Merge branch 'main' into docs/HadilO/onboarding

.github/workflows/python-testing.yml

@@ -13,8 +13,8 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ['3.10', '3.11']
-        os: [ubuntu-latest, macOS-latest] #,windows-latest]
+        python-version: ['3.10'] #3.11
+        os: [ubuntu-latest] #, macOS-latest] #,windows-latest]
     steps:
     - uses: actions/checkout@v3
     - name: Set up Python ${{ matrix.python-version }}
@@ -24,12 +24,12 @@ jobs:
     - name: Setup PDM
       uses: pdm-project/[email protected]
       with: 
-        version: 2.10.4
+        version: 2.15.4
         cache: true
 
     - name: Install dependencies
       run: |
-        pdm install -d -G tensorflow -G rebound
+        pdm install -G all
     - name: Analysing the code with pylint
       run: |
         pdm run pylint $(git ls-files '*.py')

CONTRIBUTING.md

@@ -39,7 +39,7 @@
        pyenv shell 3.10.14   #use the newly installed version
        ```
 
-   Check your installation by running `which python` which should output something along the lines of:
+   Check your installation by running `which python`. Output should be along the lines of:
    - for mise: `/Users/[user]/.local/share/mise/installs/python/3.10.14/bin/python`
 
    - for pyenv: `/Users/[user]/.pyenv/shims/python`

demos/systems/kuramoto_demo.ipynb

@@ -96,9 +96,9 @@
     }
    ],
    "source": [
-    "plot([x], legend_labels=[\"IND\"], max_oscillators=3, title=\"InD\", max_lines=2, phase_dynamics=True)\n",
+    "plot([x], labels=[\"IND\"], max_oscillators=3, title=\"InD\", max_lines=2, phase_dynamics=True)\n",
     "\n",
-    "plot([x], legend_labels=[\"IND\"], max_oscillators=3, title=\"InD\", max_lines=2, phase_dynamics=False)\n",
+    "plot([x], labels=[\"IND\"], max_oscillators=3, title=\"InD\", max_lines=2, phase_dynamics=False)\n",
     "# plot([y],  max_oscillators=3, title=\"OOD\")"
    ]
   },
@@ -126,7 +126,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.14"
+   "version": "3.10.11"
   }
  },
  "nbformat": 4,

src/dynadojo/baselines/dnn.py

@@ -10,7 +10,7 @@
 import time
 
 from ..abstractions import AbstractAlgorithm
-
+import logging
 
 
 class TorchBaseClass(AbstractAlgorithm, torch.nn.Module):
@@ -33,7 +33,8 @@ def __init__(
             torch.manual_seed(seed)
 
         self.device = device or "cuda" if torch.cuda.is_available() else "mps" if torch.backends.mps.is_available() else "cpu"
-        print(f"Using device: {self.device}")
+        logging.info(f"Using device: {self.device}")
+        # print(f"Using device: {self.device}")
         # self.model = self.create_model()
         self.criterion = torch.nn.MSELoss()
 
@@ -90,7 +91,7 @@ def fit(self, x: np.ndarray,
         losses = []
         self.train() 
         training_start_time = time.time()
-        print(f'Dataloader length: {len(dataloader)}')
+        # print(f'Dataloader length: {len(dataloader)}')
         for epoch in range(epochs):
             self.train()
             epoch_loss = 0
@@ -170,8 +171,9 @@ class DNN(TorchBaseClass):
     def __init__(self, 
             embed_dim,
             timesteps,
+            max_control_cost,
             **kwargs):
-        super().__init__(embed_dim, timesteps, **kwargs)
+        super().__init__(embed_dim, timesteps, max_control_cost, **kwargs)
         self.model = torch.nn.Sequential(
             torch.nn.Linear(self.embed_dim, embed_dim*10),
             torch.nn.ReLU(),

src/dynadojo/systems/ctln.py

@@ -69,7 +69,7 @@ def dynamics(t, state):
 
     def calc_error(self, x, y) -> float:
         error = x - y
-        return np.mean(error ** 2) / self.embed_dim
+        return np.mean(error ** 2)
 
     def calc_control_cost(self, control: np.ndarray) -> float:
         return np.linalg.norm(control, axis=(1, 2), ord=2)

src/dynadojo/systems/kuramoto.py

@@ -180,7 +180,7 @@ def solve(t, x0, U):
 
     def calc_error(self, x, y) -> float:
         error = x - y
-        return np.mean(error ** 2) / self.latent_dim
+        return np.mean(error ** 2)
 
