Merge branch 'main' of https://github.com/agosztolai/MARBLE into main

.gitignore

@@ -11,3 +11,4 @@ myjob*
 __pycache__/
 *.egg
 *.egg-info/
+log/

MARBLE/preprocessing.py

@@ -117,7 +117,7 @@ def _compute_geometric_objects(
     n_geodesic_nb=2.0,
     var_explained=0.9,
     return_spectrum=True,
-    local_gauges=True,
+    local_gauges=False,
     compute_laplacian=False,
     compute_connection_laplacian=False,
     dim_man=None,

README.md

@@ -34,29 +34,32 @@ The code is tested for both cpu and gpu (CUDA) machines running Linux or OSX. Al
 
 We recommend you install the code in a fresh Anaconda virtual environment, as follows.
 
-First, clone this repository, 
+- First, clone this repository, 
 
 ```
 git clone https://github.com/agosztolai/MARBLE
 ```
 
-Then, create an new anaconda environment using the provided environment file that matches your system.
+- Then, create an new anaconda environment using the provided environment file that matches your system.
+  - For Linux machines with CUDA:
 
-For Linux machines with CUDA: 
+  `conda env create -f environment.yml`
+  - For Intel-based Mac:
 
-```
-conda env create -f environment.yml
-```
-
-For Mac without CUDA:
+  `conda env create -f environment_osx_intel.yml`
 
-```
-conda env create -f environment_cpu_osx.yml
-```
+  - For recent M1/M2/M3 Mac:
+    - Install cmake `brew install cmake` or using the installer on the [cmake website](https://cmake.org/download/)
+    - Create the environment
 
-This will install all the requires dependencies. Finally, install by running inside the main folder
+        `conda env create -f environment_osx_arm.yml`
+    - Activate the environment `conda activate MARBLE` 
+    - Install pytorch geometric
+    `pip install -r requirements_osx_arm.txt`
+- All the required dependencies are now installed. Finally, activate the environment and install by running inside the main folder
 
 ```
+conda activate MARBLE
 pip install . 
 ```
 

environment_osx_arm.yml

@@ -0,0 +1,109 @@
+name: MARBLE
+channels:
+  - pytorch
+  - defaults
+dependencies:
+  - blas=1.0=openblas
+  - bottleneck=1.3.5=py39heec5a64_0
+  - brotli=1.0.9=h1a28f6b_7
+  - brotli-bin=1.0.9=h1a28f6b_7
+  - brotlipy=0.7.0=py39h1a28f6b_1002
+  - bzip2=1.0.8=h620ffc9_4
+  - ca-certificates=2023.08.22=hca03da5_0
+  - certifi=2023.7.22=py39hca03da5_0
+  - cffi=1.15.1=py39h80987f9_3
+  - charset-normalizer=2.0.4=pyhd3eb1b0_0
+  - contourpy=1.0.5=py39h525c30c_0
+  - cryptography=41.0.3=py39hd4332d6_0
+  - cycler=0.11.0=pyhd3eb1b0_0
+  - ffmpeg=4.2.2=h04105a8_0
+  - fonttools=4.25.0=pyhd3eb1b0_0
+  - freetype=2.12.1=h1192e45_0
+  - gettext=0.21.0=h13f89a0_1
+  - giflib=5.2.1=h80987f9_3
+  - gmp=6.2.1=hc377ac9_3
+  - gnutls=3.6.15=h887c41c_0
+  - icu=73.1=h313beb8_0
+  - idna=3.4=py39hca03da5_0
+  - importlib_resources=5.2.0=pyhd3eb1b0_1
+  - jpeg=9e=h80987f9_1
+  - kiwisolver=1.4.4=py39h313beb8_0
+  - lame=3.100=h1a28f6b_0
+  - lcms2=2.12=hba8e193_0
+  - lerc=3.0=hc377ac9_0
+  - libbrotlicommon=1.0.9=h1a28f6b_7
+  - libbrotlidec=1.0.9=h1a28f6b_7
+  - libbrotlienc=1.0.9=h1a28f6b_7
+  - libcxx=14.0.6=h848a8c0_0
+  - libdeflate=1.17=h80987f9_1
+  - libffi=3.4.4=hca03da5_0
+  - libgfortran=5.0.0=11_3_0_hca03da5_28
+  - libgfortran5=11.3.0=h009349e_28
+  - libiconv=1.16=h1a28f6b_2
+  - libidn2=2.3.4=h80987f9_0
+  - libopenblas=0.3.21=h269037a_0
+  - libopus=1.3=h1a28f6b_1
+  - libpng=1.6.39=h80987f9_0
+  - libtasn1=4.19.0=h80987f9_0
+  - libtiff=4.5.1=h313beb8_0
+  - libunistring=0.9.10=h1a28f6b_0
+  - libvpx=1.10.0=hc377ac9_0
+  - libwebp=1.3.2=ha3663a8_0
+  - libwebp-base=1.3.2=h80987f9_0
+  - libxml2=2.10.4=h0dcf63f_1
+  - llvm-openmp=14.0.6=hc6e5704_0
+  - lz4-c=1.9.4=h313beb8_0
+  - matplotlib=3.7.2=py39hca03da5_0
+  - matplotlib-base=3.7.2=py39h46d7db6_0
+  - munkres=1.1.4=py_0
+  - ncurses=6.4=h313beb8_0
+  - nettle=3.7.3=h84b5d62_1
+  - networkx=3.1=py39hca03da5_0
+  - numexpr=2.8.7=py39hecc3335_0
+  - numpy=1.26.0=py39h3b2db8e_0
+  - numpy-base=1.26.0=py39ha9811e2_0
+  - openh264=1.8.0=h98b2900_0
+  - openjpeg=2.3.0=h7a6adac_2
+  - openssl=3.0.11=h1a28f6b_2
+  - packaging=23.1=py39hca03da5_0
+  - pandas=2.1.1=py39h46d7db6_0
+  - pillow=10.0.1=py39h3b245a6_0
+  - pip=23.3=py39hca03da5_0
+  - pybind11=2.10.4=py39h48ca7d4_0
+  - pybind11-global=2.10.4=py39h48ca7d4_0
+  - pycparser=2.21=pyhd3eb1b0_0
+  - pyopenssl=23.2.0=py39hca03da5_0
+  - pyparsing=3.0.9=py39hca03da5_0
+  - pysocks=1.7.1=py39hca03da5_0
+  - python=3.9.18=hb885b13_0
+  - python-dateutil=2.8.2=pyhd3eb1b0_0
+  - python-tzdata=2023.3=pyhd3eb1b0_0
+  - pytorch=1.12.1=py3.9_0
+  - pytz=2023.3.post1=py39hca03da5_0
+  - pyyaml=6.0=py39h80987f9_1
+  - readline=8.2=h1a28f6b_0
+  - requests=2.31.0=py39hca03da5_0
+  - scipy=1.11.3=py39h20cbe94_0
+  - seaborn=0.12.2=py39hca03da5_0
+  - setuptools=68.0.0=py39hca03da5_0
+  - scikit-learn=1.3.0=py39h46d7db6_0
+  - six=1.16.0=pyhd3eb1b0_1
+  - sqlite=3.41.2=h80987f9_0
+  - tk=8.6.12=hb8d0fd4_0
+  - torchaudio=0.12.1=py39_cpu
+  - torchvision=0.13.1=py39_cpu
+  - tornado=6.3.3=py39h80987f9_0
+  - tqdm=4.65.0=py39h86d0a89_0
+  - typing_extensions=4.7.1=py39hca03da5_0
+  - tzdata=2023c=h04d1e81_0
+  - urllib3=1.26.16=py39hca03da5_0
+  - wheel=0.41.2=py39hca03da5_0
+  - x264=1!152.20180806=h1a28f6b_0
+  - xz=5.4.2=h80987f9_0
+  - yaml=0.2.5=h1a28f6b_0
+  - zipp=3.11.0=py39hca03da5_0
+  - zlib=1.2.13=h5a0b063_0
+  - zstd=1.5.5=hd90d995_0
+  - pip:
+      - ninja==1.11.1.1
+      - teaspoon==1.3.1

