Setup documentation

.gitignore

@@ -1 +1,2 @@
 Manifest.toml
+docs/build/

.travis.yml

@@ -8,31 +8,31 @@ julia:
   - 1.0
   - 1.1
   - nightly
- 
+
 matrix:
   allow_failures:
   - julia: nightly
- 
+
 notifications:
   email:
     recipients:
       - [email protected]
     on_success: never
     on_failure: always
-    
-    
+
+
 cache:
   directories:
     - /home/travis/.julia/packages/MPI/
-    
-#MPI.jl needs mpi binaries plus compilers avalible 
+
+#MPI.jl needs mpi binaries plus compilers avalible
 addons:
   apt:
     packages:
       - gfortran
       - mpich
       - libmpich-dev
-      
+
 before_script:
   - |
     if [[ "$TRAVIS_OS_NAME" == "osx" ]]; then
@@ -41,7 +41,20 @@ before_script:
       brew cask uninstall oclint # Prevent conflict with gcc
       brew install mpich
     fi
-    
+
 before_install:
   - export CC=mpicc
   - export FC=mpif90
+
+
+# generate documentation
+jobs:
+  include:
+    - stage: "Documentation"
+      julia: 1.1
+      os: linux
+      script:
+        - julia --project=docs/ -e 'using Pkg; Pkg.develop(PackageSpec(path=pwd()));
+                                               Pkg.instantiate()'
+        - julia --project=docs/ docs/make.jl
+      after_success: skip

README.md

@@ -1,6 +1,8 @@
 # JuliaPetra
 
 [![Build Status](https://travis-ci.org/Collegeville/JuliaPetra.jl.svg?branch=master)](https://travis-ci.org/Collegeville/JuliaPetra.jl)
+[![Stable Docs](https://img.shields.io/badge/docs-stable-blue.svg)](https://Collegeville.github.io/JuliaPetra.jl/stable/)
+[![Latest Docs](https://img.shields.io/badge/docs-latest-blue.svg)](https://Collegeville.github.io/JuliaPetra.jl/latest/)
 
 This is an implmentation of [Trilinos's Petra Object Model](https://trilinos.github.io/data_service.html#trilinos-packages) in Julia.
 It is a basic framework for sparse linear algebra.
@@ -12,4 +14,4 @@ JuliaPetra is licensed under the MIT "Expat" license. See LICENSE.md for more in
 ## Setup
 
 JuliaPetra is not yet a registered package.
-So to install it, run `add https://github.com/collegeville/JuliaPetra` from the Pkg REPL. 
+So to install it, run `add https://github.com/collegeville/JuliaPetra` from the Pkg REPL.

docs/Project.toml

@@ -0,0 +1,5 @@
+[deps]
+Documenter = "e30172f5-a6a5-5a46-863b-614d45cd2de4"
+
+[compat]
+Documenter = "~0.21"

docs/make.jl

@@ -0,0 +1,16 @@
+using Documenter, JuliaPetra
+
+makedocs(
+    modules = [JuliaPetra],
+    sitename="JuliaPetra",
+    pages = [
+        "Home" => "index.md",
+        "Communcation Layer" => "CommunicationLayer.md",
+        "Problem Distribution Layer" => "ProblemDistributionLayer.md",
+        "Linear Algebra Layer" => "LinearAlgebraLayer.md"
+    ])
+
+
+deploydocs(
+    repo = "github.com/collegeville/JuliaPetra.jl.git"
+)

docs/src/CommunicationLayer.md

@@ -0,0 +1,50 @@
+# Communications Layer
+
+```@meta
+CurrentModule = JuliaPetra
+```
+
+JuliaPetra abstracts communication with the [`Comm`](@ref) and [`Distributor`](@ref) interfaces.
+There are two communication implementations with JuliaPetra [`SerialComm`](@ref) and [`MPIComm`](@ref).
+Note that most objects dependent on inter-process communication support the [`getComm`](@ref) method.
+
+## Interface
+
+```@docs
+Comm
+Distributor
+```
+
+### Functions
+
+```@docs
+getComm
+barrier
+broadcastAll
+gatherAll
+sumAll
+maxAll
+minAll
+scanSum
+myPid
+numProc
+createDistributor
+createFromSends
+createFromRecvs
+resolve
+resolvePosts
+resolveWaits
+resolveReverse
+resolveReversePosts
+resolveReverseWaits
+```
+
+## Implementations
+
+```@docs
+LocalComm
+SerialComm
+SerialDistributor
+MPIComm
+MPIDistributor
+```

