Run black on tutorials

pytorch · Feb 28, 2023 · 30fb0be · 30fb0be
1 parent e5f4d14
commit 30fb0be
Show file tree

Hide file tree

Showing 32 changed files with 19,084 additions and 18,689 deletions.
diff --git a/tutorials/GIBBON_for_efficient_batch_entropy_search.ipynb b/tutorials/GIBBON_for_efficient_batch_entropy_search.ipynb
diff --git a/tutorials/Multi_objective_multi_fidelity_BO.ipynb b/tutorials/Multi_objective_multi_fidelity_BO.ipynb
diff --git a/tutorials/batch_mode_cross_validation.ipynb b/tutorials/batch_mode_cross_validation.ipynb
diff --git a/tutorials/bo_with_warped_gp.ipynb b/tutorials/bo_with_warped_gp.ipynb
@@ -84,20 +84,20 @@
     "from torch.distributions import Kumaraswamy\n",
     "import matplotlib.pyplot as plt\n",
     "\n",
-    "%matplotlib inline \n",
+    "%matplotlib inline\n",
     "\n",
     "\n",
     "fontdict = {\"fontsize\": 15}\n",
     "torch.manual_seed(1234567890)\n",
     "c1 = torch.rand(6, dtype=dtype, device=device) * 3 + 0.1\n",
     "c0 = torch.rand(6, dtype=dtype, device=device) * 3 + 0.1\n",
-    "x = torch.linspace(0,1,101, dtype=dtype, device=device)\n",
+    "x = torch.linspace(0, 1, 101, dtype=dtype, device=device)\n",
     "k = Kumaraswamy(concentration1=c1, concentration0=c0)\n",
     "k_icdfs = k.icdf(x.unsqueeze(1).expand(101, 6))\n",
-    "fig, ax = plt.subplots(1,1, figsize=(5,5))\n",
+    "fig, ax = plt.subplots(1, 1, figsize=(5, 5))\n",
     "\n",
     "for i in range(6):\n",
-    "    ax.plot(x.cpu(), k_icdfs[:,i].cpu())\n",
+    "    ax.plot(x.cpu(), k_icdfs[:, i].cpu())\n",
     "ax.set_xlabel(\"Raw Value\", **fontdict)\n",
     "ax.set_ylabel(\"Transformed Value\", **fontdict)"
    ]
@@ -112,10 +112,10 @@
     "\n",
     "neg_hartmann6 = Hartmann(negate=True)\n",
     "\n",
+    "\n",
     "def obj(X):\n",
     "    X_warp = k.icdf(X)\n",
-    "    return neg_hartmann6(X_warp)\n",
-    "    "
+    "    return neg_hartmann6(X_warp)"
    ]
   },
   {
@@ -147,11 +147,13 @@
     "\n",
     "def generate_initial_data(n=14):\n",
     "    # generate training data\n",
-    "    train_x = draw_sobol_samples(bounds=bounds, n=n, q=1, seed=torch.randint(0,10000,(1,)).item()).squeeze(1)\n",
+    "    train_x = draw_sobol_samples(\n",
+    "        bounds=bounds, n=n, q=1, seed=torch.randint(0, 10000, (1,)).item()\n",
+    "    ).squeeze(1)\n",
     "    exact_obj = obj(train_x).unsqueeze(-1)  # add output dimension\n",
-    "    \n",
+    "\n",
     "    best_observed_value = exact_obj.max().item()\n",
-    "    train_obj = exact_obj +  NOISE_SE * torch.randn_like(exact_obj)\n",
+    "    train_obj = exact_obj + NOISE_SE * torch.randn_like(exact_obj)\n",
     "    return train_x, train_obj, best_observed_value"
    ]
   },
@@ -181,17 +183,17 @@
     "            indices=list(range(train_x.shape[-1])),\n",
     "            # use a prior with median at 1.\n",
     "            # when a=1 and b=1, the Kumaraswamy CDF is the identity function\n",
-    "            concentration1_prior=LogNormalPrior(0.0, 0.75 ** 0.5),\n",
-    "            concentration0_prior=LogNormalPrior(0.0, 0.75 ** 0.5),\n",
+    "            concentration1_prior=LogNormalPrior(0.0, 0.75**0.5),\n",
+    "            concentration0_prior=LogNormalPrior(0.0, 0.75**0.5),\n",
     "        )\n",
     "    else:\n",
     "        warp_tf = None\n",
-    "    # define the model for objective \n",
+    "    # define the model for objective\n",
     "    model = FixedNoiseGP(\n",
     "        train_x,\n",
     "        standardize(train_obj),\n",
     "        train_yvar.expand_as(train_obj),\n",
-    "        input_transform=warp_tf\n",
+    "        input_transform=warp_tf,\n",
     "    ).to(train_x)\n",
     "    mll = ExactMarginalLogLikelihood(model.likelihood, model)\n",
