This repository contains the model AdjointTurbidity 1.0, described in the paper
Samuel D. Parkinson, Simon W. Funke, Jon Hill, Matthew D. Piggott, and Peter A. Allison, Application of the adjoint approach to optimise the initial conditions of a turbidity current, Geoscientific Model Development (GMD).
- Install docker
- Build docker container with
docker build -t turbidity . - Start docker container with
docker run -i -t turbidity
1. Install FEniCS 1.3 from http://fenicsproject.org/
2. Install libadjoint from https://bitbucket.org/dolfin-adjoint/libadjoint. Check out commit 3a99e45533d1ecb3a828c8072d22067d0a66d5b0.
3. Install dolfin-adjoint from https://bitbucket.org/dolfin-adjoint/dolfin-adjoint. Check out commit d9d3d28632c56e9a68b48f73e2a622a5077b918c.
4. Install the most recent version of libspud from the Fluidity project https://github.com/FluidityProject/fluidity
5. Install pyipopt from https://github.com/xuy/pyipopt.
5. Clone this repository and include its root path to $PYTHONPATH
- Section 2.5 Forward model verification Reproduce with:
#!bash
$ cd tests/similarity
$ python similarity.py
- Section 4.3 Verification of the gradient calculation
#!bash
Reproduce with:
$ cd tests/taylor
$ python taylor.py
- Section 4.4 and 4.6: Optimisation of a model with one/two sediment class(es) Reproduce with:
#!bash
$ cd jobs/bed_11
$ python bed_4_test.py "1"
$ ./MAKEFILE
$ python bed_4_test.py "1" opt_IPOPT_<last_run>.pckl
$ mv final.pckl final_end.pckl
$ python bed_4_test.py "1" opt_IPOPT_1.pckl
$ ./MAKEFILE2