WIP: Add documentation for CUQIpy-Umbridge plugin

clients/README.md

@@ -563,3 +563,79 @@ which indeed returns `Ev = 2.9948`, i.e., close to 3 as expected. The script the
 [Full example sources here.](https://github.com/UM-Bridge/umbridge/blob/main/clients/matlab/sgmkClient.m)
 
 
+## CUQIpy client
+
+CUQIpy stands for Computational Uncertainty Quantification for Inverse Problems in python. It’s a robust Python package designed for modeling and solving inverse problems using Bayesian inference. Here’s what it brings to the table:
+
+- A straightforward high-level interface for UQ analysis.
+
+- Complete control over the models and methods.
+
+- An array of predefined distributions, samplers, models, and test problems.
+
+- Easy extendability for your unique needs.
+
+A number of CUQIpy Plugins are available as separate packages that expand the functionality of CUQIpy.
+
+### CUQIpy-UMBridge
+
+The CUQIpy-Umbridge plugin allows you to use UM-Bridge models as a forward model in CUQIpy and UM-Bridge posterior models as distributions in CUQIpy.
+
+The plugin can be installed from pip
+
+```
+pip install cuqipy-umbridge
+```
+
+This installs CUQIpy and Umbridge as dependencies.
+
+### Examples
+
+Given a model server running at `http://localhost:4242`, we can connect to it and use it as a forward model in CUQIpy.
+
+TODO: ADD EXPLICIT MODEL FROM DOCKER IMAGES
+
+```python
+import cuqipy_umbridge
+import numpy as np
+
+# Create CUQIpy forward model
+model = cuqipy_umbridge.client.create_model('http://localhost:4242', 'forward')
+
+# Print type of model and domain and range geometries
+print(model)
+
+# Evaluate the model
+result = model(np.ones(model.domain_dim))
+print(result)
+```
+
+Given a model server running at `http://localhost:4242`, we can connect to it and use it as a posterior model in CUQIpy. 
+
+TODO: ADD EXPLICIT DISTRIBUTION FROM DOCKER IMAGES
+
+```python
+import cuqi
+import cuqipy_umbridge
+import numpy as np
+
+# Create CUQIpy distribution
+dist = cuqipy_umbridge.client.create_distribution('http://localhost:4242', 'posterior')
+
+# Print geometry of distribution
+print(dist)
+
+# Sample distribution using MH sampler
+sampler = cuqi.sampler.MH(dist)
+
+# Sample from distribution (1000 samples with 200 burn-in)
+samples = sampler.sample(Ns=1000, Nb=200)
+
+# Plot trace of samples
+samples.plot_trace()
+```
+
+
+
+
+

clients/python/cuqipy-client.py

@@ -0,0 +1,33 @@
+# Example depends on release of CUQIpy-Umbridge to PyPI
+
+import argparse
+import cuqipy_umbridge # CUQIpy UM-Bridge integration.
+import cuqi # pip install cuqipy
+
+
+if __name__ == "__main__":
+
+    # Read URL from command line argument
+    parser = argparse.ArgumentParser(description='Minimal HTTP model demo.')
+
+    parser.add_argument('url', metavar='url', type=str,
+                        help='the URL at which the model is running, for example http://localhost')
+
+    args = parser.parse_args()
+
+    print(f"Connecting to host URL {args.url}")
+
+    # Set up CUQIpy distribution for umbridge model named "posterior"
+    posterior = cuqipy_umbridge.client.create_distribution(args.url, "posterior")
+
+    # Print dimension of posterior
+    print(f"Dimension of posterior is {posterior.dim}")
+
+    # Sample posterior using CUQIpy
+    sampler = cuqi.sampler.MH(posterior)
+
+    # Sample from posterior
+    samples = sampler.sample(1000)
+
+    # Plot samples
+    samples.plot_trace()