examples/RNN/RNN_scripts/helpers.py

@@ -35,14 +35,13 @@ def load_network(f):
     return z, net
 
 
-def sample_network(net, f):
+def sample_network(net, f, seed=0):
 
     if os.path.exists(f):
         print('Network found with same name. Loading...')
         return torch.load(open(f, "rb"))
 
     n_pops = 2
-    seed = 0
     z, _ = clustering.gmm_fit(net, n_pops, algo="bayes", random_state=seed)
     net_sampled = clustering.to_support_net(net, z)
 
@@ -68,7 +67,7 @@ def sample_network(net, f):
 
 
 def generate_trajectories(
-    net, input=None, epochs=None, n_traj=None, fname="./outputs/RNN_trajectories.pkl"
+    net, input=None, epochs=None, n_traj=None, fname="./data/RNN_trajectories.pkl"
 ):
 
     if fname is not None:
@@ -455,7 +454,7 @@ def plot_experiment(net, input, traj, epochs, rect=(-8, 8, -6, 6), traj_to_show=
     fig.subplots_adjust(hspace=0.1, wspace=0.1)
 
 
-def aggregate_data(traj, epochs, transient=10, only_stim=False):
+def aggregate_data(traj, epochs, transient=10, only_stim=False, pca=True):
 
     n_conds = len(traj)
     n_epochs = len(epochs) - 1
@@ -468,9 +467,10 @@ def aggregate_data(traj, epochs, transient=10, only_stim=False):
             for j in range(n_traj):  # trajectories
                 pos.append(traj[i][j][k][transient:])
 
-    pca = PCA(n_components=3)
-    pca.fit(np.vstack(pos))
-    print("Explained variance: ", pca.explained_variance_ratio_)
+    if pca:
+        pca = PCA(n_components=3)
+        pca.fit(np.vstack(pos))
+        print("Explained variance: ", pca.explained_variance_ratio_)
 
     # aggregate data under baseline condition (no input)
     pos, vel = [], []
@@ -480,7 +480,8 @@ def aggregate_data(traj, epochs, transient=10, only_stim=False):
             for k in [0, 2, 4]:
                 for j in range(n_traj):  # trajectories
                     pos_proj = traj[i][j][k][transient:]
-                    pos_proj = pca.transform(pos_proj)
+                    if pca:
+                        pos_proj = pca.transform(pos_proj)
                     pos_.append(pos_proj[:-1])  # stack trajectories
                     vel_.append(np.diff(pos_proj, axis=0))  # compute differences
 
@@ -494,7 +495,8 @@ def aggregate_data(traj, epochs, transient=10, only_stim=False):
         for k in [1, 3]:
             for j in range(n_traj):  # trajectories
                 pos_proj = traj[i][j][k][transient:]
-                pos_proj = pca.transform(pos_proj)
+                if pca:
+                    pos_proj = pca.transform(pos_proj)
                 pos_.append(pos_proj[:-1])
                 vel_.append(np.diff(pos_proj, axis=0))