I've opened a ticket in PowerSeries.jl to track discussion of merging these
libraries:
https://github.com/jwmerrill/PowerSeries.jl/issues/7
Luis, I would love to continue this discussion there.
On Wednesday, March 26, 2014 8:10:48 PM UTC-7, Luis Benet wrote:
>
> Hi Jason,
>
>
>> It looks like t
Hi Jason,
> It looks like the scope of this is very similar to PowerSeries.jl. We
> should do some benchmarking and API bikeshedding, and think about combining
> efforts.
>
> Are there places your new package is obviously superior? I see
> L'Hospital's rule for division: that's a nice idea.
>
On Wednesday, March 26, 2014 12:47:47 PM UTC-7, Luis Benet wrote:
>
> Hi,
>
> I have followed the discussion, and would like to announce a package that
> we are developing:
> https://github.com/lbenet/TaylorSeries.jl
> since it does a bunch of things discussed above, though so far for 1
> indepe
Hi,
I have followed the discussion, and would like to announce a package that
we are developing:
https://github.com/lbenet/TaylorSeries.jl
since it does a bunch of things discussed above, though so far for 1
independent variable. It expands
different functions (so far a limited number of basic
>
> But once you have such a unified interface, why call the result
> Derivatives? What’s left for Calculus except the content of Derivatives?
> Will Calculus just be a symbolic calculus system?
>
I don't have a strong opinion on the naming, but it seemed reasonable.
Calculus also has simplificati
I think that kind of splitting of functionality would out well. I made this
change because I wanted to have a unified interface as soon as possible and the
potential cyclical dependency between DualNumbers and Calculus2 was something I
needed to avoid.
But once you have such a unified interface
On Sunday, March 23, 2014 11:29:37 PM UTC-7, Miles Lubin wrote:
>
> I like the unified interface for differentiation. My only qualm is with
> flattening out the hierarchy of packages by including DualNumbers. There
> was some discussion on this here:
> https://github.com/johnmyleswhite/Calculus.
I like the unified interface for differentiation. My only qualm is with
flattening out the hierarchy of packages by including DualNumbers. There
was some discussion on this here:
https://github.com/johnmyleswhite/Calculus.jl/pull/44
To start out, what is the intended scope of Calculus.jl? Once
I’d be happy to remove “xp” from the derivative rules.
— John
On Mar 23, 2014, at 6:49 PM, Tony Kelman wrote:
> While you're considering API changes, what would you think about getting rid
> of the "xp" from derivative_rules here
> https://github.com/johnmyleswhite/Calculus2.jl/blob/master/s
While you're considering API changes, what would you think about getting
rid of the "xp" from derivative_rules
here
https://github.com/johnmyleswhite/Calculus2.jl/blob/master/src/symbolic/gradexpr.jl#L121-L186
?
Derivative calculations that use xp can trivially multiply by it without it
bein
Here is a small comparison of some differentiation methods, available in
Julia, from the pre-Calculus2 era. It might still be interesting. I wrote
the 'NumericalMath' package because finite differences and the
"complex-step approach" were not exported from Calculus. I implemented both
"complex-
PowerSeries definitely gets you easy 2nd (and higher) derivatives in 1
dimension.
In higher dimensions, you can get 2nd order directional derivatives, so
it's really easy to fill out the diagonal of the hessian. It's possible to
fill out the off diagonals (i.e. the cross terms like d/dx d/dy) b
Thanks, Jason. Working on this made me realize that I need to dig into
PowerSeries more. Am I right in thinking that PowerSeries would provide a tool
for doing the equivalent of forward AD for hessians?
— John
On Mar 23, 2014, at 10:56 AM, Jason Merrill wrote:
> This looks great. Really nice
This looks great. Really nice work.
On Sunday, March 23, 2014 10:28:03 AM UTC-7, John Myles White wrote:
>
> I recently had a project where I made heavy use of the Calculus package
> and found that the interface has really fallen out of sync with the
> functionality that’s available. I also foun
14 matches
Mail list logo