Skorch emulator (#89)

This PR is a complete rewrite of the ExperimentalEmulator. It makes use of skorch to enable use of scikit-learn. This commit also removes some dead code and does some minor refactoring.

.gitignore

@@ -1,3 +1,4 @@
+
 # Byte-compiled / optimized / DLL files
 __pycache__/
 *.py[cod]
@@ -111,6 +112,7 @@ venv.bak/
 ~*.xlsx
 *.DS_STORE
 .vscode/
+.idea/
 
 # TSEMO
 .TSEMO_DATA

README.md

@@ -10,6 +10,7 @@
 
 Summit is a set of tools for optimising chemical processes. We’ve started by targeting reactions.
 
+
 ## What is Summit?
 Currently, reaction optimisation in the fine chemicals industry is done by intuition or design of experiments.  Both scale poorly with the complexity of the problem. 
 

docs/source/experiments_benchmarks/new_benchmarks.ipynb

@@ -60,7 +60,10 @@
    "source": [
     "from summit.benchmarks import ExperimentalEmulator\n",
     "from summit.domain import *\n",
-    "from summit.utils.dataset import DataSet"
+    "from summit.utils.dataset import DataSet\n",
+    "import pkg_resources\n",
+    "import pathlib\n",
+    "import pprint"
    ]
   },
   {
@@ -97,7 +100,7 @@
        "<table id='domain' width=100%><tr><td><b>Name</b></td><td><b>Type</b></td><td><b>Description</b></td><td><b>Values</b></td></tr><tr><td>catalyst</td><td>categorical, input</td><td>Catalyst type - different ligands</td><td>8 levels</td></tr><tr><td>t_res</td><td>continuous, input</td><td>Residence time in seconds (s)</td><td>[60,600]</td></tr><tr><td>temperature</td><td>continuous, input</td><td>Reactor temperature in degrees Celsius (ºC)</td><td>[30,110]</td></tr><tr><td>catalyst_loading</td><td>continuous, input</td><td>Catalyst loading in mol%</td><td>[0.5,2.5]</td></tr><tr><td>ton</td><td>continuous, maximize objective</td><td>Turnover number - moles product generated divided by moles catalyst used</td><td>[0,200]</td></tr><tr><td>yield</td><td>continuous, maximize objective</td><td>Yield</td><td>[0,100]</td></tr></table>"
       ],
       "text/plain": [
-       "<summit.domain.Domain at 0x153eca5f8>"
+       "<summit.domain.Domain at 0x13483f8d0>"
       ]
      },
      "execution_count": 3,
@@ -173,7 +176,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "If you are running this yourself, uncomment the second line."
+    "Here, we import the data that we already have in the Summit package, but you could use your own data. Change verbose to 1 if you want streaming updates of the training."
    ]
   },
   {
@@ -182,19 +185,10 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "import pathlib \n",
-    "FOLDER = pathlib.Path(\"../_static/\")  # When using this in the context of docs\n",
-    "# FOLDER = pathlib.Path(\".\")"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "emul = ExperimentalEmulator(domain=domain, model_name='my_reizman')\n",
-    "emul.train(csv_dataset=FOLDER / \"reizman_suzuki_case1_train_test.csv\", cv_fold=2, test_size=0.25)"
+    "DATA_PATH = pathlib.Path(pkg_resources.resource_filename(\"summit\", \"benchmarks/data\"))\n",
+    "ds = DataSet.read_csv(DATA_PATH / \"reizman_suzuki_case_1.csv\",)\n",
+    "emul = ExperimentalEmulator(model_name='my_reizman', domain=domain, dataset=ds)\n",
+    "res = emul.train(max_epochs=100, cv_fold=2, test_size=0.25, verbose=0)"
    ]
   },
   {
@@ -206,7 +200,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
+   "execution_count": 5,
    "metadata": {},
    "outputs": [
     {
@@ -215,10 +209,10 @@
        "<table id='domain' width=100%><tr><td><b>Name</b></td><td><b>Type</b></td><td><b>Description</b></td><td><b>Values</b></td></tr><tr><td>catalyst</td><td>categorical, input</td><td>Catalyst type - different ligands</td><td>8 levels</td></tr><tr><td>t_res</td><td>continuous, input</td><td>Residence time in seconds (s)</td><td>[60,600]</td></tr><tr><td>temperature</td><td>continuous, input</td><td>Reactor temperature in degrees Celsius (ºC)</td><td>[30,110]</td></tr><tr><td>catalyst_loading</td><td>continuous, input</td><td>Catalyst loading in mol%</td><td>[0.5,2.5]</td></tr><tr><td>ton</td><td>continuous, maximize objective</td><td>Turnover number - moles product generated divided by moles catalyst used</td><td>[0,200]</td></tr><tr><td>yield</td><td>continuous, maximize objective</td><td>Yield</td><td>[0,100]</td></tr></table>"
       ],
       "text/plain": [
-       "<summit.domain.Domain at 0x153eca5f8>"
+       "<summit.domain.Domain at 0x13483f8d0>"
       ]
      },
-     "execution_count": 7,
+     "execution_count": 5,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -229,7 +223,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 8,
+   "execution_count": 6,
    "metadata": {},
    "outputs": [
     {
@@ -271,27 +265,27 @@
        "      <td>60</td>\n",
        "      <td>100</td>\n",
        "      <td>1.0</td>\n",
-       "      <td>29.972519</td>\n",
-       "      <td>43.924999</td>\n",
+       "      <td>23.364954</td>\n",
+       "      <td>33.13002</td>\n",
        "      <td>0.0</td>\n",
-       "      <td>0.063283</td>\n",
+       "      <td>0.058378</td>\n",
        "      <td>NaN</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
        "</div>"
       ],
       "text/plain": [
-       "NAME catalyst t_res temperature catalyst_loading        ton      yield  \\\n",
-       "TYPE     DATA  DATA        DATA             DATA       DATA       DATA   \n",
-       "0       P1-L1    60         100              1.0  29.972519  43.924999   \n",
+       "NAME catalyst t_res temperature catalyst_loading        ton     yield  \\\n",
+       "TYPE     DATA  DATA        DATA             DATA       DATA      DATA   \n",
+       "0       P1-L1    60         100              1.0  23.364954  33.13002   \n",
        "\n",
        "NAME computation_t experiment_t strategy  \n",
        "TYPE      METADATA     METADATA METADATA  \n",
-       "0              0.0     0.063283      NaN  "
+       "0              0.0     0.058378      NaN  "
       ]
      },
-     "execution_count": 8,
+     "execution_count": 6,
      "metadata": {},
      "output_type": "execute_result"
     }