Increase the data for stick-probability substantially, do some more maths, add data to gitignore for now

.gitignore

@@ -1 +1,2 @@
 .idea
+/data/

notebooks/minimal-viable-alteration.ipynb

@@ -337,12 +337,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 121,
+   "execution_count": 126,
    "outputs": [],
    "source": [
     "def compute_fd(df, tol = 0.05):\n",
     "    tail_index = convergent_tail_index(df.fd, tol)\n",
-    "    return np.mean(df.fd[tail_index:])"
+    "    return np.mean(df.fd[tail_index:]), np.std(df.fd[tail_index:])"
    ],
    "metadata": {
     "collapsed": false
@@ -350,21 +350,11 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 122,
-   "outputs": [
-    {
-     "data": {
-      "text/plain": "[1.6972188780546005,\n 1.741736844787361,\n 1.732135584521786,\n 1.7152613146304656,\n 1.728836880434697,\n 1.7300514834031597,\n 1.748421194691431,\n 1.7287640430180986,\n 1.7123919318562837,\n 1.7739930888022366,\n 1.7584717697837762,\n 1.7009044749656368,\n 1.7316986067539795,\n 1.7397504422026655,\n 1.714607675170586,\n 1.7541336557600284,\n 1.6944449486837003,\n 1.7413017386225582,\n 1.7586824633086955,\n 1.7621333897577298,\n 1.731409604427308,\n 1.7040785354660994,\n 1.720970435833901,\n 1.7493817381438628,\n 1.7194584712458039,\n 1.7371705652881426,\n 1.7476735580577902,\n 1.7453410966666576,\n 1.7405812458466108,\n 1.7318723528446751,\n 1.7363887686944106,\n 1.7082895630235775,\n 1.7398588474316838,\n 1.7071844051847547,\n 1.730417361421785,\n 1.7562821657346117,\n 1.6970380009288513,\n 1.727847271689809,\n 1.7344987347694254,\n 1.6917005260543316,\n 1.736465228691966,\n 1.7463067343476375,\n 1.7327131263137254,\n 1.7320567534823674,\n 1.7119737361873444,\n 1.7157273693563924,\n 1.7290841054382464,\n 1.7647982477471926,\n 1.7562017718859915,\n 1.6950468500096474,\n 1.747725564486872,\n 1.7252896867263523,\n 1.7671445124216143,\n 1.7693744566230807,\n 1.7059184138844365,\n 1.7090900521487329,\n 1.7231096018794785,\n 1.7209694721543376,\n 1.7678252294196048,\n 1.7316721866238294,\n 1.7262902265337214,\n 1.7377991339529295,\n 1.770592027972686,\n 1.6973207132946366,\n 1.7180222330856505,\n 1.7350278763310734,\n 1.7371312165639443,\n 1.7199230950584823,\n 1.7102785432584966,\n 1.7244790713772096,\n 1.7351634054665168,\n 1.713486860504405,\n 1.7568987349317238,\n 1.7136332033996875,\n 1.7343403134354431,\n 1.7238926108882633,\n 1.7090981817183717,\n 1.703302194004836,\n 1.7446644371287279,\n 1.7379078424212124,\n 1.7644224392368066,\n 1.7411749383649393,\n 1.7387938040773043,\n 1.7587053030420463,\n 1.7723295858624455,\n 1.7207158077979385,\n 1.712275781859355,\n 1.684724599409851,\n 1.7184172924395567,\n 1.7291500326282283,\n 1.764249287056465,\n 1.6995609808553767,\n 1.7141798221829627,\n 1.7337439700388064,\n 1.7367204835781522,\n 1.7532684630386097,\n 1.722663761530065,\n 1.7281346013788716]"
-     },
-     "execution_count": 122,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
+   "execution_count": 131,
+   "outputs": [],
    "source": [
     "fds = [compute_fd(df, 0.001) for df in dfs]\n",
-    "fds_clean = [f for f in fds if f < np.inf]\n",
-    "fds_clean"
+    "fds_clean = [f for f in fds if f[0] < np.inf]"
    ],
    "metadata": {
     "collapsed": false
@@ -372,19 +362,30 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 124,
+   "execution_count": 137,
+   "outputs": [],
+   "source": [
+    "means, stds = list(zip(*fds_clean))"
+   ],
+   "metadata": {
+    "collapsed": false
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 140,
    "outputs": [
     {
      "data": {
       "text/plain": "(1.7311567309540636, 0.020648406595153657)"
      },
-     "execution_count": 124,
+     "execution_count": 140,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "np.mean(fds_clean), np.std(fds_clean)"
+    "np.mean(means), np.std(means)"
    ],
    "metadata": {
     "collapsed": false

tools/stick-probability-harness.bash

@@ -4,7 +4,7 @@
 
 MODEL="models/stick-probability"
 mkdir -p data/stick-probability
-PROBABILITIES=$(seq 0.1 0.1 1)
+PROBABILITIES=$(seq 0.05 0.1 1)
 
 for i in $PROBABILITIES; do
   mkdir data/stick-probability/"$i"
@@ -12,4 +12,4 @@ done
 
 MAX_PARTICLES=2000
 
-parallel "$MODEL" '{1}' $MAX_PARTICLES '{2}' 'data/stick-probability/{2}/{1}.csv' ::: {1..20} ::: $PROBABILITIES
+parallel "$MODEL" '{1}' $MAX_PARTICLES '{2}' 'data/stick-probability/{2}/{1}.csv' ::: {1..100} ::: $PROBABILITIES