models/README.md

@@ -143,4 +143,75 @@ The [MIT Uncertainty Quantification library (MUQ)](https://mituq.bitbucket.io),
 muq::Modeling::serveModPiece(mod_piece, "0.0.0.0", 4242);
 ```
 
-See MUQ's documentation for more in-depth documentation on model graphs and UM-Bridge integration.
+See MUQ's documentation for more in-depth documentation on model graphs and UM-Bridge integration.
+
+## CUQIpy server
+
+CUQIpy stands for Computational Uncertainty Quantification for Inverse Problems in python. It’s a robust Python package designed for modeling and solving inverse problems using Bayesian inference. Here’s what it brings to the table:
+
+- A straightforward high-level interface for UQ analysis.
+
+- Complete control over the models and methods.
+
+- An array of predefined distributions, samplers, models, and test problems.
+
+- Easy extendability for your unique needs.
+
+A number of CUQIpy Plugins are available as separate packages that expand the functionality of CUQIpy.
+
+### CUQIpy-UMBridge
+
+The CUQIpy-Umbridge plugin allows you to serve CUQIpy models and distributions via UM-Bridge.
+
+The plugin can be installed from pip
+
+```
+pip install cuqipy-umbridge
+```
+
+This installs CUQIpy and Umbridge as dependencies.
+
+### Examples
+
+Serving a CUQIpy model via UM-Bridge is as simple as:
+
+```python
+import cuqi
+import cuqipy_umbridge
+
+# This is the forward model to serve
+# This can be changed to any other CUQIpy forward model
+# Current example is the forward model of a 1D deconvolution test problem
+
+model = cuqi.testproblem.Deconvolution1D().model
+
+# Serve the distribution as UM-Bridge model
+# Name will be name of CUQIpy class for distribution
+# In this case "LinearModel"
+cuqipy_umbridge.server.serve_model(model)
+
+```
+
+
+Serving a CUQIpy distribution via UM-Bridge is as simple as:
+
+```python
+
+import cuqi
+import cuqipy_umbridge
+
+# This is the distribution to serve
+# This can be changed to any other CUQIpy distribution
+# Current example is the posterior of a 1D deconvolution test problem
+
+dist = cuqi.testproblem.Deconvolution1D().posterior
+
+# Serve the distribution as UM-Bridge model
+# Name will be name of CUQIpy class for distribution
+# In this case "Posterior"
+cuqipy_umbridge.server.serve_distribution(dist)
+
+```
+
+
+

models/cuqipy-distribution/Dockerfile

@@ -0,0 +1,9 @@
+FROM ubuntu:latest
+
+COPY distribution-server.py /
+
+RUN apt update && \
+    DEBIAN_FRONTEND="noninteractive" apt install -y python3-pip && \
+    pip install cuqipy-umbridge
+
+CMD python3 distribution-server.py

models/cuqipy-distribution/distribution-server.py

@@ -0,0 +1,14 @@
+import cuqi
+import cuqipy_umbridge
+
+# This is the distribution to serve
+# This can be changed to any other CUQIpy distribution
+# Current example is the posterior of a 1D deconvolution test problem
+
+dist = cuqi.testproblem.Deconvolution1D().posterior
+
+# Serve the distribution as UM-Bridge model
+# Name will be name of CUQIpy class for distribution
+# In this case "Posterior"
+cuqipy_umbridge.server.serve_distribution(dist)
+

models/cuqipy-model/Dockerfile

@@ -0,0 +1,9 @@
+FROM ubuntu:latest
+
+COPY model-server.py /
+
+RUN apt update && \
+    DEBIAN_FRONTEND="noninteractive" apt install -y python3-pip && \
+    pip install cuqipy-umbridge
+
+CMD python3 model-server.py

models/cuqipy-model/model-server.py

@@ -0,0 +1,14 @@
+import cuqi
+import cuqipy_umbridge
+
+# This is the forward model to serve
+# This can be changed to any other CUQIpy forward model
+# Current example is the forward model of a 1D deconvolution test problem
+
+model = cuqi.testproblem.Deconvolution1D().model
+
+# Serve the distribution as UM-Bridge model
+# Name will be name of CUQIpy class for distribution
+# In this case "LinearModel"
+cuqipy_umbridge.server.serve_model(model)
+