Reference documentation for deal.II version 9.4.0

Namespaces  
namespace  ConsensusAlgorithms 
namespace  internal 
namespace  LargeCount 
Classes  
class  CollectiveMutex 
class  CommunicationPatternBase 
class  DuplicatedCommunicator 
struct  MinMaxAvg 
class  MPI_InitFinalize 
class  NoncontiguousPartitioner 
class  Partitioner 
class  ProcessGrid 
class  RemotePointEvaluation 
Functions  
unsigned int  n_mpi_processes (const MPI_Comm &mpi_communicator) 
unsigned int  this_mpi_process (const MPI_Comm &mpi_communicator) 
const std::vector< unsigned int >  mpi_processes_within_communicator (const MPI_Comm &comm_large, const MPI_Comm &comm_small) 
std::vector< unsigned int >  compute_point_to_point_communication_pattern (const MPI_Comm &mpi_comm, const std::vector< unsigned int > &destinations) 
unsigned int  compute_n_point_to_point_communications (const MPI_Comm &mpi_comm, const std::vector< unsigned int > &destinations) 
MPI_Comm  duplicate_communicator (const MPI_Comm &mpi_communicator) 
void  free_communicator (MPI_Comm &mpi_communicator) 
int  create_group (const MPI_Comm &comm, const MPI_Group &group, const int tag, MPI_Comm *new_comm) 
std::vector< IndexSet >  create_ascending_partitioning (const MPI_Comm &comm, const IndexSet::size_type locally_owned_size) 
IndexSet  create_evenly_distributed_partitioning (const MPI_Comm &comm, const IndexSet::size_type total_size) 
template<class Iterator , typename Number = long double>  
std::pair< Number, typename numbers::NumberTraits< Number >::real_type >  mean_and_standard_deviation (const Iterator begin, const Iterator end, const MPI_Comm &comm) 
std::unique_ptr< MPI_Datatype, void(*)(MPI_Datatype *)>  create_mpi_data_type_n_bytes (const std::size_t n_bytes) 
template<typename T >  
T  sum (const T &t, const MPI_Comm &mpi_communicator) 
template<typename T , typename U >  
void  sum (const T &values, const MPI_Comm &mpi_communicator, U &sums) 
template<typename T >  
void  sum (const ArrayView< const T > &values, const MPI_Comm &mpi_communicator, const ArrayView< T > &sums) 
template<int rank, int dim, typename Number >  
SymmetricTensor< rank, dim, Number >  sum (const SymmetricTensor< rank, dim, Number > &local, const MPI_Comm &mpi_communicator) 
template<int rank, int dim, typename Number >  
Tensor< rank, dim, Number >  sum (const Tensor< rank, dim, Number > &local, const MPI_Comm &mpi_communicator) 
template<typename Number >  
void  sum (const SparseMatrix< Number > &local, const MPI_Comm &mpi_communicator, SparseMatrix< Number > &global) 
template<typename T >  
T  max (const T &t, const MPI_Comm &mpi_communicator) 
template<typename T , typename U >  
void  max (const T &values, const MPI_Comm &mpi_communicator, U &maxima) 
template<typename T >  
void  max (const ArrayView< const T > &values, const MPI_Comm &mpi_communicator, const ArrayView< T > &maxima) 
template<typename T >  
T  min (const T &t, const MPI_Comm &mpi_communicator) 
template<typename T , typename U >  
void  min (const T &values, const MPI_Comm &mpi_communicator, U &minima) 
template<typename T >  
void  min (const ArrayView< const T > &values, const MPI_Comm &mpi_communicator, const ArrayView< T > &minima) 
template<typename T >  
T  logical_or (const T &t, const MPI_Comm &mpi_communicator) 
template<typename T , typename U >  
void  logical_or (const T &values, const MPI_Comm &mpi_communicator, U &results) 
template<typename T >  
void  logical_or (const ArrayView< const T > &values, const MPI_Comm &mpi_communicator, const ArrayView< T > &results) 
MinMaxAvg  min_max_avg (const double my_value, const MPI_Comm &mpi_communicator) 
std::vector< MinMaxAvg >  min_max_avg (const std::vector< double > &my_value, const MPI_Comm &mpi_communicator) 
void  min_max_avg (const ArrayView< const double > &my_values, const ArrayView< MinMaxAvg > &result, const MPI_Comm &mpi_communicator) 
bool  job_supports_mpi () 
template<typename T >  
std::map< unsigned int, T >  some_to_some (const MPI_Comm &comm, const std::map< unsigned int, T > &objects_to_send) 
template<typename T >  
std::vector< T >  all_gather (const MPI_Comm &comm, const T &object_to_send) 
template<typename T >  
std::vector< T >  gather (const MPI_Comm &comm, const T &object_to_send, const unsigned int root_process=0) 
