact rationals where possible; a banked problem set with Wolfram answers as the oracle, pass count printed | Organ | 2 u |
| Pick-and-place placement (R6) | nx_pcb_place | Component placement with courtyard collision checks and centroid CSV for assembly; the existing DRC reads the result | Organ | 1.5 u |
Milestones
| Milestone | Rungs | Cumulative |
|---|---|---|
| M0 · Validate before hardware | R0,R1 | 3.5 u |
| M1 · Physics depth | R2,R3 | 9.5 u |
| M2 · Authoring | R4,R5 | 14.5 u |
| M3 · Fab completeness | R6 | 16 u |
Layer 3 · Engineering
How this is scored. Every Nishi mark is measured: the generator reads the real organ source on disk and requires the implementing symbol to exist (no self-grading). A watching tag names the organ and symbol contracted to close a gap — the mark flips itself on the next compare beat when that workstream ships, and the
comparewatch- plane row flips with it. The flip is necessary, not sufficient: it proves the symbol exists, never that the capability is good. The bar is the rung's pre-declared done-rule, proven by its gate — a symbol shipped without the behaviour behind it is a defect, and the flip is exactly what makes that defect visible instead of quiet. Competitor marks record documented capability presence — presence, not depth or scale. Adoption is measured too: every measured row carries where its organ stands on the estate's ladder (source → built → promoted → registered → invoked; libraries by importer reach minus validation importers; gates by the execution surfaces that run them). A row is fully adopted only at the top of its ladder; anything short is tagged partial with the exact remedy, so a build nobody promoted can no longer read as shipped. Census stamps: importers asof 1787849099, gate census asof 1787855507 (unix seconds; -1 = census absent).Capability matrix — measured against source
◉ leads / measured exceed● present◐ partial○ absent · click any capability for its evidence
| Capability | Nishi | Wolfram Language | Fusion 360 | KiCad | Verilator + QEMU |
|---|---|---|---|---|---|
| Math | |||||
symbolic calculus and integrationMeasured:nx_calc_integrate exists in runtime/nx_calc_integrate.nx, verified at emit. Wolfram and SymPy lead CAS depth; Nishi has sovereign integer-exact symbolic integration but DLMF breadth is only about 30 percent [wolfram-docs] Adoption: LIB-GATE-ONLY importers=1 — PARTIAL: imported only by validation organs (gates, tests, benches): wire it into a shipping program. | ● | ◉ | ○ | ○ | ○ |
equation solvingMeasured:nx_calc_solve_quadratic exists in runtime/nx_calc_solve.nx, verified at emit. Wolfram Solve is Best; Nishi quadratic and linear present, general symbolic solve is the named gap Adoption: LIB-GATE-ONLY importers=1 — PARTIAL: imported only by validation organs (gates, tests, benches): wire it into a shipping program. | ● | ◉ | ○ | ○ | ○ |
automatic differentiationMeasured:ta_leaf exists in runtime/nx_autograd_tensor.nx, verified at emit. vs Wolfram and JAX; Nishi reverse-mode autodiff is bit-exact no-float, first LM beat the analytic midpoint Adoption: SOURCE-ONLY — PARTIAL: source exists, never compiled: /api/build it. | ● | ● | ○ | ○ | ○ |
number theoryMeasured:nx_check_prop exists in runtime/nx_triangulation_number_theory.nx, verified at emit. Wolfram Best; Nishi primes, gcd and modular arithmetic present Adoption: SOURCE-ONLY — PARTIAL: source exists, never compiled: /api/build it. | ● | ◉ | ○ | ○ | ○ |
| Chem | |||||
periodic, molecule graph, SMILES, isotopesMeasured:nx_chem_periodic exists in runtime/nx_chem_periodic.nx, verified at emit. vs Wolfram and RDKit; Z1-118, SMILES, Morgan canonicalization, isotopes NIST and CIAAW verified [nist-ciaaw-atomic-weights]; thermo and kinetics absent Adoption: LIB-WIRED importers=20 nonval=6 — fully adopted (top of its ladder). | ● | ● | ○ | ○ | ○ |
| Bio | |||||
sequence alignment Gotoh and FASTQ to VCFMeasured:nx_align_affine exists in runtime/nx_align_affine.nx, verified at emit. vs BWA and GATK; Smith-Waterman, Gotoh affine [gotoh1982] and variant pipeline present; phylogenetics and protein structure absent Adoption: LIB-WIRED importers=2 nonval=1 — fully adopted (top of its ladder). | ● | ○ | ○ | ○ | ○ |
