GitHub - Ibtisam-Mohammad/Fly.exe: Runs the complete released Drosophila male CNS connectome (165,122 neurons, 25.5M edges) inside a physical fly body, in closed loop, on one GPU.

MaleCNS Virtual Fly A Drosophila brain simulator that runs the whole Traced universe of the released male CNS connectome — 165,122 neurons and every one of the 25,563,197 edges between them — inside a physical fly body, in closed loop, on one GPU. That universe is 165,122 of the 166,700 bodies the MaleCNS paper annotates: the runtime graph is the subgraph induced by annotation status Traced, a decision recorded in ADR-2026-002. Widening it to Traced+Assign+Anchor would add 2,443 bodies (1.48%) and 60,281 edges (0.24%). It is not a recovered copy of the imaged fly, a complete biological emulation, or a digital twin. What it is, and the evidence behind every number on this page, is set out in What is and is not claimed. Twelve flies, twelve connectomes, one scene Twelve full NeuroMechFly bodies in one MuJoCo scene, scattered at random over a 30 mm disc and aimed at random. Each fly executes its own copy of that whole graph every 15 ms coupling interval, over one shared connectivity allocation with independent membrane, adaptation, spike-counter and noise state. Food spheres and pillars are geoms with real contact pairs, and the bodies share one solver step, so an object stops a fly through the physics rather than through a rule. Flies do not collide with each other — FlyGym gives every fly geom contype 0 and this world writes no fly-fly contact pair, so two bodies pass through one another. What couples them is vision: each fly enters the others' encoders as a sphere of one declared radius. That vision is analytic, not optical. The encoder is handed each object's exact position and radius and computes a bearing and an angular size from them; there are no camera pixels, no ray casting, no occlusion, no colour and no texture anywhere in the loop. Nothing steers them but two descending population rates read out of their own network. Onset is on a timer and the video says so: every fly stands still for 1.5 s by construction and then needs its drive held above threshold for 150 ms, so all twelve start walking at 1.665 s. The timer decides when. What the stimulus decides is whether — the control below never leaves the standing state at all. GIF Twelve independent neural states over one connectivity allocation. Brightness is spike count with a declared decay. 0.30x speed. Six legs in a tripod gait, adhesion switching per leg, every contact solved by MuJoCo. Exact against its control — same seed, same bodies, encoder held at baseline. Watch the full 1:38 video (1920x1080, 19 MB). What it measured, against the identical-seed stimulus-absent control: exact stimulus-absent flies that entered the locomoting state 12 / 12 0 / 12 flies that ended within 1 mm of an object's surface 12 / 12 0 / 12 neurons spiking per fly per coupling interval 9,337 164 straight-line displacement 5.0 to 25.6 mm 3.0 to 7.3 mm ended nearer a food object 12 / 12, median +11.48 mm 8 / 12, median +2.25 mm Eleven of twelve ended against a food sphere and the twelfth against a pillar. They were stopped by the object, they did not decide to stop: the decoder has no transition out of its locomoting state, so a blocked fly keeps being commanded forward. "Within 1 mm" is a two-dimensional thorax-centre distance minus the object radius, not MuJoCo contact telemetry. The last row is the one to be careful about, and the video says so on screen: a standing body drifts forward along its own axis, and headings are bounded so that food lies inside the encoder's mapped visual field, so the control leans the same way. Closing distance to food is not the discriminator. Ending against an object is — 12 of 12 against 0 of 12. This is one seed and one arena — the third arena, designed after measuring two that failed, with starting headings bounded so that food falls inside the encoder's mapped field. There is no multi-seed matrix, and no acceptance contract scores any of it. Two properties are checked rather than asserted, because the swarm reuses the frozen DEMO-01 network and reimplements its per-fly actuation. One fly driven through SwarmWorld and through Demo01VisualBody under an identical command sequence agree to 0.0 on every one of 133 qpos components; the multi-object encoder reproduces the frozen single-cue encoder to 1.1e-13 Hz on rates spanning 1 to 400 Hz. Both are in tests/test_swarm3d.py. Operator handoff: docs/showcase/SWARM3D.md. Decision record: ADR-2026-023, which also records the two arenas that were built, measured and discarded first — each one exposed a property of the frozen visual route that no previous experiment had tested. How it works MaleCNS v1.0 (CC-BY) 165,122 neurons, 25,563,197 edges, checksum-locked | transmitter sign per-neuron; unresolved signs are zeroed, not guessed | sparse graph build hashed arrays, verified on every load | GeNN / CUDA one connectivity