template<typename T >  
std::enable_if< is_mpi_type< T >==false, T >::type  broadcast (const MPI_Comm &comm, const T &object_to_send, const unsigned int root_process=0) 
template<typename T >  
std::enable_if< is_mpi_type< T >==true, T >::type  broadcast (const MPI_Comm &comm, const T &object_to_send, const unsigned int root_process=0) 
template<typename T >  
void  broadcast (T *buffer, const size_t count, const unsigned int root, const MPI_Comm &comm) 
template<typename T >  
T  reduce (const T &local_value, const MPI_Comm &comm, const std::function< T(const T &, const T &)> &combiner, const unsigned int root_process=0) 
template<typename T >  
T  all_reduce (const T &local_value, const MPI_Comm &comm, const std::function< T(const T &, const T &)> &combiner) 
std::vector< unsigned int >  compute_index_owner (const IndexSet &owned_indices, const IndexSet &indices_to_look_up, const MPI_Comm &comm) 
template<typename T >  
std::vector< T >  compute_set_union (const std::vector< T > &vec, const MPI_Comm &comm) 
template<typename T >  
std::set< T >  compute_set_union (const std::set< T > &set, const MPI_Comm &comm) 
Variables  
template<typename T >  
constexpr bool  is_mpi_type 
template<typename T >  
const MPI_Datatype  mpi_type_id_for_type 
template const MPI_Datatype  mpi_type_id_for_type< bool > 
template const MPI_Datatype  mpi_type_id_for_type< char > 
template const MPI_Datatype  mpi_type_id_for_type< signed char > 
template const MPI_Datatype  mpi_type_id_for_type< short > 
template const MPI_Datatype  mpi_type_id_for_type< int > 
template const MPI_Datatype  mpi_type_id_for_type< long int > 
template const MPI_Datatype  mpi_type_id_for_type< unsigned char > 
template const MPI_Datatype  mpi_type_id_for_type< unsigned short > 
template const MPI_Datatype  mpi_type_id_for_type< unsigned long int > 
template const MPI_Datatype  mpi_type_id_for_type< unsigned long long int > 
template const MPI_Datatype  mpi_type_id_for_type< float > 
template const MPI_Datatype  mpi_type_id_for_type< double > 
template const MPI_Datatype  mpi_type_id_for_type< long double > 
A namespace for utility functions that abstract certain operations using the Message Passing Interface (MPI) or provide fallback operations in case deal.II is configured not to use MPI at all.
Return the number of MPI processes there exist in the given communicator object. If this is a sequential job (i.e., the program is not using MPI at all, or is using MPI but has been started with only one MPI process), then the communicator necessarily involves only one process and the function returns 1.
Return the rank of the present MPI process in the space of processes described by the given communicator. This will be a unique value for each process between zero and (less than) the number of all processes (given by get_n_mpi_processes()).
std::vector< unsigned int > Utilities::MPI::compute_point_to_point_communication_pattern  (  const MPI_Comm &  mpi_comm, 
const std::vector< unsigned int > &  destinations  
) 
Consider an unstructured communication pattern where every process in an MPI universe wants to send some data to a subset of the other processors. To do that, the other processors need to know who to expect messages from. This function computes this information.
mpi_comm  A communicator that describes the processors that are going to communicate with each other. 
destinations  The list of processors the current process wants to send information to. This list need not be sorted in any way. If it contains duplicate entries that means that multiple messages are intended for a given destination. 
unsigned int Utilities::MPI::compute_n_point_to_point_communications  (  const MPI_Comm &  mpi_comm, 
const std::vector< unsigned int > &  destinations  
) 
Simplified (for efficiency) version of the compute_point_to_point_communication_pattern() which only computes the number of processes in an MPI universe to expect communication from.
mpi_comm  A communicator that describes the processors that are going to communicate with each other. 
destinations  The list of processors the current process wants to send information to. This list need not be sorted in any way. If it contains duplicate entries that means that multiple messages are intended for a given destination. 