     "    return mll, model"
@@ -229,7 +231,7 @@
     "        raw_samples=raw_samples,  # used for intialization heuristic\n",
     "        options={\"batch_limit\": 5, \"maxiter\": 200},\n",
     "    )\n",
-    "    # observe new values \n",
+    "    # observe new values\n",
     "    new_x = candidates.detach()\n",
     "    exact_obj = obj(new_x).unsqueeze(-1)  # add output dimension\n",
     "    train_obj = exact_obj + NOISE_SE * torch.randn_like(exact_obj)\n",
@@ -242,7 +244,7 @@
     "    \"\"\"\n",
     "    rand_x = draw_sobol_samples(bounds=bounds, n=1, q=1).squeeze(1)\n",
     "    next_random_best = obj(rand_x).max().item()\n",
-    "    best_random.append(max(best_random[-1], next_random_best))       \n",
+    "    best_random.append(max(best_random[-1], next_random_best))\n",
     "    return best_random"
    ]
   },
@@ -271,8 +273,8 @@
     "import warnings\n",
     "\n",
     "\n",
-    "warnings.filterwarnings('ignore', category=BadInitialCandidatesWarning)\n",
-    "warnings.filterwarnings('ignore', category=RuntimeWarning)\n",
+    "warnings.filterwarnings(\"ignore\", category=BadInitialCandidatesWarning)\n",
+    "warnings.filterwarnings(\"ignore\", category=RuntimeWarning)\n",
     "\n",
     "\n",
     "N_TRIALS = 3 if not SMOKE_TEST else 2\n",
@@ -287,46 +289,53 @@
     "\n",
     "# average over multiple trials\n",
     "for trial in range(1, N_TRIALS + 1):\n",
-    "    \n",
+    "\n",
     "    print(f\"\\nTrial {trial:>2} of {N_TRIALS} \", end=\"\")\n",
     "    best_observed_ei, best_observed_warp, best_random = [], [], []\n",
-    "    \n",
+    "\n",
     "    # call helper functions to generate initial training data and initialize model\n",
     "    train_x_ei, train_obj_ei, best_observed_value_ei = generate_initial_data(n=14)\n",
-    "    mll_ei, model_ei = initialize_model(train_x_ei, train_obj_ei, use_input_warping=False)\n",
-    "    \n",
-    "    train_x_warp, train_obj_warp,  = train_x_ei, train_obj_ei\n",
+    "    mll_ei, model_ei = initialize_model(\n",
+    "        train_x_ei, train_obj_ei, use_input_warping=False\n",
+    "    )\n",
+    "\n",
+    "    train_x_warp, train_obj_warp, = (\n",
+    "        train_x_ei,\n",
+    "        train_obj_ei,\n",
+    "    )\n",
     "    best_observed_value_warp = best_observed_value_ei\n",
     "    # use input warping\n",
-    "    mll_warp, model_warp = initialize_model(train_x_warp, train_obj_warp, use_input_warping=True)\n",
-    "    \n",
+    "    mll_warp, model_warp = initialize_model(\n",
+    "        train_x_warp, train_obj_warp, use_input_warping=True\n",
+    "    )\n",
+    "\n",
     "    best_observed_ei.append(best_observed_value_ei)\n",
     "    best_observed_warp.append(best_observed_value_warp)\n",
     "    best_random.append(best_observed_value_ei)\n",
-    "    \n",
+    "\n",
     "    # run N_BATCH rounds of BayesOpt after the initial random batch\n",
-    "    for iteration in range(1, N_BATCH + 1):    \n",
-    "        \n",
+    "    for iteration in range(1, N_BATCH + 1):\n",
+    "\n",
     "        t0 = time.monotonic()\n",
-    "        \n",
+    "\n",
     "        # fit the models\n",
     "        fit_gpytorch_mll(mll_ei)\n",
     "        fit_gpytorch_mll(mll_warp)\n",
-    "        \n",
+    "\n",
     "        ei = qNoisyExpectedImprovement(\n",
-    "            model=model_ei, \n",
+    "            model=model_ei,\n",
     "            X_baseline=train_x_ei,\n",
     "        )\n",
-    "        \n",
+    "\n",
     "        ei_warp = qNoisyExpectedImprovement(\n",
-    "            model=model_warp, \n",
+    "            model=model_warp,\n",
     "            X_baseline=train_x_warp,\n",
     "        )\n",
-    "        \n",
+    "\n",