| Physics | |||||
statics, stress and fluid primitivesMeasured:nx_stress exists in runtime/nx_stress.nx, verified at emit. vs Wolfram and Ansys; force, stress and fluid primitives present but a unified mechanics, thermal, EM and ODE engine is absent Adoption: LIB-GATE-ONLY importers=1 — PARTIAL: imported only by validation organs (gates, tests, benches): wire it into a shipping program. | ● | ● | ◐ | ○ | ○ |
| Fab-CAD | |||||
kernel deterministic no-float CSG and filletsMeasured exceed:nx_cad_exceed in runtime/nx_cad_exceed.nx, verified at emit. EXCEED measured: byte-identical determinism plus robust SDF CSG and fillets where Fusion and FreeCAD B-rep abort on edge cases Adoption: LIB-WIRED importers=2 nonval=1 — fully adopted (top of its ladder). | ◉ | ○ | ◉ | ● | ○ |
STL and mesh exportMeasured:nx_stl_write exists in runtime/nx_stl_write.nx, verified at emit. vs Fusion and FreeCAD exporters; units-exact binary STL writer, gate 12 of 12 Adoption: LIB-WIRED importers=26 nonval=4 — fully adopted (top of its ladder). | ● | ○ | ● | ● | ○ |
interactive GUI editor and constraint solverOpen — watchingruntime/nx_cad_gui.nx : cg_sketch_solve, re-measured on every compare beat. Ship that symbol and this mark flips itself; the comparewatch- plane row flips with it. the number one behind axis vs Fusion and FreeCAD and SolveSpace [fusion360-docs]; sliders-to-regen GUI and sketch constraints are absent | ○ | ○ | ◉ | ◉ | ○ |
| Fab-Print | |||||
slicer STL to G-code watertightMeasured:nx_slice_pipeline exists in runtime/nx_slice_pipeline.nx, verified at emit. true rival Cura, PrusaSlicer, OrcaSlicer; Nishi watertight adaptive load-aware slicing present, composes 3 caps no single slicer emits together Adoption: LIB-WIRED importers=25 nonval=5 — fully adopted (top of its ladder). | ● | ○ | ○ | ○ | ○ |
G-code and Klipper validationMeasured:nx_klipper_gcode_validator exists in runtime/nx_klipper_gcode_validator.nx, verified at emit. true rival Klipper and Moonraker; Nishi validates the whole job before it is ever sent to the machine Adoption: LIB-GATE-ONLY importers=3 — PARTIAL: imported only by validation organs (gates, tests, benches): wire it into a shipping program. | ● | ○ | ○ | ○ | ○ |
printer transport Moonraker HTTPMeasured:nx_mr_build_post_request exists in runtime/nx_moonraker_client.nx, verified at emit. true rival Moonraker and OctoPrint; sovereign Klipper API client, ready-check and end-to-end loop proven green Adoption: LIB-WIRED importers=12 nonval=5 — fully adopted (top of its ladder). | ● | ○ | ○ | ○ | ○ |
| Fab-PCB | |||||
Gerber, Excellon, DRC, autoroute, CAMMeasured:nx_pcb_gerber exists in runtime/nx_pcb_gerber.nx, verified at emit. vs KiCad [kicad-docs]; full manufacturable package plus in-house isolation-mill CAM present, pick-and-place placement absent Adoption: LIB-WIRED importers=3 nonval=2 — fully adopted (top of its ladder). | ● | ○ | ○ | ● | ○ |
| Fab-Silicon | |||||
sovereign synthesis to nxgate netlistMeasured:nx_nxgate_sim exists in runtime/nx_nxgate_sim.nx, verified at emit. true rival Yosys; sovereign divider-backed gate synthesis emit, whole RV64IM decoder emittable Adoption: LIB-WIRED importers=9 nonval=5 — fully adopted (top of its ladder). | ● | ○ | ○ | ○ | ● |
| Fab-Robotics | |||||
kinematics, control, signal, timingMeasured:nx_robot_kinematics exists in runtime/nx_robot_kinematics.nx, verified at emit. true rival ROS, Klipper and GRBL; 7 rungs sim-proven with never-brick clamps, physical bench is the gap Adoption: LIB-GATE-ONLY importers=1 — PARTIAL: imported only by validation organs (gates, tests, benches): wire it into a shipping program. | ● | ○ | ○ | ○ | ○ |
| Twin-Print | |||||
3D-print sim thermal flow mech adhesionMeasured:nx_print_sim exists in runtime/nx_print_sim.nx, verified at emit. true rival Simufact Additive and Cura preview; predicts sag, buckling, collision and ringing; NOT yet validated against a real print Adoption: LIB-GATE-ONLY importers=2 — PARTIAL: imported only by validation organs (gates, tests, benches): wire it into a shipping program. | ● | ○ | ◐ | ○ | ○ |