Given a communicator, generate a new communicator that contains the same set of processors but that has a different, unique identifier.
This functionality can be used to ensure that different objects, such as distributed matrices, each have unique communicators over which they can interact without interfering with each other.
When no longer needed, the communicator created here needs to be destroyed using free_communicator().
This function is equivalent to calling MPI_Comm_dup(mpi_communicator, &return_value);
.
void Utilities::MPI::free_communicator  (  MPI_Comm &  mpi_communicator  ) 
Free the given communicator mpi_communicator
that was duplicated using duplicate_communicator().
The argument is passed by reference and will be invalidated and set to the MPI null handle. This function is equivalent to calling MPI_Comm_free(&mpi_communicator);
.
int Utilities::MPI::create_group  (  const MPI_Comm &  comm, 
const MPI_Group &  group,  
const int  tag,  
MPI_Comm *  new_comm  
) 
If comm
is an intracommunicator, this function returns a new communicator newcomm
with communication group defined by the group
argument. The function is only collective over the group of processes that actually want to create the communicator, i.e., that are named in the group
argument. If multiple threads at a given process perform concurrent create_group() operations, the user must distinguish these operations by providing different tag
or comm
arguments.
This function was introduced in the MPI3.0 standard. If available, the corresponding function in the provided MPI implementation is used. Otherwise, the implementation follows the one described in the following publication:
std::vector< IndexSet > Utilities::MPI::create_ascending_partitioning  (  const MPI_Comm &  comm, 
const IndexSet::size_type  locally_owned_size  
) 
Given the number of locally owned elements locally_owned_size
, create a 1:1 partitioning of the of elements across the MPI communicator comm
. The total size of elements is the sum of locally_owned_size
across the MPI communicator. Each process will store contiguous subset of indices, and the index set on process p+1 starts at the index one larger than the last one stored on process p.
IndexSet Utilities::MPI::create_evenly_distributed_partitioning  (  const MPI_Comm &  comm, 
const IndexSet::size_type  total_size  
) 
Given the total number of elements total_size
, create an evenly distributed 1:1 partitioning of the elements across the MPI communicator comm
. Uses comm
to determine number of partitions and processor ID to call the create_evenly_distributed_partitioning()
function above.
std::pair< Number, typename numbers::NumberTraits< Number >::real_type > Utilities::MPI::mean_and_standard_deviation  (  const Iterator  begin, 
const Iterator  end,  
const MPI_Comm &  comm  
) 
Calculate mean and standard deviation across the MPI communicator comm
for values provided as a range [begin,end)
. The mean is computed as \(\bar x=\frac 1N \sum x_k\) where the \(x_k\) are the elements pointed to by the begin
and end
iterators on all processors (i.e., each processor's [begin,end)
range points to a subset of the overall number of elements). The standard deviation is calculated as \(\sigma=\sqrt{\frac {1}{N1} \sum x_k \bar x^2}\), which is known as unbiased sample variance.
Number  specifies the type to store the mean value. The standard deviation is stored as the corresponding real type. This allows, for example, to calculate statistics from integer input values. 
std::unique_ptr< MPI_Datatype, void(*)(MPI_Datatype *)> Utilities::MPI::create_mpi_data_type_n_bytes  (  const std::size_t  n_bytes  ) 
Create a object that contains an MPI_Datatype
that represents n_bytes
bytes.
The resulting data type can be used in MPI send/receive or MPI IO to process messages of sizes larger than 2 GB with MPI_Byte as the underlying data type. This helper is required for MPI versions before 4.0 because routines like MPI_Send
use a signed integer for the count
variable. Instead, you can use this data type with the appropriate size set to the size of your message and by passing 1 as the count
.
