Mathieu Blondel
Mathieu Blondel
Currently, `ScipyMinimize` doesn't expose a `maxiter` option like all other solvers in JAXopt. Instead, it's part of the `options` dict.
Currently, the interface for QPs is like this: ```python params_obj = (Q, c) params_eq = (A, b) params_ineq = (G, h) osqp = jaxopt.OSQP() osqp.run(init, params_obj, params_eq, params_ineq) ``` I...
Some of our examples in the narrative documentation are not completely runnable (e.g., missing data or initialization). Also, examples are not tested as part of continuous integration, which makes it...
In the never-ending list of nice algorithms to add, there is [SR1](https://en.wikipedia.org/wiki/Symmetric_rank-one).
Some methods, such as line-search based methods, require more function / gradient evaluations than others. It would be great to keep track of the number of such calls in the...
With some solvers (e.g. Newton's method), it is possible to reuse some of the computations for solving the implicit differentiation linear system more efficiently. Since `@custom_root` and `@custom_fixed_point` accept a...