validated vs real outcome CAXTON-classOpen — watchingruntime/nx_print_sim_validate.nx : psv_compare, re-measured on every compare beat. Ship that symbol and this mark flips itself; the comparewatch- plane row flips with it. the honest gap: the twin must predict a REAL failure the SOTA tool misses to earn the exceed label [brion2022-caxton] | ○ | ○ | ○ | ○ | ○ |
| Twin-Silicon | |||||
RV64 SoC sim mmu uart virtio kernel-bootMeasured:nx_rv64_sim exists in runtime/nx_rv64_sim.nx, verified at emit. Verilator and QEMU are Best; Nishi RV64 SoC sim boots a real kernel with peripherals [riscv-isa-spec] Adoption: SOURCE-ONLY — PARTIAL: source exists, never compiled: /api/build it. | ● | ○ | ○ | ○ | ◉ |
validated vs QEMU oracle difftest and fuzzMeasured exceed:nx_rv64_qemu_oracle in runtime/_hdl_build/nx_rv64_qemu_oracle.nx, verified at emit. EXCEED method: differential test against the QEMU golden model [qemu-docs] plus fuzzing is the SOTA validation method; this leg can test the core BEFORE the ULX3S is bought Adoption: BUILT-UNPROMOTED — PARTIAL: compiled, never promoted to the serving root: /api/promote it. | ◉ | ○ | ○ | ○ | ◉ |
gate-level nxgate netlist simulatorMeasured:nx_nxgate_sim exists in runtime/nx_nxgate_sim.nx, verified at emit. Verilator gate sim is Best [verilator]; Nishi has a sovereign netlist simulator, functional equivalence proven Adoption: LIB-WIRED importers=9 nonval=5 — fully adopted (top of its ladder). | ● | ○ | ○ | ○ | ◉ |
| Twin-PCB | |||||
board digital twin electricalMeasured:nx_pcb_twin exists in runtime/_hdl_build/nx_pcb_twin.nx, verified at emit. true rival ngspice and KiCad SPICE; board twin present, SPICE-oracle validation is the next rung Adoption: LIB-GATE-ONLY importers=1 — PARTIAL: imported only by validation organs (gates, tests, benches): wire it into a shipping program. | ● | ○ | ○ | ● | ○ |
| Twin-CAD | |||||
assembly twin explode and part-fitOpen — no implementing organ is measured for this axis yet. vs Fusion assembly; sovereign explode and part-fit twin, benchmarked on a car model | ○ | ○ | ● | ○ | ○ |
| Twin-Robotics | |||||
rigid-body and contact physics twinOpen — watchingruntime/nx_robot_phys_twin.nx : rpt_step, re-measured on every compare beat. Ship that symbol and this mark flips itself; the comparewatch- plane row flips with it. the gap vs Gazebo, MuJoCo and Isaac Sim; needed to test real motion and contact before committing hardware | ○ | ○ | ○ | ○ | ○ |
Twin-Multiphysics: FEA solver warp and stress before buildOpen — watchingruntime/nx_fea_solver.nx : fea_solve, re-measured on every compare beat. Ship that symbol and this mark flips itself; the comparewatch- plane row flips with it. Ansys is Best; Nishi has stress and fluid primitives but no full finite-element solver yet | ○ | ◉ | ◐ | ○ | ○ |
| Math | |||||
general symbolic solveOpen — watchingruntime/nx_calc_solve.nx : nx_calc_solve_general, re-measured on every compare beat. Ship that symbol and this mark flips itself; the comparewatch- plane row flips with it. The named gap of the equation-solving row: polynomial and transcendental solve beyond quadratic and linear | ○ | ◉ | ○ | ○ | ○ |
| Twin-PCB | |||||
SPICE-oracle validation of the board twinOpen — watchingruntime/_hdl_build/nx_pcb_twin.nx : nx_pcb_twin_spice_oracle, re-measured on every compare beat. Ship that symbol and this mark flips itself; the comparewatch- plane row flips with it. The next rung named on the board-twin row: differential test against ngspice the way the silicon twin tests against QEMU | ○ | ○ | ○ | ● | ○ |
| Fab-PCB | |||||
pick-and-place placementOpen — watchingruntime/nx_pcb_gerber.nx : nx_pcb_place, re-measured on every compare beat. Ship that symbol and this mark flips itself; the comparewatch- plane row flips with it. Named absent on the Gerber row: component placement and centroid export for assembly | ○ | ○ | ○ | ● | ○ |
On these two registers. Rows are declared in the domain's plan file and carry the debt id, which is the join key back to the sovereign debt plane — that plane, not this page, is the authority on state. Reconciling them automatically (the regen reading the plane and refreshing these rows) is a named, owed rung; until it lands, treat an id here as a pointer to look up, not a status to trust.