MPI_Datatype
because such objects need to be destroyed by a call to MPI_Type_free
and it is easy to forget to do so (thereby creating a resource leak). Rather, the function returns an object that points to such an MPI_Datatype
object, but also has a "deleter" function that ensures that MPI_Type_free
is called whenever the object returned by this function goes out of scope.Usage example: std::vector<char> buffer; [...] if (buffer.size()<(1U<<31)) { // less than 2GB of data MPI_Send(buffer.data(), buffer.size(), MPI_BYTE, dest, tag, comm); } else { // more than 2GB of data const auto bigtype = Utilities::MPI::create_mpi_data_type_n_bytes(buffer.size()); MPI_Send(buffer.data(), 1, *bigtype, dest, tag, comm); }
Alternatively, the code in the else
branch can be simplified to the following: [...] else { // more than 2GB of data MPI_Send(buffer.data(), 1, *UtilitiesMPI::create_mpi_data_type_n_bytes(buffer.size()), dest, tag, comm); }
T Utilities::MPI::sum  (  const T &  t, 
const MPI_Comm &  mpi_communicator  
) 
Return the sum over all processors of the value t
. This function is collective over all processors given in the communicator. If deal.II is not configured for use of MPI, this function simply returns the value of t
. This function corresponds to the MPI_Allreduce
function, i.e. all processors receive the result of this operation.
MPI_Reduce
function instead of the MPI_Allreduce
function. The latter is at most twice as expensive, so if you are concerned about performance, it may be worthwhile investigating whether your algorithm indeed needs the result everywhere.T
, namely float, double, int, unsigned int
. void Utilities::MPI::sum  (  const T &  values, 
const MPI_Comm &  mpi_communicator,  
U &  sums  
) 
Like the previous function, but take the sums over the elements of an array of type T. In other words, the ith element of the results array is the sum over the ith entries of the input arrays from each processor. T and U must decay to the same type, e.g. they just differ by one of them having a const type qualifier and the other not.
Input and output arrays may be the same.
void Utilities::MPI::sum  (  const ArrayView< const T > &  values, 
const MPI_Comm &  mpi_communicator,  
const ArrayView< T > &  sums  
) 
Like the previous function, but take the sums over the elements of an array as specified by the ArrayView arguments. In other words, the ith element of the results array is the sum over the ith entries of the input arrays from each processor.
Input and output arrays may be the same.
SymmetricTensor< rank, dim, Number > Utilities::MPI::sum  (  const SymmetricTensor< rank, dim, Number > &  local, 
const MPI_Comm &  mpi_communicator  
) 
Perform an MPI sum of the entries of a symmetric tensor.
Tensor< rank, dim, Number > Utilities::MPI::sum  (  const Tensor< rank, dim, Number > &  local, 
const MPI_Comm &  mpi_communicator  
) 
Perform an MPI sum of the entries of a tensor.
void Utilities::MPI::sum  (  const SparseMatrix< Number > &  local, 
const MPI_Comm &  mpi_communicator,  
SparseMatrix< Number > &  global  
) 
Perform an MPI sum of the entries of a SparseMatrix.
local
and global
should have the same sparsity pattern and it should be the same for all MPI processes. T Utilities::MPI::max  (  const T &  t, 
const MPI_Comm &  mpi_communicator  
) 
Return the maximum over all processors of the value t
. This function is collective over all processors given in the communicator. If deal.II is not configured for use of MPI, this function simply returns the value of t
. This function corresponds to the MPI_Allreduce
function, i.e. all processors receive the result of this operation.
MPI_Reduce
function instead of the MPI_Allreduce
function. The latter is at most twice as expensive, so if you are concerned about performance, it may be worthwhile investigating whether your algorithm indeed needs the result everywhere.T
, namely float, double, int, unsigned int
. void Utilities::MPI::max  (  const T &  values, 
const MPI_Comm &  mpi_communicator,  
U &  maxima  
) 
Like the previous function, but take the maximum over the elements of an array of type T. In other words, the ith element of the results array is the maximum over the ith entries of the input arrays from each processor. T and U must decay to the same type, e.g. they just differ by one of them having a const type qualifier and the other not.
Input and output vectors may be the same.
void Utilities::MPI::max  (  const ArrayView< const T > &  values, 
const MPI_Comm &  mpi_communicator,  
const ArrayView< T > &  maxima  
) 
Like the previous function, but take the maximum over the elements of an array as specified by the ArrayView arguments. In other words, the ith element of the results array is the maximum over the ith entries of the input arrays from each processor.
Input and output arrays may be the same.
