diff --git a/.ipynb_checkpoints/hierarchical_sisso-checkpoint.ipynb b/.ipynb_checkpoints/hierarchical_sisso-checkpoint.ipynb
index 6cb0b7620a24dc7f61a7b63e600b2663797c6625..4b65365d9eb6c0825c97ddd7eecf5048ec7db827 100644
--- a/.ipynb_checkpoints/hierarchical_sisso-checkpoint.ipynb
+++ b/.ipynb_checkpoints/hierarchical_sisso-checkpoint.ipynb
@@ -11,9 +11,9 @@
"cell_type": "markdown",
"metadata": {},
"source": [
- "<img style=\"float: left;\" src=\"assets/hisisso/logo_MPG.png\" width=150>\n",
- "<img style=\"float: left; margin-top: -10px\" src=\"assets/hisisso/logo_NOMAD.png\" width=250>\n",
- "<img style=\"float: left; margin-top: -5px\" src=\"assets/hisisso/logo_HU.png\" width=130>"
+ "<img style=\"float: left;\" src=\"assets/hierarchical_sisso/logo_MPG.png\" width=150>\n",
+ "<img style=\"float: left; margin-top: -10px\" src=\"assets/hierarchical_sisso/logo_NOMAD.png\" width=250>\n",
+ "<img style=\"float: left; margin-top: -5px\" src=\"assets/hierarchical_sisso/logo_HU.png\" width=130>"
]
},
{
@@ -285,7 +285,51 @@
" avg_cv_errors = cv_errors.mean(axis=0)\n",
" std_cv_errors = cv_errors.std(axis=0)\n",
" \n",
- " return(train_errors, avg_cv_errors, std_cv_errors) "
+ " return(train_errors, avg_cv_errors, std_cv_errors)\n",
+ "\n",
+ "def get_model(path,dim,rung,prop,unit,mode):\n",
+ " \"\"\"\n",
+ " reads cpp-sisso output and returns the model and model components\n",
+ " arguments: path(str): directory containing the output files\n",
+ " dim(int): model dimension\n",
+ " rung(int): number of iterations for operator application (e.g. 1, 2 or 3)\n",
+ " prop(str): property label\n",
+ " unit(str): property unit\n",
+ " mode(str): 'train' or 'test' to obtain the models evaluated for training and test materials, respectively\n",
+ " \"\"\"\n",
+ " if mode == \"train\":\n",
+ " model = load_model(f\"{path}/models/train_dim_{dim}_model_0.dat\")\n",
+ " else:\n",
+ " model = load_model(\n",
+ " f\"{path}/models/train_dim_{dim}_model_0.dat\", f\"{path}/models/test_dim_{dim}_model_0.dat\"\n",
+ " )\n",
+ " \n",
+ " coefficients = model.coefs[0]\n",
+ " columns = [f\"{prop} {unit}\", f\"pred_{prop}_r{rung} {unit}\"]\n",
+ " columns += [f\"f{ii}\" for ii in range(1, len(model.feats) + 1)]\n",
+ "\n",
+ " if mode == \"train\":\n",
+ " df = pd.DataFrame(\n",
+ " index = [mat_id.strip() for mat_id in model.sample_ids_train],\n",
+ " data = np.column_stack((model.prop_train.reshape(-1, 1), model.fit.reshape(-1, 1), np.array([feat.value for feat in model.feats]).T)),\n",
+ " columns=columns,\n",
+ " )\n",
+ " else:\n",
+ " df = pd.DataFrame(\n",
+ " index = [mat_id.strip() for mat_id in model.sample_ids_test],\n",
+ " data = np.column_stack((model.prop_test.reshape(-1, 1), model.predict.reshape(-1, 1), np.array([feat.test_value for feat in model.feats]).T)),\n",
+ " columns=columns,\n",
+ " )\n",
+ " \n",
+ " df.dropna(axis=1, inplace=True)\n",
+ " df.drop(labels=f\"{prop} {unit}\", axis=1, inplace=True)\n",
+ " \n",
+ " for i in range(len(coefficients)-1):\n",
+ " j=i+1\n",
+ " df[f\"a{j}_r{rung}_{prop} {unit}\"]=df[f\"f{j}\"] * coefficients[i]\n",
+ " df.drop(labels=f\"f{j}\", axis=1, inplace=True)\n",
+ "\n",
+ " return(df)"
]
},
{
@@ -808,57 +852,6 @@
"The case of bulk modulus will be treated separately below, since the information on lattice constants and cohesive energy will be used to model the bulk modulus. "
]
},
- {
- "cell_type": "code",
- "execution_count": null,
- "metadata": {},
- "outputs": [],
- "source": [
- "def get_model(path,dim,rung,prop,unit,mode):\n",
- " \"\"\"\n",
- " reads cpp-sisso output and returns the model and model components\n",
- " arguments: path(str): directory containing the output files\n",
- " dim(int): model dimension\n",