allocation, N independent neuron states | ^ retinotopic | | two descending population rates lamina | | encoder | v MuJoCo + NeuroMechFly 133 qpos / 132 DOF, 42 actuated, contacts solved The full frame chain, including the causal queues that keep the body one interval behind the neural engine, is in docs/architecture.md. The loop is deliberately narrow and every narrowing is declared. Light enters one synapse downstream of the photoreceptors, because all 66,533 photoreceptor output edges are zeroed by the frozen unresolved-sign policy. The gait is a published pattern generator, not the simulated ventral nerve cord. The readout is two numbers. Everything else the body and the connectome could do is inert in this demonstration, and the inventory of what is inert is part of the artifact rather than a footnote. What is and is not claimed The project sits at tier V0 Structural on its own V0–V8 ladder. The swarm demonstration awards no tier at all: both the run summary and the render manifest carry validation_tier_awarded: null and evidence_grade: false, because no preregistered biological hypothesis and no acceptance contract exists for a swarm. It demonstrates machinery and validates no biology. Specifically not claimed, and printed on the video frames rather than hidden here: No social behaviour. Flies aggregate because a nearby fly is a large object in the visual field and the network approaches large objects. No foraging. Food is a coloured sphere with a radius. The encoder has no colour channel and cannot tell food from a pillar; what separates them is angular size. There is no ingestion, no proboscis extension, no taste channel in this run. The walking is engineered. No part of the simulated ventral nerve cord contributes to leg movement. The flies are not individuals. Twelve parameterised copies of one specimen, differing in where they start, what they see from there, and their independent noise stream. No fly-fly contact and no optical vision, as above: the bodies pass through one another and the encoder reads exact coordinates rather than pixels. The controls that would make this a swarm result do not exist yet. There is no flies-invisible arm, no swarm-specific readout ablation, no matched controller-only arm, no activity-matched shuffle, no multi-seed matrix and no equal-angular-size food-versus-pillar preference test. The published run is exact against stimulus-absent, and nothing more. Rendering is software rasterisation on the machine that produced these files, and the manifest records it. The discipline that produces those statements is the point of the project as much as the simulation is: every parameter carries a provenance class (M measured, P population prior, F fitted, E engineering scaffold, I irrecoverable) and an assumption ID in configs/assumptions.json; evidence-grade runs refuse to start from a dirty worktree; and criteria are registered before they are scored. Results that failed are kept — see docs/STATUS.md for the current state, including Track A's 0-of-30 grooming-displacement failure and a withdrawn evidence round. Quick start The lightweight reference engine runs before FlyGym, CUDA or the MaleCNS data are installed: uv python install 3.12 uv sync --python 3.12 --extra data --extra render uv run flysim run eon-demo --seed 1 --headless uv run flysim render runs/<run-id> uv run pytest # tests needing FlyGym, MuJoCo or CUDA skip That produces the semantic engineering storyboard, which is also the project's neural-bypass control. It is an E engineering scaffold and its manifest says so. For anything that executes the real connectome you need the dataset and an NVIDIA GPU: export FLYSIM_DATA_ROOT=/path/with/room # dataset + run outputs flysim data sync --profile starter # public, no credentials flysim data validate flysim data import-aggregate GeNN is a native CUDA source build rather than a registry package (scripts/install_genn.sh), and the pinned production environment is Linux. On Windows it is reached through WSL2. Full operator documentation, including the dataset profiles, the preregistered matrices and the evidence-bundle path, is in docs/OPERATIONS.md. Reproduce the swarm video Recording and rendering are separate by construction: the run records whole-scene qpos once per coupling interval and no frames at all, and the replay path refuses a world that has ever been stepped, so rendering cannot advance a simulation. PYTHONPATH=src python scripts/run_swarm3d_showcase.py \ --duration-s 30 --variant exact --variant stimulus-absent --progress PYTHONPATH=src python scripts/render_swarm3d_showcase.py \ --run RUN_DIR/exact --control RUN_DIR/stimulus-absent \ --out artifacts/showcase/swarm3d-v1/swarm3d-showcase.mp4 --progress Cost, measured: twelve flies at 0.015x biological real time, about 33 minutes of wall clock per variant on one RTX 3060 at 4.2 GB of device memory. That is not the GPU being slow — 30 s of biology at the registered 100 µs neural step is 300,000 timesteps over 165,122 x 12 neuron