Skip to content

Releases: ddrous/Updes

Version 1.0.2

28 May 08:33
0ee5c69
Compare
Choose a tag to compare

Exciting upgrades coming with version 1.0.2 🎉

  • Faster linear solves with Lineax
  • Support for both vals and coeffs as fields inputs to differential and rhs operators
  • Fixed the args inputs to construct the local matrix for the nodal_div_grad: (if array, else, etc.)
  • Added colorbar to animate fields
  • Implemented the Darcy flow problem in demos

Version 1.0.1

18 Apr 13:25
Compare
Choose a tag to compare

This release includes:

  • A patch to fix the problem that occurred due to non-staking of the field arguments when passed to the differentiable operators
  • Separate readmes for PyPI and GitHub
  • Nicer documentation with 3 pages and a longer tutorial

Version 1.0.0

14 Apr 22:44
Compare
Choose a tag to compare

Finally, Updes is available in its first official release. This release comes with:

  • A way to simulate and control PDEs with ease
  • A documentation with QuartoDoc
  • Detailed readme and reference paper
  • Further example PDEs in the demos
  • Etc.

We are officially accepting issues and pull-requests 🎉

Version 0.1.4

04 Apr 20:06
Compare
Choose a tag to compare

Major pre-release before first official release !

Version 0.1.3

19 Mar 20:50
Compare
Choose a tag to compare
Version 0.1.3 Pre-release
Pre-release

Miles stone reached.
Periodic boundary conditions !

Updec v0.1.2

16 Jul 23:52
Compare
Choose a tag to compare
Updec v0.1.2 Pre-release
Pre-release

This release focuses on demos for the Navier-Stokes test case:

  • Derivation of various schemes for the NS equations
  • Training PINNs and collecting performance metrics
  • Benchmarking CPU times and peak memory usage
  • Etc.

Updec v0.1.1

19 Jun 10:59
Compare
Choose a tag to compare
Updec v0.1.1 Pre-release
Pre-release

This release focuses on providing accurate examples on how to use Updec:

  • Formulation of the adjoint equations for Laplace
  • 3 different formulations of the adjoint equations for Navier-Stokes (the best has a negated loss)
  • Find appropriate hyperparams for differentiable solvers on Navier-Stokes for better memory efficiency
  • Implements finite difference in lieu of the differentiable solver on Navier-Stokes
  • Integrated Adam optimiser to DAL and DP gradient descents
  • Improved plotting, small bugs, etc.

Version 0.1.0

11 Feb 03:57
Compare
Choose a tag to compare
Version 0.1.0 Pre-release
Pre-release

Very first release. The code is working smoothly, but isn't vectorized. Major features include:

  • PDE solving with RBF e.g. The Laplace equation with Neumann boundary conditions with super-linear convergence
  • Load cloud points from Gmsh
  • Implementation of RBF-PINNs (not stable)

N.B. The current API is not stable as the code is under active development.