docs/src/LinearAlgebraLayer.md

@@ -0,0 +1,49 @@
+# Linear Algebra Layer
+
+```@meta
+CurrentModule = JuliaPetra
+```
+
+The Linear Algebra layer provides the main abstractions for linear algebra codes.
+The two top level interfaces are [`MultiVector`](@ref), for groups of vectors, and [`Operator`](@ref), for operations on `MultiVector`s.
+
+## MultiVectors
+
+MutliVectors support many basic array functions, including broadcasting.
+Additionally, [`dot`] and [`norm`] are supported, however they return arrays since [`MultiVector`]s may have multiple dot products and norms.
+
+```@docs
+MultiVector
+DenseMultiVector
+localLength
+globalLength
+numVectors
+getVectorView
+getVectorCopy
+getLocalArray
+commReduce
+```
+
+## Operators
+
+`Operator`s represent an operation on a [`MultiVector`](@ref), such as a matrix which applies a matrix-vector product.
+
+```@docs
+Operator
+getRangeMap
+getDomainMap
+apply!
+apply
+TransposeMode
+isTransposed
+applyConjugation
+```
+
+### Matrices
+
+Sparse matrices are the primary [`Operator`](@ref) in JuliaPetra.
+
+```@docs
+RowMatrix
+CSRMatrix
+```

docs/src/ProblemDistributionLayer.md

@@ -0,0 +1,77 @@
+# Problem Distribution Layer
+
+```@meta
+CurrentModule = JuliaPetra
+```
+
+The Problem Distribution Layer managers how the problem is distributed across processes.
+The main type is [`BlockMap`](@ref) which represents a problem distribution.
+
+## BlockMap
+
+```@docs
+BlockMap
+lid
+gid
+myLID
+myGID
+remoteIDList
+minAllGID
+maxAllGID
+minMyGID
+maxMyGID
+minLID
+maxLID
+numGlobalElements
+numMyElements
+myGlobalElements
+myGlobalElementIDs
+uniqueGIDs
+sameBlockMapDataAs
+sameAs
+globalIndicesType
+linearMap
+distributedGlobal
+```
+
+## Directory
+
+```@docs
+Directory
+BasicDirectory
+getDirectoryEntries
+gidsAllUniquelyOwned
+createDirectory
+```
+
+## Converting IDs Between Maps
+
+```@docs
+Export
+Import
+sourceMap
+targetMap
+distributor
+isLocallyComplete
+permuteToLIDs
+permuteFromLIDs
+exportLIDs
+remoteLIDs
+remotePIDs
+numSameIDs
+```
+
+## Converting Data Structures Between Maps
+
+Converting data structures between maps is built on the [`DistObject`](@ref) and [`SrcDistObject`](@ref) interfaces.
+
+```@docs
+DistObject
+SrcDistObject
+CombineMode
+getMap
+checkSizes
+copyAndPermute
+packAndPrepare
+unpackAndCombine
+```

docs/src/index.md

@@ -0,0 +1,11 @@
+# JuliaPetra
+
+JuliaPetra is an implmentation of [Trilinos's Petra Object Model](https://trilinos.github.io/data_service.html#trilinos-packages) in Julia.
+It is a basic framework for distributed sparse linear algebra.
+Note that JuliaPetra uses Single Program Multiple Data parallelism instead of the master/worker parallelism often used in Julia.
+
+# Organization
+JuliaPetra is organized into a series of layers.
+* The [Communications Layer](@ref) contains an interface for Single Program Multiple Data parallel systems
+* The [Problem Distribution Layer](@ref) manages how the problem is distributed across the processes
+* The [Linear Algebra Layer](@ref) provides the interfaces and implementations for linear algebra objects

src/BlockMap.jl

@@ -1,9 +1,9 @@
 export BlockMap
-export remoteIDList, lid, gid, findLocalElementID
+export remoteIDList, lid, gid
 export minAllGID, maxAllGID, minMyGID, maxMyGID, minLID, maxLID
 export numGlobalElements, myGlobalElements
 export uniqueGIDs, globalIndicesType, sameBlockMapDataAs, sameAs
-export linearMap, myGlobalElementIDs, comm
+export linearMap, myGlobalElementIDs
 export myGID, myLID, distributedGlobal, numMyElements
 
 

src/CSRMatrix.jl

@@ -2,6 +2,9 @@ export CSRMatrix, insertGlobalValues
 
 using TypeStability
 
+"""
+An implementation of [`RowMatrix`](@ref) that uses CSR format
+"""
 mutable struct CSRMatrix{Data <: Number, GID <: Integer, PID <: Integer, LID <: Integer} <: RowMatrix{Data, GID, PID, LID}
     rowMap::BlockMap{GID, PID, LID}
     colMap::Union{BlockMap{GID, PID, LID}, Nothing}

src/Comm.jl

@@ -121,7 +121,7 @@ scanSum(comm::Comm, val) = scanSum(comm, [val])[1]
 """
     getComm(obj)
 
-Gets the Comm for the object, if aplicable
+Gets the Comm for the object, if applicable
 """
 getComm(comm::Comm) = comm
 

src/Directory.jl

@@ -3,7 +3,7 @@ export Directory, getDirectoryEntries, gidsAllUniquelyOwned
 # methods and docs based straight off Epetra_Directory to match Comm
 
 """
-A base type as an interface to allow Map and BlockMap objects to reference non-local
+A base type as an interface to allow BlockMap objects to reference non-local
 elements.
 
 All subtypes must have the following methods, with DirectoryImpl standing in for