T Utilities::MPI::min  (  const T &  t, 
const MPI_Comm &  mpi_communicator  
) 
Return the minimum over all processors of the value t
. This function is collective over all processors given in the communicator. If deal.II is not configured for use of MPI, this function simply returns the value of t
. This function corresponds to the MPI_Allreduce
function, i.e. all processors receive the result of this operation.
MPI_Reduce
function instead of the MPI_Allreduce
function. The latter is at most twice as expensive, so if you are concerned about performance, it may be worthwhile investigating whether your algorithm indeed needs the result everywhere.T
, namely float, double, int, unsigned int
. void Utilities::MPI::min  (  const T &  values, 
const MPI_Comm &  mpi_communicator,  
U &  minima  
) 
Like the previous function, but take the minima over the elements of an array of type T. In other words, the ith element of the results array is the minimum of the ith entries of the input arrays from each processor. T and U must decay to the same type, e.g. they just differ by one of them having a const type qualifier and the other not.
Input and output arrays may be the same.
void Utilities::MPI::min  (  const ArrayView< const T > &  values, 
const MPI_Comm &  mpi_communicator,  
const ArrayView< T > &  minima  
) 
Like the previous function, but take the minimum over the elements of an array as specified by the ArrayView arguments. In other words, the ith element of the results array is the minimum over the ith entries of the input arrays from each processor.
Input and output arrays may be the same.
T Utilities::MPI::logical_or  (  const T &  t, 
const MPI_Comm &  mpi_communicator  
) 
Performs a logical or operation over all processors of the value t
. The logical or operator 
returns the boolean value true
if either or all operands are true
and returns false
otherwise. If the provided value t
corresponds to 0
in its associated data type T
, it will be interpreted as false
, and true
otherwise. Data type T
must be of type integral
, i.e., bool
, char
, short
, int
, long
, or any of their variations.
This function is collective over all processors given in the communicator. If deal.II is not configured for use of MPI, this function simply returns the value of value
. This function corresponds to the MPI_Allreduce
function, i.e., all processors receive the result of this operation.
MPI_Reduce
function instead of the MPI_Allreduce
function. The latter is at most twice as expensive, so if you are concerned about performance, it may be worthwhile investigating whether your algorithm indeed needs the result everywhere. void Utilities::MPI::logical_or  (  const T &  values, 
const MPI_Comm &  mpi_communicator,  
U &  results  
) 
Like the previous function, but performs the logical or operation on each element of an array. In other words, the ith element of the results array is the result of the logical or operation applied on the ith entries of the input arrays from each processor. T and U must decay to the same type, e.g., they just differ by one of them having a const type qualifier and the other not.
Input and output arrays may be the same.
std::vector
for the data type bool
. In that case, use a different container or data type. void Utilities::MPI::logical_or  (  const ArrayView< const T > &  values, 
const MPI_Comm &  mpi_communicator,  
const ArrayView< T > &  results  
) 
Like the previous function, but performs the logical or operation on each element of an array as specified by the ArrayView arguments. In other words, the ith element of the results array is the result of the logical or operation applied on the ith entries of the input arrays from each processor.
Input and output arrays may be the same.
Return sum, average, minimum, maximum, processor id of minimum and maximum as a collective operation of on the given MPI communicator mpi_communicator
. Each processor's value is given in my_value
and the result will be returned. The result is available on all machines.
MPI_Reduce
function instead of the MPI_Allreduce
function. The latter is at most twice as expensive, so if you are concerned about performance, it may be worthwhile investigating whether your algorithm indeed needs the result everywhere. std::vector< MinMaxAvg > Utilities::MPI::min_max_avg  (  const std::vector< double > &  my_value, 
const MPI_Comm &  mpi_communicator  
) 
Same as above but returning the sum, average, minimum, maximum, process id of minimum and maximum as a collective operation on the given MPI communicator mpi_communicator
for each entry of the vector.
void Utilities::MPI::min_max_avg  (  const ArrayView< const double > &  my_values, 
const ArrayView< MinMaxAvg > &  result,  
const MPI_Comm &  mpi_communicator  
) 
Same as above but returning the sum, average, minimum, maximum, process id of minimum and maximum as a collective operation on the given MPI communicator mpi_communicator
for each entry of the ArrayView.
bool Utilities::MPI::job_supports_mpi  (  ) 
Return whether (i) deal.II has been compiled to support MPI (for example by compiling with CXX=mpiCC
) and if so whether (ii) MPI_Init()
has been called (for example using the Utilities::MPI::MPI_InitFinalize class). In other words, the result indicates whether the current job is running under MPI.
std::map< unsigned int, T > Utilities::MPI::some_to_some  (  const MPI_Comm &  comm, 
const std::map< unsigned int, T > &  objects_to_send  
) 
Initiate a sometosome communication, and exchange arbitrary objects (the class T should be serializable using boost::serialize) between processors.
