diff options
author | Edoardo Pasca <edo.paskino@gmail.com> | 2019-07-08 09:58:21 +0100 |
---|---|---|
committer | Edoardo Pasca <edo.paskino@gmail.com> | 2019-07-08 09:58:21 +0100 |
commit | d9c125f8a32fdb9bc6d705e9cd076ac3b1456bc5 (patch) | |
tree | b4dfc9f6b9a788d8cf1a6b069933b7f1f7996367 /README.md | |
parent | 99fe2a62f23dde3eb1ffb6d29a7b0bea410ef88c (diff) | |
download | framework-d9c125f8a32fdb9bc6d705e9cd076ac3b1456bc5.tar.gz framework-d9c125f8a32fdb9bc6d705e9cd076ac3b1456bc5.tar.bz2 framework-d9c125f8a32fdb9bc6d705e9cd076ac3b1456bc5.tar.xz framework-d9c125f8a32fdb9bc6d705e9cd076ac3b1456bc5.zip |
added min info on Block Framework
Diffstat (limited to 'README.md')
-rw-r--r-- | README.md | 13 |
1 files changed, 13 insertions, 0 deletions
@@ -120,6 +120,19 @@ In `ccpi.framework` we define a number of common classes normally used in tomogr The `Algorithm` provides the infrastructure to continue iteration, to access the values of the objective function in subsequent iterations, the time for each iteration. + ### Block Framework + +Block Framework is a generic strategy to treat variational problems in the following form: `\min \text{Regulariser} + \text{Fidelity} ` +where Regulariser and Fidelity are convex functions. While Fidelity is differentiable Regulariser may not. + +We have now a number of algorithms that can address this: + + * GradientDescent + * CGLS + * SIRT + * FISTA + * PDHG + #### Examples Please see the [demos](https://github.com/vais-ral/CIL-Demos) for examples of defining and using operators, functions and algorithms |