new framework

.github/workflows/run.yml

@@ -0,0 +1,42 @@
+name: run
+on:
+  push:
+    branches: ['**']
+    tags: ['**']
+    paths-ignore: ['LICENSE*', 'LICENCE*', 'README.md']
+jobs:
+  check:
+    defaults: {run: {shell: 'bash -el {0}'}}
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v4
+    - uses: conda-incubator/setup-miniconda@v3
+      with: {mamba-version: "*", channels: conda-forge}
+    - name: Install dependencies
+      run: |
+        conda activate test
+        if test -f apt.txt; then
+          sudo apt-get update
+          xargs -a apt.txt sudo apt-get install -y
+        fi
+        if test -f environment.yml; then
+          conda install --file environment.yml
+        fi
+        if test -f requirements.txt; then
+          pip install -r requirements.txt
+        fi
+    - name: Test imports
+      run:
+        conda activate test
+        python <<EOF
+        from main import Submission
+        from cil.optimisation.algorithms import Algorithm
+        assert issubclass(Submission, Algorithm)
+        EOF
+  full:
+    if: startsWith(github.ref, 'refs/tags')
+    runs-on: [self-hosted, docker, cuda]
+    steps:
+    - uses: actions/checkout@v4
+      with: {fetch-depth: 0, submodules: recursive}
+    - run: petric

README.md

@@ -1,40 +1,30 @@
 # SyneRBI Challenge
 
+## TL;DR
+
 The organisers will execute:
 
 ```python
-test = main.Test("/some/private/input_patients_dir")
+from main import Submission
+assert issubclass(Submission, cil.optimisation.algorithms.Algorithm)
 with Timeout(minutes=5):
-    test.run_all_patients("/some/private/output_patients_dir")
-evaluate_metrics("/some/private/output_patients_dir")
+    Submission(data).run(np.inf, callbacks=metrics)
 ```
 
-To avoid timing out, please disable any debugging/plotting code before submitting!
-
-Layout:
-
-- /challenge/
-  - input_patients_dir/
-    - patient_01/
-      - sinogram.npy
-      - reference.npy
-    - patient_02/
-    - ...
-  - output_patients_dir/
-    - patient_01/
-      - 000.npy
-      - 001.npy
-      - ...
-
-Private (organiser) test machine:
-
-- /$JOB_ID/
-  - input_patients_dir/
-    - patient_01/
-      - sinogram.npy
-  - output_patients_dir/
-    - patient_01/
-      - 000.npy
-- /hidden/patients/reference/dir/
-  - patient_01/
-    - reference.npy
+> [!WARNING]
+> To avoid timing out, please disable any debugging/plotting code before submitting!
+
+The organisers will have private versions of `data` and `metrics`.
+Smaller test (public) versions of `data` and `metrics` are defined in the [`notebook.ipynb`](notebook.ipynb).
+
+## Layout
+
+Only [`main.py`](main.py) is required.
+[SIRF](https://github.com/SyneRBI/SIRF), [CIL](https://github.com/TomographicImaging/CIL), and CUDA are already installed.
+Additional dependencies may be specified via `apt.txt`, `environment.yml`, and/or `requirements.txt`.
+
+- (required) `main.py`: must define a `class Submission(cil.optimisation.algorithms.Algorithm)`
+- `notebook.ipynb`: can be used for experimenting. Runs `main.Submission()` on test `data` with basic `metrics`
+- `apt.txt`: passed to `apt install`
+- `environment.yml`: passed to `conda install`
+- `requirements.txt`: passed to `pip install`

main.py

@@ -1,43 +1,11 @@
-from pathlib import Path, PurePath
+from cil.optimisation.algorithms import GD
 
-import numpy as np
-from tqdm import trange
+class Submission(GD):
+    def __init__(self, data, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        # Your code here
+        self.data = data
 
-
-class PatientData:
-    def __init__(self, patient_dir: PurePath):
-        self.patient_dir = Path(patient_dir)
-        self.sinogram = np.random.random((252, 344, 344)) # TODO
-        self.image_size = (127, 344, 344) # TODO
-        self.image_reference = np.random.random(self.image_size) # TODO
-
-
-class Test:
-    def __init__(self, input_patients_dir: PurePath):
-        self.input_patients_dir = Path(input_patients_dir)
-
-    def run_all_patients(self, output_patients_dir: PurePath | None = None):
-        for input_dir in self.input_patients_dir.glob("*/"):
-            output_dir = Path(output_patients_dir) / input_dir.name if output_patients_dir else None
-            self.run(input_dir, output_dir=output_dir)
-
-    def run(self, input_dir: PurePath, output_dir: PurePath | None = None, max_iterations=100):
-        data = PatientData(input_dir)
-        img = np.ones(data.image_size)
-        with trange(max_iterations) as t:
-            for iter in t:
-                img = self.reconstruct(data.sinogram, img)
-                if output_dir:
-                    np.save(Path(output_dir) / f"{iter:03d}.npy", img)
-                nrmse = (((img - data.image_reference) ** 2).sum() / (data.image_reference ** 2).sum()) ** 0.5
-                t.set_postfix(nrmse=nrmse, refresh=False)
-
-    def reconstruct(self, sinogram: np.ndarray, image: np.ndarray) -> np.ndarray:
-        raise NotImplementedError("""TODO: write your code for 1 iteration here, e.g.
-            return image * backward(sinogram / (forward(image) + randoms_and_scatter)) / backward(np.ones_like(image))
-        """)
-
-
-if __name__ == "__main__":
-    test = Test("input_patients_dir")
-    test.run_all_patients("output_patients_dir")
+    def update(self):
+        # Your code here
+        return super().update()