- " rung(int): number of iterations for operator application (e.g. 1, 2 or 3)\n",
- " prop(str): property label\n",
- " unit(str): property unit\n",
- " mode(str): 'train' or 'test' to obtain the models evaluated for training and test materials, respectively\n",
- " \"\"\"\n",
- " if mode == \"train\":\n",
- " model = load_model(f\"{path}/models/train_dim_{dim}_model_0.dat\")\n",
- " else:\n",
- " model = load_model(\n",
- " f\"{path}/models/train_dim_{dim}_model_0.dat\", f\"{path}/models/test_dim_{dim}_model_0.dat\"\n",
- " )\n",
- " \n",
- " coefficients = model.coefs[0]\n",
- " columns = [f\"{prop} {unit}\", f\"pred_{prop}_r{rung} {unit}\"]\n",
- " columns += [f\"f{ii}\" for ii in range(1, len(model.feats) + 1)]\n",
- "\n",
- " if mode == \"train\":\n",
- " df = pd.DataFrame(\n",
- " index = [mat_id.strip() for mat_id in model.sample_ids_train],\n",
- " data = np.column_stack((model.prop_train.reshape(-1, 1), model.fit.reshape(-1, 1), np.array([feat.value for feat in model.feats]).T)),\n",
- " columns=columns,\n",
- " )\n",
- " else:\n",
- " df = pd.DataFrame(\n",
- " index = [mat_id.strip() for mat_id in model.sample_ids_test],\n",
- " data = np.column_stack((model.prop_test.reshape(-1, 1), model.predict.reshape(-1, 1), np.array([feat.test_value for feat in model.feats]).T)),\n",
- " columns=columns,\n",
- " )\n",
- " \n",
- " df.dropna(axis=1, inplace=True)\n",
- " df.drop(labels=f\"{prop} {unit}\", axis=1, inplace=True)\n",
- " \n",
- " for i in range(len(coefficients)-1):\n",
- " j=i+1\n",
- " df[f\"a{j}_r{rung}_{prop} {unit}\"]=df[f\"f{j}\"] * coefficients[i]\n",
- " df.drop(labels=f\"f{j}\", axis=1, inplace=True)\n",
- "\n",
- " return(df)"
- ]
- },
{
"cell_type": "code",
"execution_count": null,
diff --git a/hierarchical_sisso.ipynb b/hierarchical_sisso.ipynb
index 6cb0b7620a24dc7f61a7b63e600b2663797c6625..4b65365d9eb6c0825c97ddd7eecf5048ec7db827 100644
--- a/hierarchical_sisso.ipynb
+++ b/hierarchical_sisso.ipynb
@@ -11,9 +11,9 @@
"cell_type": "markdown",
"metadata": {},
"source": [
- "<img style=\"float: left;\" src=\"assets/hisisso/logo_MPG.png\" width=150>\n",
- "<img style=\"float: left; margin-top: -10px\" src=\"assets/hisisso/logo_NOMAD.png\" width=250>\n",
- "<img style=\"float: left; margin-top: -5px\" src=\"assets/hisisso/logo_HU.png\" width=130>"
+ "<img style=\"float: left;\" src=\"assets/hierarchical_sisso/logo_MPG.png\" width=150>\n",
+ "<img style=\"float: left; margin-top: -10px\" src=\"assets/hierarchical_sisso/logo_NOMAD.png\" width=250>\n",
+ "<img style=\"float: left; margin-top: -5px\" src=\"assets/hierarchical_sisso/logo_HU.png\" width=130>"
]
},
{
@@ -285,7 +285,51 @@
" avg_cv_errors = cv_errors.mean(axis=0)\n",
" std_cv_errors = cv_errors.std(axis=0)\n",
" \n",
- " return(train_errors, avg_cv_errors, std_cv_errors) "
+ " return(train_errors, avg_cv_errors, std_cv_errors)\n",
+ "\n",
+ "def get_model(path,dim,rung,prop,unit,mode):\n",
+ " \"\"\"\n",
+ " reads cpp-sisso output and returns the model and model components\n",
+ " arguments: path(str): directory containing the output files\n",
+ " dim(int): model dimension\n",
+ " rung(int): number of iterations for operator application (e.g. 1, 2 or 3)\n",
+ " prop(str): property label\n",
+ " unit(str): property unit\n",
+ " mode(str): 'train' or 'test' to obtain the models evaluated for training and test materials, respectively\n",
+ " \"\"\"\n",
+ " if mode == \"train\":\n",
+ " model = load_model(f\"{path}/models/train_dim_{dim}_model_0.dat\")\n",
+ " else:\n",
+ " model = load_model(\n",
+ " f\"{path}/models/train_dim_{dim}_model_0.dat\", f\"{path}/models/test_dim_{dim}_model_0.dat\"\n",
+ " )\n",
+ " \n",
+ " coefficients = model.coefs[0]\n",
+ " columns = [f\"{prop} {unit}\", f\"pred_{prop}_r{rung} {unit}\"]\n",