Honest verdict. Nishi is strong on the sovereign fabrication and digital-twin axes: a deterministic no-float CAD kernel that exceeds B-rep on robustness, and a silicon twin validated against QEMU (difftest plus fuzz) that lets us test the core BEFORE buying the board. It is NOT yet validated on metal -- the print and PCB twins need external-oracle validation, and the robotics-physics twin and multiphysics FEA are honest gaps. The climb is validate-before-hardware: prove each twin predicts a REAL failure the SOTA tool misses, then close the on-metal loop.
Person · product · place — not yet measured for this domain
Every compare carries this layer. Declare
knowledge/compare/stem.ppp (rows surface|nishi or c1..c4|label|url|connect naming OUR live surface and each rival's front door), run nx_ppp_probe domain stem, and this section fills itself on the next beat: the same ruler on both sides — privacy and CX (third-party hosts, tracker classes, cookies, security headers), design and longevity (design hygiene, computed WCAG contrast, render-blocking resources, unsized media, script weight, theme and motion queries), findability (landmarks, skip link, on-site search, breadcrumb, headings, internal links).References
Beyond a link list. Every reference below resolves twice — the publisher's copy and, where banked, the estate's own non-rottable library mirror with a content pin — and carries its evidence class plus the exact claim on this page it grounds. Keyed marks like [key] in the matrix notes jump here. A dash means honestly absent, never assumed.
- [wolfram-docs] Wolfram Research. Wolfram Language and System Documentation Center (reference.wolfram.com) -- Integrate, Solve, number theory and the full symbolic stack. Accessed August 2026. publisher · read in our library
knowledge/fetched/cmp_stem_wolfram-docs.html· pinh2ddd3b6d60f0f714fc5515a7ad4ecf4c29d567599952e9ac655e5a15c25b0eb7· accessed 2026-08-18 · vendor-docGrounds: The Wolfram Language column (c1) on the Math rows -- Math: symbolic calculus and integration, Math: equation solving, Math: number theory -- coded Best from Wolfram's own documented Integrate, Solve and number-theory coverage; the note's DLMF-breadth-about-30-percent is the honest distance our nx_calc_integrate has to close. - [fusion360-docs] Autodesk. Autodesk Fusion -- integrated cloud-based 3D CAD, CAM, CAE and PCB platform (product overview page). Accessed August 2026; the vendor page answers 403 to non-browser fetchers, so no mirror is claimed -- URL and title confirmed by an independent search-index read. publisher · accessed 2026-08-18 · vendor-docGrounds: The Fusion 360 column (c2): Fab-CAD: kernel deterministic no-float CSG and fillets (Fusion B-rep is the rival our SDF CSG exceed is measured against), Fab-CAD: interactive GUI editor and constraint solver (Fusion Best, the number-one behind axis) and Twin-Multiphysics: FEA solver warp and stress before build (Fusion simulation beside Ansys).
