v1.2.0

Breaking changes

• z_s is now a param of lens objects (rather than an argument to every function). Should generally be treated like z_l. A new tutorial has been added to show how to do multi source modelling
• Several params have been renamed to normalize conventions. Various b and th_ein parameters have now been changed to Rein, and for EPL this involved changing b by sqrt(q). Others include m -> mass and scale_radius -> Rs etc.

These changes will likely affect all caustics codes, but are very easy/quick to adjust for.

Big positives

• NFW and TNFW no longer require specifying batchable or differentiable they are now capable of both simultaneously. Also TNFW potential has been fixed.
• Forward raytrace had a bug where it was possible to return multiple copies of the same coordinates, that is now fixed
• The jacobian of the lens equation (used for various calculations) is now far more memory efficient

What's Changed

• refactor: normalize variable names across caustics objects by @ConnorStoneAstro in #302
• refactor: Change z_s from argument to caskade param by @ConnorStoneAstro in #305
• Guassian to torch by @ConnorStoneAstro in #288
• fix: ensure the forward raytrace points are unique by @ConnorStoneAstro in #315
• refactor: better use of autograd for base lens quantities by @ConnorStoneAstro in #306
• fix: handle q=1 for sie by @ConnorStoneAstro in #301
• feat: New parametrizations for external shear and point mass by @ConnorStoneAstro in #304
• docs: update docstrings remove z_s and params by @ConnorStoneAstro in #317
• chore: update pre-commit hooks by @pre-commit-ci in #310
• chore: update pre-commit hooks by @pre-commit-ci in #312

Full Changelog: v1.1.0...v1.2.0