+ " columns += [f\"f{ii}\" for ii in range(1, len(model.feats) + 1)]\n",
+ "\n",
+ " if mode == \"train\":\n",
+ " df = pd.DataFrame(\n",
+ " index = [mat_id.strip() for mat_id in model.sample_ids_train],\n",
+ " data = np.column_stack((model.prop_train.reshape(-1, 1), model.fit.reshape(-1, 1), np.array([feat.value for feat in model.feats]).T)),\n",
+ " columns=columns,\n",
+ " )\n",
+ " else:\n",
+ " df = pd.DataFrame(\n",
+ " index = [mat_id.strip() for mat_id in model.sample_ids_test],\n",
+ " data = np.column_stack((model.prop_test.reshape(-1, 1), model.predict.reshape(-1, 1), np.array([feat.test_value for feat in model.feats]).T)),\n",
+ " columns=columns,\n",
+ " )\n",
+ " \n",
+ " df.dropna(axis=1, inplace=True)\n",
+ " df.drop(labels=f\"{prop} {unit}\", axis=1, inplace=True)\n",
+ " \n",
+ " for i in range(len(coefficients)-1):\n",
+ " j=i+1\n",
+ " df[f\"a{j}_r{rung}_{prop} {unit}\"]=df[f\"f{j}\"] * coefficients[i]\n",
+ " df.drop(labels=f\"f{j}\", axis=1, inplace=True)\n",
+ "\n",
+ " return(df)"
]
},
{
@@ -808,57 +852,6 @@
"The case of bulk modulus will be treated separately below, since the information on lattice constants and cohesive energy will be used to model the bulk modulus. "
]
},
- {
- "cell_type": "code",
- "execution_count": null,
- "metadata": {},
- "outputs": [],
- "source": [
- "def get_model(path,dim,rung,prop,unit,mode):\n",
- " \"\"\"\n",
- " reads cpp-sisso output and returns the model and model components\n",
- " arguments: path(str): directory containing the output files\n",
- " dim(int): model dimension\n",
- " rung(int): number of iterations for operator application (e.g. 1, 2 or 3)\n",
- " prop(str): property label\n",
- " unit(str): property unit\n",
- " mode(str): 'train' or 'test' to obtain the models evaluated for training and test materials, respectively\n",
- " \"\"\"\n",
- " if mode == \"train\":\n",
- " model = load_model(f\"{path}/models/train_dim_{dim}_model_0.dat\")\n",
- " else:\n",
- " model = load_model(\n",
- " f\"{path}/models/train_dim_{dim}_model_0.dat\", f\"{path}/models/test_dim_{dim}_model_0.dat\"\n",
- " )\n",
- " \n",
- " coefficients = model.coefs[0]\n",
- " columns = [f\"{prop} {unit}\", f\"pred_{prop}_r{rung} {unit}\"]\n",
- " columns += [f\"f{ii}\" for ii in range(1, len(model.feats) + 1)]\n",
- "\n",
- " if mode == \"train\":\n",
- " df = pd.DataFrame(\n",
- " index = [mat_id.strip() for mat_id in model.sample_ids_train],\n",
- " data = np.column_stack((model.prop_train.reshape(-1, 1), model.fit.reshape(-1, 1), np.array([feat.value for feat in model.feats]).T)),\n",
- " columns=columns,\n",
- " )\n",
- " else:\n",
- " df = pd.DataFrame(\n",
- " index = [mat_id.strip() for mat_id in model.sample_ids_test],\n",
- " data = np.column_stack((model.prop_test.reshape(-1, 1), model.predict.reshape(-1, 1), np.array([feat.test_value for feat in model.feats]).T)),\n",
- " columns=columns,\n",
- " )\n",
- " \n",
- " df.dropna(axis=1, inplace=True)\n",
- " df.drop(labels=f\"{prop} {unit}\", axis=1, inplace=True)\n",
- " \n",
- " for i in range(len(coefficients)-1):\n",
- " j=i+1\n",
- " df[f\"a{j}_r{rung}_{prop} {unit}\"]=df[f\"f{j}\"] * coefficients[i]\n",
- " df.drop(labels=f\"f{j}\", axis=1, inplace=True)\n",
- "\n",
- " return(df)"
- ]
- },
{
"cell_type": "code",
"execution_count": null,
diff --git a/setup.py b/setup.py
index 6cec3c40cc897eaa81e7cc62ca0bc4e962c2eadb..19639afc490da995ff6c62af21d5026be31158e3 100644
--- a/setup.py
+++ b/setup.py
@@ -13,6 +13,6 @@ setup(
description=metainfo['title'],
long_description=metainfo['description'],
packages=find_packages(),
- install_requires=['pandas', 'numpy', 'matplotlib', 'scikit-learn', 'scipy', 'json', 'seaborn', 'sissopp'],
+ install_requires=['pandas', 'numpy', 'matplotlib', 'scikit-learn', 'scipy', 'json', 'seaborn'],
)