Influence cells · v17

This is a cellular automaton: a grid of small cells that change according to rules. The colored cells alter the influence passing through them, and that influence affects which cells survive or appear.

Try this: watch the pattern, select Pause, then use Step once to see one update. Choose a cell type below the picture and draw on the grid. Select Resume to see the result.

In v17, the sources at the top use four regular rhythms. “New random world” changes their positions and numbers.

What do the colors and controls mean?

Black: reduces influence. Blue and green: turn it left or right. Orange: increases it. Purple: sends some upward while using a stored allowance.

Entity feedback: lets cells change the influence around them. Field: shows the influence map. Mutation: lets local conditions change the type of a new cell. Active: the number of occupied cells. The other measurements and settings are for closer study.

Proof of concept: this is an abstract model of feedback and pattern formation. Its influence values are not measurements of water or physical energy.

Sourcecol 0

Strength 0.45

Period 18

Noise level 0.00

Noise level controls enabled stochastic channels. Stable default: 0.00.

Hover a top source node to adjust it. Use Stable sources to reset all sources to the same deterministic forcing.

Grid Speed 1.0×

Source scale: Stochastic forcing: All four source scales coexist. Source count falls geometrically as strength and period rise and noise falls. Seeded stochastic forcing is the default; Entropy RNG remains optional.

Draw: Brush Drag on the grid to paint. Drawing pauses the simulation. The top source row and bottom outlet row are protected.

Optional experiment notes

Mutation is OFF. GIF history starts after a new world or your latest drawing intervention.

Step0

Active entities0

Birth / death0 / 0

Clustering0.000

Recent activity0%

Type diversity0.000

Estimated macro periodnone

Period score0.000

Forcing base period0

Seam imbalance0.000

Reflectors0

Reflector energy0.000

Advanced model details and settings

Stochastic forcing

Three independent source perturbations can be enabled without changing the deterministic default:

Strength: step-to-step multiplicative amplitude noise.
Period: step-to-step effective-period jitter around the selected period.
Timing: step-to-step phase/timing jitter around the periodic waveform.

Strength: A*=A(1+rξ)
Period: T*=T(1+0.35rξ)
Timing: φ*=φ+2π(0.25rξ)

Entropy RNG uses browser/OS cryptographic entropy and is intentionally non-reproducible. This is not a claim of quantum or physically certified true randomness. With Entropy RNG OFF, the same seed reproduces the same pseudo-random forcing.

Boundary conditions

Top: periodic source boundary.
Bottom: open outlet.
Left and right: toroidal wrap.

The new ingredient is a rare Reflector entity. A Reflector can send part of its incoming influence one row upward, but every reflection consumes stored energy. When energy is low, reflection weakens and the Reflector becomes easier to lose.

Top = source
Bottom = open outlet
Left ↔ Right = torus

Entropy-randomized source realization

New random world redraws both source locations and source-type counts using browser/OS cryptographic entropy when available.

E[Nℓ] ∝ 2^-ℓ

The realized counts vary around the scaling-law expectation. Strength, period, and noise still follow the source scale. After the counts are drawn, strengths are renormalized so the total nominal source budget remains comparable across worlds.

Replay same world reuses the exact source realization, including locations, counts, scale classes, strengths, periods, and noise levels.

This is entropy-based randomness, not a claim of certified physical or quantum randomness.

Mixed multiscale forcing

All source scales coexist in one world. For hierarchy level ℓ = 0,1,2,3:

Nℓ ≈ N₀ / 2^ℓ
Aℓ ∝ 2^ℓ
Tℓ = 9 · 2^ℓ
rℓ = 0.32 / 2^ℓ

After integer source counts are chosen, all amplitudes receive one common normalization factor so the hierarchy has the same nominal summed source strength as the stable reference.

F - Fine: common, weak, fast, noisy.
M - Meso: fewer, stronger, slower.
C - Coarse: rare, strong determinants.
L - Macro: very few, strongest, slowest, lowest-noise determinants.

The source waveform is mean-centered, so the periodic cycle does not shut the whole world off:

A(t) = a [1 + 0.40 sin(2πt/T)] × stochastic terms

Entity lexicon

Absorberg=.62, 90°

Left deflectorg=.90, 58°

Right deflectorg=.90, 122°

Amplifierg=1.12, 90°

Reflectorg=.82, ρ↑=.38

Reflector only:
A↑ = ρ A · energy
energy(t+1) = energy + recharge - maintenance - reflection cost

Reflectors start rare and are rarely recruited. Their upward feedback is therefore a costly ecological strategy rather than a universal property of every entity.

Field rule

Main packets propagate downward.
Reflective packets are created only at Reflector cells with available energy.
Reflected packets move one row up and are then re-propagated downward in a few echo passes.
This creates local memory and recirculation.

E = 1 - exp(-1.15 A)

User-designed initial conditions

You can now construct your own entity configuration directly on the grid. Draw while paused, or start drawing during a run and the simulation will pause automatically.

Paint any of the five v7 entity types.
Reflectors drawn by the user receive the standard reflector birth energy.
Eraser removes an entity and its stored energy.
Brush sizes of 1, 3×3, and 5×5 cells are available.
The source row and outlet row cannot be painted.

This changes only the initial or intervened configuration. The v7 survival, birth, feedback, and energy rules are unchanged.

Record manual regimes

Manually drawn configurations can generate outcomes that a random initial condition does not, including cases where Deflectors persist. Use the recorder to preserve those interventions exactly.

Starting config JSON: saves every entity cell, viability, Reflector energy, source parameters, rule parameters, current step, and your notes.
Stable-period GIF: exports the latest detected cycle from a rolling frame history.
If automatic period detection is not satisfactory, type a period in simulation steps before saving the GIF.

The GIF uses a built-in encoder, so no external library or internet connection is required.

Stable-state reproduction mode

Mutation is OFF by default. In this mode, a newborn inherits one of the entity types already present in its local neighborhood. Environmental direction and exposure can affect whether a birth occurs, but they cannot create a new strategy.

Mutation OFF:
new type ∈ neighboring parent types

Turn Mutation ON only when you want the earlier evolutionary rule, where exposure/direction can generate Amplifiers and rare spontaneous Reflectors.

Cellular fate rule

Survival: a slightly narrower exposure band and 1 to 4 neighbors.
Death: viability reaches 0 after repeated unfavorable ticks; Reflectors also suffer when energy is depleted.
Birth: empty cell, suitable exposure, and 2 to 4 neighbors.