issues
search
patrick-kidger
/
optimistix
Nonlinear optimisation (root-finding, least squares, ...) in JAX+Equinox. https://docs.kidger.site/optimistix/
Apache License 2.0
265
stars
12
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Parallel multi start
#67
MaAl13
opened
1 week ago
5
Question: errorhandling, BFGS minimization, vmap, and best practices
#66
HansN87
closed
3 weeks ago
2
Fix JAX deprecation warning
#65
patrick-kidger
closed
3 weeks ago
0
Draft:: Implement box constraints for the Nelder Mead solver, to increase its robustness on messy data where bounds are known.
#64
johannahaffner
opened
3 weeks ago
13
Can't vmap across input using Gauss Newton fwd
#63
packquickly
opened
1 month ago
11
Slow compile of least_squares with large dict parameters
#62
MaximilianJHuber
closed
1 month ago
2
Switched from reverse mode to forward mode where possible.
#61
patrick-kidger
opened
1 month ago
8
Unexpected behaviour with JAX version
#60
itk22
closed
1 month ago
3
Possibly of interest
#59
nstarman
opened
1 month ago
1
v0.0.7
#58
patrick-kidger
closed
1 month ago
0
Incompatibility with jax 0.4.27
#57
tjltjl
opened
1 month ago
1
Using `optimistix` with an `equinox` model
#56
frostedoyster
opened
1 month ago
2
Would an exhaustive grid search have a place in `optimistix`?
#55
mjo22
closed
3 months ago
2
Zero implicit gradients when using `ImplicitAdjoint` with CG solver
#54
itk22
opened
3 months ago
4
Fix some complex-related issues
#53
Randl
closed
3 months ago
3
Extracting intermediate function values/ losses from the solve
#52
itk22
opened
3 months ago
4
AbstractGaussNewton now supports reverse-autodiff for Jacobians.
#51
patrick-kidger
closed
2 months ago
14
Incompatibility of least_squares and custom_vjp
#50
ahwillia
opened
3 months ago
2
Adds a check in order to prevent terminating on first step for functions with initial condition yielding 0 in function evaluation. Corresponding tests.
#49
NightWinkle
closed
3 months ago
1
pytree output structure mismatch error in backprop during vmap
#48
FFroehlich
closed
2 months ago
9
Behavior of BFGS
#47
NightWinkle
opened
4 months ago
2
added citation to readme
#46
packquickly
closed
4 months ago
0
Will constrained optimization be supported?
#45
zhouxian
opened
4 months ago
4
Non-finite values in the root function are not handled well
#44
FFroehlich
closed
4 months ago
2
Fixed compatibility with no-jit. See https://github.com/patrick-kidger/diffrax/issues/368
#43
patrick-kidger
closed
4 months ago
0
Classical newton methods
#42
ataras2
closed
4 months ago
6
`BestSoFar...` wanted behavior ?
#41
thibmonsel
closed
4 months ago
1
Removed copyright notice from the head of each file
#40
packquickly
closed
5 months ago
0
grad of vmap of function which wraps an optax solver occasionally fails
#39
TomClarkMassSpec
closed
5 months ago
2
Issue with vmap `optx.least_squares`.
#38
bheijden
closed
4 months ago
2
Error in "optimistix/docs/examples /optimise_diffeq.ipynb"
#37
KhayrullevJokhongir
opened
5 months ago
1
Now using strict=True
#36
patrick-kidger
closed
5 months ago
0
correct name of the exception class that Equinox uses for runtime errors
#35
TomClarkMassSpec
closed
5 months ago
1
Check last point when using best so far minimiser
#34
ColCarroll
opened
6 months ago
1
BestSoFarMinimiser behavior
#33
ColCarroll
opened
6 months ago
1
Fix small typo in index.md
#32
ColCarroll
closed
6 months ago
1
Now using strict rank and dtype promotions when testing.
#31
patrick-kidger
closed
6 months ago
0
Now using the same jaxpr in the state.
#30
patrick-kidger
closed
6 months ago
0
improved OptaxMinimiser docs
#29
packquickly
closed
6 months ago
1
dev
#28
patrick-kidger
closed
6 months ago
0
Fixed {Newton,Chord}(cauchy_termination=False) failing when started close to the solution
#27
patrick-kidger
closed
6 months ago
0
Can't use Optimistix solvers with `eqx.Module`s and filtered transformations
#26
eringrant
closed
7 months ago
2
Pytree inputs for `rtol` and `atol` or custom termination condition?
#25
LouisDesdoigts
closed
7 months ago
7
Efficient NewtonCG Implementation
#24
quattro
opened
7 months ago
3
Compat with Lineax successful message chnage
#23
patrick-kidger
closed
4 months ago
0
Removed usage of a private lineax stub.
#22
patrick-kidger
closed
8 months ago
0
Using OptaxMinimiser results in AttributeError
#21
hkortier
closed
6 months ago
5
Qs: Mapping different solvers to leaves, parameter normalisation, different parameter scales
#20
LouisDesdoigts
closed
8 months ago
2
State how to use `options` more clearly.
#19
patrick-kidger
closed
8 months ago
0
API is not clear to me about notational entry of options: upper: and lower: for root finding optimistix.Bisection/Newton
#18
TomClarkMassSpec
closed
8 months ago
3
Next