Add reverse for vectors.

[ghost]

Submisison Checklist

• Run unit tests: ./runTests.py test/unit
• Run cpplint: make cpplint
• Declare copyright holder and open-source license: see below

Summary:

Just a quick resolution of #344 adding a reverse utility for std::vectors. Like our sort functions, it mutates its argument. We ok with that?

Intended Effect:

Add a missing minor feature.

How to Verify:

Code is short and there's a unit test.

Side Effects:

None, adds a function.

Documentation:

doxygen

Reviewer Suggestions:

Copyright and Licensing

Please list the copyright holder for the work you are submitting (this will be you or your assignee, such as a university or company): Columbia University

By submitting this pull request, the copyright holder is agreeing to license the submitted work under the following licenses:

• Code: BSD 3-clause (https://opensource.org/licenses/BSD-3-Clause)
• Documentation: CC-BY 4.0 (https://creativecommons.org/licenses/by/4.0/)

[bob-carpenter]

Three things:

1. It needs to be a constant reference, which is how arguments are passed in from the Stan language. So it needs a copy.

2. It also needs tests for reverse, forward, and mixed modes. That'll just test that the appropriate comparisons are defined for std::reverse to apply to the iterators.

3. It should be implemented for row vectors and column vector types, too. If you take an ordered vector, it's an Eigen vector, and that's the kind of thing we'll have users reversing.

[ghost]

@bob-carpenter I hopefully fixed the const issue and am proceeding to the rest, I'm curious whether there are any good examples of of functions from vector to vector already in the language? I'm looking at fun/sum (in prim, rev, fwd), the autodiff paper, and a few other places but if there's an idiom it'd be nice to match

Also: I'm intending to put a little bit of noise on this pull request. Is there a nice way to avoid triggering CI builds on intermediate pushes?

[syclik]

In your commit, you should be able to include "[skip ci]", but I don't know how well that actually works as documented. Both only build on push, so you can prevent constant builds if you push when it's actually ready to test. Although, you really shouldn't worry about the builds that are fired off. Most of the time we're idle.

…

On Jan 26, 2017, at 6:10 PM, thelseraphim ***@***.***> wrote: @bob-carpenter I hopefully fixed the const issue and am proceeding to the rest, I'm curious whether there are any good examples of of functions from vector to vector already in the language? I'm looking at fun/sum (in prim, rev, fwd), the autodiff paper, and a few other places but if there's an idiom it'd be nice to match Also: I'm intending to put a little bit of noise on this pull request. Is there a nice way to avoid triggering CI builds on intermediate pushes? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

[jgabry]

[ci skip] in a commit message has always worked as advertised for me. It definitely comes in handy. On Thu, Jan 26, 2017 at 11:05 PM Daniel Lee <notifications@github.com> wrote:

…

In your commit, you should be able to include "[skip ci]", but I don't know how well that actually works as documented. Both only build on push, so you can prevent constant builds if you push when it's actually ready to test. Although, you really shouldn't worry about the builds that are fired off. Most of the time we're idle. > On Jan 26, 2017, at 6:10 PM, thelseraphim ***@***.***> wrote: > > @bob-carpenter I hopefully fixed the const issue and am proceeding to the rest, I'm curious whether there are any good examples of of functions from vector to vector already in the language? I'm looking at fun/sum (in prim, rev, fwd), the autodiff paper, and a few other places but if there's an idiom it'd be nice to match > > Also: I'm intending to put a little bit of noise on this pull request. Is there a nice way to avoid triggering CI builds on intermediate pushes? > > — > You are receiving this because you are subscribed to this thread. > Reply to this email directly, view it on GitHub, or mute the thread. > — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#479 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AHb4Q2MJ5709ZIiKziA7rNFNWjst0M0pks5rWWzxgaJpZM4Ls-lr> .

[seantalts]

I've just been killing off the builds that I know I've triggered if I have updates to the PR that makes running those tests obsolete.

[bob-carpenter]

You also don't need to create a pull request until it's ready to go. We can review branches until then. But sometimes it's easier with pull requests. - Bob

…

On Jan 26, 2017, at 11:05 PM, Daniel Lee ***@***.***> wrote: In your commit, you should be able to include "[skip ci]", but I don't know how well that actually works as documented. Both only build on push, so you can prevent constant builds if you push when it's actually ready to test. Although, you really shouldn't worry about the builds that are fired off. Most of the time we're idle. > On Jan 26, 2017, at 6:10 PM, thelseraphim ***@***.***> wrote: > > @bob-carpenter I hopefully fixed the const issue and am proceeding to the rest, I'm curious whether there are any good examples of of functions from vector to vector already in the language? I'm looking at fun/sum (in prim, rev, fwd), the autodiff paper, and a few other places but if there's an idiom it'd be nice to match > > Also: I'm intending to put a little bit of noise on this pull request. Is there a nice way to avoid triggering CI builds on intermediate pushes? > > — > You are receiving this because you are subscribed to this thread. > Reply to this email directly, view it on GitHub, or mute the thread. > — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

[bob-carpenter]

You can find the signatures in the math library functions in the index of the manual. So you're looking for something that returns real[] or vector and takes a real[] or vector argument. For example, cumulative_sum. But cumulative_sum doesn't have specialized derivatives, so you're better off looking at stan/math/rev/mat/fun/softmax.hpp Also, we have an array allocator in memalloc_ that should've been used in softmax. You could replace that, too vari** alpha_vi_array = reinterpret_cast<vari**>(ChainableStack::memalloc_ .alloc(sizeof(vari*) * alpha.size())); should be vari** alpha_vi_array = ChainableStack::memalloc_.alloc_array<vari*>(alpha.size()); At least I think I have my types in a row here. The memaalloc_ object with alloc_array method is defined in stan/math/memory/stack_alloc.hpp. - Bob

…

On Jan 26, 2017, at 6:10 PM, thelseraphim ***@***.***> wrote: @bob-carpenter I hopefully fixed the const issue and am proceeding to the rest, I'm curious whether there are any good examples of of functions from vector to vector already in the language? I'm looking at fun/sum (in prim, rev, fwd), the autodiff paper, and a few other places but if there's an idiom it'd be nice to match Also: I'm intending to put a little bit of noise on this pull request. Is there a nice way to avoid triggering CI builds on intermediate pushes? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

[bob-carpenter]

jenkins, retest this please

[bob-carpenter]

I shouldn't have just kicked off retesting. Looks like the reverse file isn't including the header for std::begin and std::end.

[ghost]

@bob-carpenter it wasn't using it because I shouldn't have had it there. pushing a micro-change this time just to see the [ci skip] in action

[bob-carpenter]

The reverse function looks good now. But the pull request still needs rev and fwd autodiff tests just to make sure that everything needed by std::reverse_copy is available for the autodiff classes.

[bob-carpenter]

Was this all ready for final review?

[bob-carpenter]

@thelseraphim Would you mind adding the autodiff tests for this? Just one for fwd, one for rev, and one for rev nested in fwd (three test files in rev, fwd, and mix directories).

[syclik]

@thelseraphim mind finishing out this pull request?