[in]  comm  MPI communicator. 
[in]  objects_to_send  A map from the rank (unsigned int) of the process meant to receive the data and the object to send (the type T must be serializable for this function to work properly). If this map contains an entry with a key equal to the rank of the current process (i.e., an instruction to a process to send data to itself), then this data item is simply copied to the returned object. 
std::vector< T > Utilities::MPI::all_gather  (  const MPI_Comm &  comm, 
const T &  object_to_send  
) 
A generalization of the classic MPI_Allgather function, that accepts arbitrary data types T, as long as boost::serialize accepts T as an argument.
[in]  comm  MPI communicator. 
[in]  object_to_send  An object to send to all other processes 
std::vector< T > Utilities::MPI::gather  (  const MPI_Comm &  comm, 
const T &  object_to_send,  
const unsigned int  root_process = 0 

) 
A generalization of the classic MPI_Gather function, that accepts arbitrary data types T, as long as boost::serialize accepts T as an argument.
[in]  comm  MPI communicator. 
[in]  object_to_send  an object to send to the root process 
[in]  root_process  The process, which receives the objects from all processes. By default the process with rank 0 is the root process. 
root_process
receives a vector of objects, with size equal to the number of processes in the MPI communicator. Each entry contains the object received from the processor with the corresponding rank within the communicator. All other processes receive an empty vector. std::enable_if< is_mpi_type< T >==false, T >::type Utilities::MPI::broadcast  (  const MPI_Comm &  comm, 
const T &  object_to_send,  
const unsigned int  root_process = 0 

) 
This function sends an object object_to_send
from the process root_process
to all other processes.
This function is a generalization of the classic MPI_Bcast
function that accepts arbitrary data types T
, as long as Utilities::pack() (which in turn uses boost::serialize
, see in Utilities::pack() for details) accepts T
as an argument.
MPI_Bcast
because the function will use boost::serialization to (de)serialize, and execute a second MPI_Bcast
to transmit the size before sending the data itself. On the other hand, if you have a single element of a data type T
that is natively supported by MPI, then the compiler will choose another broadcast() overload that is efficient. If you have an array of such elements, you should use the other broadcast() function in this namespace that takes a pointer and a count argument.[in]  comm  MPI communicator. 
[in]  object_to_send  An object to send to all processes. 
[in]  root_process  The process that sends the object to all processes. By default the process with rank 0 is the root process. 
T  Any type for which the Utilities::pack() and Utilities::unpack() functions can be used to convert the object into an array of char . The compiler will not select this function if T is a type that is natively supported by MPI and instead use a more efficient overload. 
object_to_send
. On every other process, return a copy of the object sent by the root_process
. std::enable_if< is_mpi_type< T >==true, T >::type Utilities::MPI::broadcast  (  const MPI_Comm &  comm, 
const T &  object_to_send,  
const unsigned int  root_process = 0 

) 
This function sends an object object_to_send
from the process root_process
to all other processes.
This function is wrapper around the MPI_Bcast
function selected by the compiler whenever T
is a data type natively supported by MPI.
[in]  comm  MPI communicator. 
[in]  object_to_send  An object to send to all processes. 
[in]  root_process  The process that sends the object to all processes. By default the process with rank 0 is the root process. 
T  Any type. The compiler will only select this function if T is a type that is natively supported by MPI. It will choose the other overloaded version of this function if that is not the case. 
object_to_send
. On every other process, return a copy of the object sent by the root_process
. void Utilities::MPI::broadcast  (  T *  buffer, 
const size_t  count,  
const unsigned int  root,  
const MPI_Comm &  comm  
) 
Broadcast the information in buffer
from root
to all other ranks.
Like MPI_Bcast
but with support to send data with a count
bigger than 2^31. The datatype to send needs to be supported directly by MPI and is automatically deduced from T.
