artificial-life op• 資料種類:image → image
• 呼叫:fullseye.apply(img, "alife_sandpile", a=0.5, b=0.5)(2-D 的模型是一張圖 + 兩個純量旋鈕 a,b∈[0,1])

*圖為在 128×128 合成輸入上實際執行的輸出。左為輸入,右為輸出(非影像的回傳值直接顯示其值)。*
阿貝爾沙堆模型(Abelian sandpile)/自組織臨界性(self-organised criticality,Bak-Tang-Wiesenfeld 1987)。
> 以下的詳細說明為原文 —— 摘要與標題已翻譯。
The image is quantised to integer grain heights h = round(K * image) with
K = 4 + int(12a), then relaxed by the BTW toppling rule: a cell with 4 or
more grains gives one grain to each orthogonal neighbour and keeps the rest.
The boundary is dissipative (grains leaving the grid are lost), which is what
lets the pile settle into the self-organised critical state where every cell
holds at most 3 grains.
`a` sets the initial grain scale K (how supercritical the pile starts);
`b` sets the number of parallel relaxation sweeps 1 + int(50b). For
b >= 0.9 the pile relaxes *toward* stability with early termination, but the
sweep count is bounded by a total-work budget (`_SANDPILE_BUDGET` grain-
updates) so the op stays fast on any image size: a small or varied pile
reaches the stable critical state (every cell <= 3), while a very large
maximally-supercritical pile is only partially relaxed (full BTW
stabilisation is O(L^2) sweeps). Returns h / max(h) in [0, 1]; a fully
relaxed pile (max 3) takes values in {0, 1/3, 2/3, 1} (or {0, 1/2, 1} when
the stable maximum is 2).
• gallery2d_physics_alife_3d 族使用指南
• 範例資料目錄(下載 URL / 授權) —— 2-D 用 skimage.data(BSD/公有領域)加合成圖,3-D 給出真實資料源(Stanford/PDS 等)的下載 URL。
• 運算子來歷與參考文獻 —— 該運算子族所依據的研究/方法出處。
• 演算法的正典(作者・年份)與用途見上面的族使用指南。
下面的程式已確認可以執行(與圖相同的輸入)。在 Studio 說明中,此區塊會變成按鈕,可當場載入並執行。
alife_sandpile 0.50 0.50
▸ Load this pipeline · Load & run
• gallery2d_physics_alife_3d — py -3.11 examples/gallery2d_physics_alife_3d.py
image 作為輸入)identity · gaussian · mean_box · bilateral · unsharp · median · min_filter · max_filter
artificial-life)alife_gray_scott · alife_turing · alife_life_step · alife_cyclic_ca · alife_perona_malik · alife_curvature_flow · alife_dla · alife_reaction_bz
*Provenance: ops.py — 2D 運算子登記表。本條目由 tools/opdocs.py md 自動產生(請勿手動編輯)。*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.