31:[] 2f:{"title":"reaction diffusion","cells":[{"id":0,"value":"md`# reaction diffusion`","pinned":false,"hidden":false,"mode":"ojs"},{"id":117,"value":"viewof D_a_s = html``","pinned":true,"hidden":false,"mode":"ojs"},{"id":119,"value":"viewof D_b_s = html``","pinned":true,"hidden":false,"mode":"ojs"},{"id":125,"value":"viewof feed = html``","pinned":true,"hidden":false,"mode":"ojs"},{"id":127,"value":"viewof kill = html``","pinned":true,"hidden":false,"mode":"ojs"},{"id":106,"value":"viewof itersPerFrame = html``","pinned":true,"hidden":false,"mode":"ojs"},{"id":42,"value":"canvas = html`

`","pinned":false,"hidden":false,"mode":"ojs"},{"id":172,"value":"md`## good params \n\n D_a_s = 0.72, D_b_s = 0.37, feed = 0.022, kill = 0.052\n`","pinned":false,"hidden":false,"mode":"ojs"},{"id":155,"value":"t = tf.tidy.bind(tf)","pinned":false,"hidden":false,"mode":"ojs"},{"id":4,"value":"tf = require('@tensorflow/tfjs@0.14')","pinned":false,"hidden":false,"mode":"ojs"},{"id":15,"value":"function reactionDiffusionIter(a, b) {\n return t(() => {\n const D_a = tf.scalar(D_a_s), D_b = tf.scalar(D_b_s), f = tf.scalar(feed), k = tf.scalar(kill)\n const dt = tf.scalar(1.0)\n const kf = tf.scalar(kill + feed)\n const gradA = tf.conv2d(a, lapKern, [1,1], 'same')\n const gradB = tf.conv2d(b, lapKern, [1,1], 'same')\n \n const ab2 = a.mul(b.square())\n const da = D_a.mul(gradA).sub(ab2).add(f.mul(tf.scalar(1.0).sub(a))).mul(dt)\n const db = D_b.mul(gradB).add(ab2).sub(kf.mul(b)).mul(dt)\n return [a.add(da), b.add(db)]\n })\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":8,"value":"lapKern = tf.tensor2d([[0.05, 0.20, 0.05],\n [0.20, -1.0, 0.20],\n [0.05, 0.20, 0.05]])\n.expandDims(2)\n.expandDims(3)","pinned":false,"hidden":false,"mode":"ojs"},{"id":18,"value":"{\n const s = 512\n let a = tf.ones([1, s, s, 1], 'float32').add(tf.randomNormal([1, s, s, 1], 0.0, 0.0))\n a = tf.clipByValue(a, 0.0, 1.0)\n let bBuf = tf.buffer([1, s, s, 1], 'float32')\n const d = 5\n for(let i = (s/2) - d; i < (s/2) + d; i++) {\n for(let j = (s/2) - d; j < (s/2) + d; j++) {\n bBuf.values[i * s + j] = 1.0\n }\n }\n let b = tf.clipByValue(bBuf.toTensor().add(tf.randomNormal([1, s, s, 1], 0.0, 0.0)), 0.0, 1.0)\n invalidation.then(() => [a, b].map(t => t.dispose()))\n for(let i = 0; i < 10000; i++) {\n for(let j = 0; j < itersPerFrame; j++) {\n let [a_prime, b_prime] = t(() => reactionDiffusionIter(a, b))\n a.dispose()\n b.dispose()\n a = a_prime\n b = b_prime\n }\n const vis = t(() => tf.clipByValue(a.sub(b).squeeze(), 0.0, 1.0))\n invalidation.then(() => vis.dispose())\n await tf.toPixels(vis, canvas)\n vis.dispose()\n yield md`frame num: ${i}`\n await tf.nextFrame()\n }\n\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":84,"value":"{ while(true) { yield tf.memory() } }","pinned":false,"hidden":false,"mode":"ojs"}],"files":"$W31","stdlib":"1","titling":"auto"}