Throws an exception if any MPI command fails.
buffer  Buffer of count objects. 
count  The number of objects to send. All processes need to specify the correct size. 
root  The rank of the process with the data. 
comm  The MPI communicator to use. 
T Utilities::MPI::reduce  (  const T &  local_value, 
const MPI_Comm &  comm,  
const std::function< T(const T &, const T &)> &  combiner,  
const unsigned int  root_process = 0 

) 
A function that combines values local_value
from all processes via a userspecified binary operation combiner
on the root_process
. As such this function is similar to MPI_Reduce (and Utilities::MPI::min/max()): however on the one hand due to the userspecified binary operation it is slower for builtin types but on the other hand general object types, including ones that store variable amounts of data, can be handled.
In contrast to all_reduce, the result will be only available on a single rank. On all other processes, the returned value is undefined.
T Utilities::MPI::all_reduce  (  const T &  local_value, 
const MPI_Comm &  comm,  
const std::function< T(const T &, const T &)> &  combiner  
) 
A function that combines values local_value
from all processes via a userspecified binary operation combiner
and distributes the result back to all processes. As such this function is similar to MPI_Allreduce (if it were implemented by a global reduction followed by a broadcast step) but due to the userspecified binary operation also general object types, including ones that store variable amounts of data, can be handled.
std::vector< unsigned int > Utilities::MPI::compute_index_owner  (  const IndexSet &  owned_indices, 
const IndexSet &  indices_to_look_up,  
const MPI_Comm &  comm  
) 
Given a partitioned index set space, compute the owning MPI process rank of each element of a second index set according to the partitioned index set. A natural usage of this function is to compute for each ghosted degree of freedom the MPI rank of the process owning that index.
One might think: "But we know which rank a ghost DoF belongs to based on the subdomain id of the cell it is on". But this heuristic fails for DoFs on interfaces between ghost cells with different subdomain_ids, or between a ghost cell and an artificial cell. Furthermore, this function enables a completely abstract exchange of information without the help of the mesh in terms of neighbors.
The first argument passed to this function, owned_indices
, must uniquely partition an index space between all processes. Otherwise, there are no limitations on this argument: In particular, there is no need in partitioning the index space into contiguous subsets. Furthermore, there are no limitations on the second index set indices_to_look_up
as long as the size matches the first one. It can be chosen arbitrarily and independently on each process. In the case that the second index set also contains locally owned indices, these indices will be treated correctly and the rank of this process is returned for those entries.
[in]  owned_indices  Index set with indices locally owned by this process. 
[in]  indices_to_look_up  Index set containing indices of which the user is interested the rank of the owning process. 
[in]  comm  MPI communicator. 
indices_to_look_up
. The order coincides with the order within the ElementIterator. std::vector< T > Utilities::MPI::compute_set_union  (  const std::vector< T > &  vec, 
const MPI_Comm &  comm  
) 
Compute the union of the input vectors vec
of all processes in the MPI communicator comm
.
std::set< T > Utilities::MPI::compute_set_union  (  const std::set< T > &  set, 
const MPI_Comm &  comm  
) 
The same as above but for std::set.

constexpr 
A template variable that is true
if the template argument T
is a data type that is natively supported by MPI, and false
otherwise. This variable can be used together with std::enable_if
to selectively allow template functions only for those data types for which the template type is supported by MPI. The variable is, in essence, a concept in the sense of C++20.
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< std::complex< double > > 
A template variable that translates from the data type given as template argument to the corresponding MPI_Datatype
to be used for MPI communication.
As an example, the value of mpi_type_id_for_type<int>
is MPI_INT
. A common way to use this variable is when sending an object obj
via MPI functions to another process, and using mpi_type_id_for_type<decltype(obj)>
to infer the correct MPI type to use for the communication.
The type T
given here must be one of the data types supported by MPI, such as int
or double
. It may not be an array of objects of such a type, or a pointer to an object of such a type. The compiler will produce an error if this requirement is not satisfied.
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< bool > 
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< char > 
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< signed char > 
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< short > 
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< int > 
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< long int > 
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< unsigned char > 
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< unsigned short > 
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< unsigned long int > 
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< unsigned long long int > 
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< float > 
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< double > 
template const MPI_Datatype Utilities::MPI::mpi_type_id_for_type< long double > 