     "        # optimize and get new observation\n",
     "        new_x_ei, new_obj_ei = optimize_acqf_and_get_observation(ei)\n",
     "        new_x_warp, new_obj_warp = optimize_acqf_and_get_observation(ei_warp)\n",
-    "                \n",
+    "\n",
     "        # update training points\n",
     "        train_x_ei = torch.cat([train_x_ei, new_x_ei])\n",
     "        train_obj_ei = torch.cat([train_obj_ei, new_obj_ei])\n",
@@ -341,20 +350,25 @@
     "        best_observed_ei.append(best_value_ei)\n",
     "        best_observed_warp.append(best_value_warp)\n",
     "\n",
-    "        mll_ei, model_ei = initialize_model(train_x_ei, train_obj_ei, use_input_warping=False)\n",
-    "        mll_warp, model_warp = initialize_model(train_x_warp, train_obj_warp, use_input_warping=True)\n",
-    "        \n",
+    "        mll_ei, model_ei = initialize_model(\n",
+    "            train_x_ei, train_obj_ei, use_input_warping=False\n",
+    "        )\n",
+    "        mll_warp, model_warp = initialize_model(\n",
+    "            train_x_warp, train_obj_warp, use_input_warping=True\n",
+    "        )\n",
+    "\n",
     "        t1 = time.monotonic()\n",
-    "        \n",
+    "\n",
     "        if verbose:\n",
     "            print(\n",
     "                f\"\\nBatch {iteration:>2}: best_value (random, ei, ei_warp) = \"\n",
     "                f\"({max(best_random):>4.2f}, {best_value_ei:>4.2f}, {best_value_warp:>4.2f}), \"\n",
-    "                f\"time = {t1-t0:>4.2f}.\", end=\"\"\n",
+    "                f\"time = {t1-t0:>4.2f}.\",\n",
+    "                end=\"\",\n",
     "            )\n",
     "        else:\n",
     "            print(\".\", end=\"\")\n",
-    "   \n",
+    "\n",
     "    best_observed_all_ei.append(best_observed_ei)\n",
     "    best_observed_all_warp.append(best_observed_warp)\n",
     "    best_random_all.append(best_random)"
@@ -428,21 +442,38 @@
     "\n",
     "iters = np.arange(N_BATCH + 1)\n",
     "y_ei = np.log10(GLOBAL_MAXIMUM - np.asarray(best_observed_all_ei))\n",
-    "y_ei_warp =  np.log10(GLOBAL_MAXIMUM - np.asarray(best_observed_all_warp))\n",
-    "y_rnd =  np.log10(GLOBAL_MAXIMUM - np.asarray(best_random_all))\n",
+    "y_ei_warp = np.log10(GLOBAL_MAXIMUM - np.asarray(best_observed_all_warp))\n",
+    "y_rnd = np.log10(GLOBAL_MAXIMUM - np.asarray(best_random_all))\n",
     "\n",
     "fig, ax = plt.subplots(1, 1, figsize=(8, 6))\n",
     "ax.errorbar(\n",
-    "    iters, y_rnd.mean(axis=0), yerr=ci(y_rnd), label=\"Sobol\", linewidth=1.5, capsize=3, alpha=0.6\n",
+    "    iters,\n",
+    "    y_rnd.mean(axis=0),\n",
+    "    yerr=ci(y_rnd),\n",
+    "    label=\"Sobol\",\n",
+    "    linewidth=1.5,\n",
+    "    capsize=3,\n",
+    "    alpha=0.6,\n",
     ")\n",
     "ax.errorbar(\n",
-    "    iters, y_ei.mean(axis=0), yerr=ci(y_ei), label=\"NEI\", linewidth=1.5, capsize=3, alpha=0.6,\n",
+    "    iters,\n",
+    "    y_ei.mean(axis=0),\n",
+    "    yerr=ci(y_ei),\n",
+    "    label=\"NEI\",\n",
+    "    linewidth=1.5,\n",
+    "    capsize=3,\n",
+    "    alpha=0.6,\n",
     ")\n",
     "ax.errorbar(\n",
-    "    iters, y_ei_warp.mean(axis=0), yerr=ci(y_ei_warp), label=\"NEI + Input Warping\",\n",
-    "    linewidth=1.5, capsize=3, alpha=0.6,\n",
+    "    iters,\n",
+    "    y_ei_warp.mean(axis=0),\n",
+    "    yerr=ci(y_ei_warp),\n",
+    "    label=\"NEI + Input Warping\",\n",
+    "    linewidth=1.5,\n",
+    "    capsize=3,\n",
+    "    alpha=0.6,\n",
     ")\n",
-    "ax.set(xlabel='number of observations (beyond initial points)', ylabel='Log10 Regret')\n",
+    "ax.set(xlabel=\"number of observations (beyond initial points)\", ylabel=\"Log10 Regret\")\n",
     "ax.legend(loc=\"lower left\")"
    ]
   }