     def calc_control_cost(self, control: np.ndarray) -> float:
         return np.linalg.norm(control, axis=(1, 2), ord=2)

src/dynadojo/systems/lv/competitive.py

@@ -168,7 +168,7 @@ def dynamics(t, X, u):
 
     def calc_error(self, x, y) -> float:
         error = x - y
-        return np.mean(error ** 2) / self.latent_dim
+        return np.mean(error ** 2)
 
     def calc_control_cost(self, control: np.ndarray) -> float:
         return np.linalg.norm(control, axis=(1, 2), ord=2)

src/dynadojo/systems/lv/prey_predator.py

@@ -237,7 +237,7 @@ def dynamics(t, X, u):
 
     def calc_error(self, x, y) -> float:
         error = x - y
-        return np.mean(error ** 2) / self.latent_dim
+        return np.mean(error ** 2)
 
     def calc_control_cost(self, control: np.ndarray) -> float:
         return np.linalg.norm(control, axis=(1, 2), ord=2)

src/dynadojo/systems/opinion/media_bias.py

@@ -93,7 +93,7 @@ def __init__(self, latent_dim=31, embed_dim=31,
         self.config.add_model_parameter("gamma_media", bias_media)
 
     def create_model(self, x0):
-        self.model = op.AlgorithmicBiasMediaModel(self.g)
+        self.model = op.AlgorithmicBiasModel(self.g)
         self.model.set_initial_status(self.config)
         self.model.status = x0
         self.model.initial_status = x0

src/dynadojo/systems/utils/epidemic.py

@@ -165,7 +165,7 @@ def dynamics(x0):
 
     def calc_error(self, x, y) -> float:
         error = x - y
-        return np.mean(error ** 2) / self.latent_dim
+        return np.mean(error ** 2)
 
     def calc_control_cost(self, control: np.ndarray) -> float:
         return np.linalg.norm(control, axis=(1, 2), ord=2)

src/dynadojo/systems/utils/fbsnn.py

@@ -209,7 +209,7 @@ def make_data(self, init_conds: np.ndarray, control: np.ndarray, timesteps: int,
 
     def calc_error(self, x, y) -> float:
         error = x - y
-        return np.mean(error ** 2) / self.latent_dim
+        return np.mean(error ** 2)
 
     def calc_control_cost(self, control: np.ndarray) -> float:
         return np.linalg.norm(control, axis=(1, 2), ord=2)

src/dynadojo/systems/utils/opinion.py

@@ -79,7 +79,7 @@ def dynamics(x0):
 
     def calc_error(self, x, y) -> float:
         error = x - y
-        return np.mean(error ** 2) / self.latent_dim
+        return np.mean(error ** 2)
 
     def calc_control_cost(self, control: np.ndarray) -> float:
         return np.linalg.norm(control, axis=(1, 2), ord=2)

src/dynadojo/systems/utils/simple.py

@@ -122,7 +122,7 @@ def make_init_conds(self, n: int, in_dist=True) -> np.ndarray:
     def calc_error(self, x, y) -> float:
         """Returns the MSE error normalized by the embedded dimension."""
         error = x - y
-        return np.mean(error ** 2) / self.embed_dim
+        return np.mean(error ** 2)
 
     def calc_control_cost(self, control: np.ndarray) -> float:
         """Calculates the L2 norm / dimension of every vector in the control"""

tests/test_deterministic.py

@@ -15,15 +15,15 @@
 # baselines
 from dynadojo.baselines.aug_ode import AugODE
 from dynadojo.baselines.cnn import CNN
-from dynadojo.baselines.dmd import DMD
+# from dynadojo.baselines.dmd import DMD
 from dynadojo.baselines.dnn import DNN
 from dynadojo.baselines.ode import ODE
 # from dynadojo.baselines.sindy import SINDy
 
 ALL_BASELINES = [
     # AugODE,
     # CNN,
-    DMD,
+    # DMD,
     DNN,
     # ODE,
     # SINDy
@@ -157,7 +157,7 @@ def test_with_trials(self, algo):
                                  algo_kwargs=None)
         cols = ['trial', 'latent_dim', 'embed_dim', 'timesteps', 'n', 'error', 'ood_error', 'total_cost',
                 'system_seed', 'algo_seed']
-        df1 = df1[cols]
+        df1 = df1[cols].loc[df2['trial'] == 1]
         df2 = df2[cols].loc[df2['trial'] == 1]
         self.assertEqual(df1, df2)