Taylor mode automatic differentiation
Axect opened this issue · comments
Peroxide
has two AD methods - Dual
, HyperDual
. But these are only 1st and 2nd order methods, and they are distinct. So, we need not only higher order AD but also generic.
- Struct
AD
-
fmt::Display
-
Copy
? -
From<T>
-
-
std::ops
forAD
-
Add
,Sub
,Neg
-
Mul
-
Div
-
-
PowOps
forAD
-
powi
-
powf
-
pow
-
-
TrigOps
forAD
-
sin
,cos
,tan
-
asin
,acos
,atan
-
sinh
,cosh
,tanh
-
asinh
,acosh
,atanh
-
sin_cos
,sinh_cosh
-
-
ExpLogOps
forAD
-
exp
,ln
-
log
,log2
,log10
-
-
std::ops
withf64
-
Add
,Sub
-
Mul
,Div
-
- Change the structure -
proc_macro
toEnum
(to reduce compile time) - Generic AD function -
ADFn
- Apply to numerical things
-
numerical/root
-
numerical/ode
-
numerical/optimize
-
numerical/utils
-
Hi, am interested in your library. Can you make some todo tasks so we can we pick that if we want to contribute?
I think creating tasks on the project board will be good so that you can assign to a person who is doing https://github.com/Axect/Peroxide/projects
@nixar0x0 You could pick up anything from https://github.com/Axect/Peroxide/blob/master/TODO.md and have your pull request reviewed almost right away (judging from https://github.com/Axect/Peroxide/pulls?q=is%3Apr+is%3Aclosed ).
@nixar0x0 Thank you for interests and @tbmreza thank you for good advice.
I'll update TODO.md more specifically soon.