Five simulations, no faked runs

Training simulations, run before anyone flies. Scores stay on this device.

Every simulation here has a pure core — a plain function that takes a state and an input and returns the next state — sitting under unit tests that replay real ticks. The code the tests exercise is the code you are about to play. Nothing pauses on a hidden timer, nothing is scripted, and no run is pre-recorded.

Simulations: 5 loaded
Pure cores: 5 under test
Scores: Local only
Network: No upload

DinoMind Arena
Re-Entry Corridor
Go / No-Go Drill
Neural Tic-Tac-Toe
Typing Challenge

SIM-01

DinoMind Arena

Race a visualized AI runner over the identical seeded course.

Space or ↑ to jump · ↓ to duck · P to pause

The opponent is a hand-tuned linear policy over NEAT-style sensors, written for this page — a demonstration policy, not the network trained in the original DinoMindEvolution project.

Press Start, Space, or tap to begin

AI inputs → actions

distance: 1.00
obstacle height: 0.00
flying: 0.00
speed: 0.00
altitude: 0.00

JUMP —DUCK —

What is neuroevolution?

In the original DinoMind Evolution project, NEAT (NeuroEvolution of Augmenting Topologies) evolves a population of neural networks: each dino senses the next obstacle's distance, size, and its own speed, and networks that survive longer pass their weights — with mutations — to the next generation. No gradients, no labels: selection pressure alone shapes the policy. The panel on the right shows the same sensor → activation → action loop this browser demo uses.

Loading DinoMind Arena…

SIM-02

Re-Entry Corridor

Hold the descent inside the corridor on a fixed propellant budget.

↑ retro burn · ← → adjust attitude · P to pause

Deterministic two-degree-of-freedom flight model on a fixed timestep, run at 7x time acceleration so a descent takes a minute rather than ten. Heat load is defined by the corridor itself, not by simulated thermodynamics, and the numbers are tuned for playability rather than for any specific vehicle.

Altitude: 45.0 km
Speed: 1402 m/s
Sink rate: -80.0 m/s
Range to pad: 105.0 km
Pitch: 12°
Propellant: 100 %
Heat load: 0 %
Corridor: Inside

Stay inside the green corridor: heat builds whenever you are above its ceiling. Pitch up to stretch the glide toward the pad, then save propellant for the last kilometre.

Loading Re-Entry Corridor…

SIM-03

Go / No-Go Drill

Read four gauges and call the burn before the clock runs out.

G to call GO · N to call NO-GO · or use the buttons

Limits are stated on the panel before each call, so the drill tests reading speed, never guesswork.

Call GO only when every reading is inside its stated limits

SCORE 0CALLS 0STRIKES 0/3

Four channels come up at once, each with its own limits printed beside it. Read them and make the call before the clock runs out. Three missed calls ends the drill.

Loading Go / No-Go Drill…

SIM-04

Neural Tic-Tac-Toe

Exact alpha-beta on 3×3, depth-limited search on 3×3×3.

Click or Tab + Enter · every square is a button

On the 3×3 board the search is exhaustive, so perfect play is a draw. The 3×3×3 board is depth-limited with a line heuristic; node counts are shown per move.

Difficulty

Your move (X)

Loading Neural Tic-Tac-Toe…

SIM-05

Typing Challenge

Machine-learning terms and real code, scored on live WPM and accuracy.

Just type · Esc to reset

Time left: 30s
Words per minute: —
Accuracy: 100%

The timer starts on your first keystroke. Mistakes are marked and feed the heat map.

Mistake heat map

qwertyuiop

asdfghjkl

zxcvbnm

Loading Typing Challenge…