- [kicad-docs] KiCad EDA. KiCad Documentation (docs.kicad.org): schematic editor, PCB editor, Gerber and Excellon plotting, design-rule check and interactive router. Accessed August 2026. publisher · read in our library
knowledge/fetched/cmp_stem_kicad-docs.html· pinh7874bccd50f7524671d4582189cb26e0888c8e6ec8192e0f1ccfb2e7a063d8c2· accessed 2026-08-18 · vendor-docGrounds: The KiCad column (c3): Fab-PCB: Gerber, Excellon, DRC, autoroute, CAM (KiCad coded Yes, our manufacturable package is graded against its plot and DRC outputs) and Twin-PCB: board digital twin electrical (KiCad SPICE with ngspice is the named rival for the next SPICE-oracle rung). - [verilator] Snyder, W. et al. Verilator -- the open-source SystemVerilog simulator that compiles HDL to a cycle-accurate C++ or SystemC model; repository README plus the Verilator 5.x User's Guide (verilator.org/guide/latest). Accessed August 2026. publisher · read in our library
knowledge/fetched/cmp_stem_verilator-docs.html· pinh13ca8c49218287cc006199c3b83770cfbc5dc75b36ff50b41ee5736305d027d3· accessed 2026-08-18 · vendor-docGrounds: Half of the Verilator + QEMU column (c4): Twin-Silicon: gate-level nxgate netlist simulator (Verilator gate sim coded Best; our sovereign netlist simulator's functional-equivalence proof is graded against it) and Fab-Silicon: sovereign synthesis to nxgate netlist. - [qemu-docs] QEMU Project. QEMU documentation (system emulation, TCG, RISC-V machine targets). Accessed August 2026. publisher · read in our library
knowledge/fetched/cmp_stem_qemu-docs.html· pinhcd28374cfd9b1c35b6850e800a27ee831f401b0c5e35b6732634ba284c903040· accessed 2026-08-18 · vendor-docGrounds: The other half of c4: Twin-Silicon: RV64 SoC sim mmu uart virtio kernel-boot (QEMU coded Best) and Twin-Silicon: validated vs QEMU oracle difftest and fuzz -- QEMU is the golden model nx_rv64_qemu_oracle differential-tests against, which is what lets the core be tested BEFORE the ULX3S is bought. - [riscv-isa-spec] RISC-V International. The RISC-V Instruction Set Manual, Volume I Unprivileged ISA and Volume II Privileged Architecture -- ratified specifications page. Accessed August 2026. publisher · read in our library
knowledge/fetched/cmp_stem_riscv-isa-spec.html· pinh33461bf0413542c87c32a430032d8dacfb57f8a454f8c3c3aaa925e9b7e6d15a· accessed 2026-08-18 · published-standardGrounds: The contract every Twin-Silicon row is measured against: the RV64 SoC sim, the QEMU-oracle difftest and the RV64IM decoder emitted by Fab-Silicon: sovereign synthesis to nxgate netlist all implement this ISA -- conformance to it is what makes QEMU a valid oracle rather than just another simulator. - [nist-ciaaw-atomic-weights] NIST Physical Measurement Laboratory. Atomic Weights and Isotopic Compositions with Relative Atomic Masses (Coursey, Schwab, Tsai, Dragoset), with the IUPAC CIAAW Standard Atomic Weights table (ciaaw.org/atomic-weights.htm, fetched and mirrored the same day as knowledge/fetched/cmp_stem_ciaaw-atomic-weights.html). Accessed August 2026. publisher · read in our library
knowledge/fetched/cmp_stem_nist-atomic-weights.html· pinh65f4063b547aa1dab9aa71e31cb622be6d68788a03fa77c07f5efd5bfbd9e697· accessed 2026-08-18 · published-standardGrounds: Chem: periodic, molecule graph, SMILES, isotopes -- the isotopic masses and abundances nx_chem_periodic carries for Z1-118 were verified against the NIST table and the standard atomic weights against CIAAW, which is exactly the note's isotopes NIST and CIAAW verified claim. - [gotoh1982] Gotoh, O. An improved algorithm for matching biological sequences. Journal of Molecular Biology 162(3):705-708, 1982. PMID 7166760, doi 10.1016/0022-2836(82)90398-9. Confirmed by search-index read; the PubMed page serves a JS bot-challenge stub to non-browser fetchers, so no mirror is claimed. publisher · accessed 2026-08-18 · published-paperGrounds: Bio: sequence alignment Gotoh and FASTQ to VCF -- the affine-gap O(mn) alignment nx_align_affine implements is this paper's algorithm, the Smith-Waterman-Gotoh form that BWA-class aligners (the row's named rivals) descend from.
- [brion2022-caxton] Brion, D.A.J., Pattinson, S.W. Generalisable 3D printing error detection and correction via multi-head neural networks. Nature Communications 13, 4654 (2022); the CAXTON system and dataset (1.2 million labelled extrusion images from 192 parts). Open-access copy at PMC9378646. publisher · read in our library
knowledge/fetched/cmp_stem_brion2022-caxton.html· pinhdcaacc706ef6d0250a47641d688a8f1775843ea28aa38d8399e6ee70a25bb68e· accessed 2026-08-18 · published-paperGrounds: Twin-Print: validated vs real outcome CAXTON-class -- the honest-gap row names this work as the bar: a print twin earns the exceed label only when it predicts a REAL failure of the kind CAXTON detects and corrects on camera, which is what nx_print_sim_validate must be measured against before Twin-Print: 3D-print sim can claim more than sim-proven.
generated by nx_swcompare_matrix (sovereign NishiLang organ) from knowledge/compare/stem.matrix · every Nishi cell verified against organ source at emit time · watch cells re-measured on every compare beat · zero JS, zero trackers