ó nTMc@sEdZdefd„ƒYZddlZejjdƒrµdd.d„ƒYZdd/d „ƒYZeƒZd e_ d e_ d d0d „ƒYZ de fd„ƒYZ de fd„ƒYZ de fd„ƒYZde fd„ƒYZde fd„ƒYZde fd„ƒYZde fd„ƒYZde fd„ƒYZde fd„ƒYZd e fd!„ƒYZd"e fd#„ƒYZd$e fd%„ƒYZd&e fd'„ƒYZdZn‰y1dd(lTdd)lmZmZeeƒ[WnUek r=ddljZ dZd*d1d+„ƒYZ!d,d2d-„ƒYZ"e!ƒZnX[dS(3sq Python interface to the Message Passing Interface (MPI) This module contains a Python interface to the Message Passing Interface (MPI), and standardized library for message-passing parallel computing. Please read an introduction to MPI before using this module; some terms in the documentation do not make much sense unless you understand the principles of MPI. This module contains an object, 'world', which represents the default communicator in MPI. This communicator can be used directly for sending and receiving data, or other communicators can be derived from it. A number of global constants are also defined (L{max}, L{min}, L{prod}, L{sum}, L{land}, L{lor}, L{lxor}, L{band}, L{bor}, L{bxor}, L{maxloc}), and L{minloc}). They are used to specify the desired operator in calls to the 'reduce' and 'allreduce' methods of the communicator objects. @undocumented: core* tMPIErrorcBseZdZRS(s MPI call failed (t__name__t __module__t__doc__(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR siÿÿÿÿNtepydoctMPICommunicatorcBs¤eZdZd„Zd„Zd„Zd„Zddd„Zddd„Z ddd„Z ddd„Z d „Z d „Z d „Zd „Zd „Zd„ZRS(s MPI Communicator There is no constructor for MPI Communicator objects. The default communicator is given by Scientific.MPI.world, and other communicators can only be created by methods on an existing communicator object. A communicator object has two read-only attributes: 'rank' is an integer which indicates the rank of the current process in the communicator, and 'size' is an integer equal to the number of processes that participate in the communicator. cCsdS(s  @returns: a new communicator with the same properties as the original one @rtype: L{MPICommunicator} N((tself((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyt duplicate<scCsdS(s` Create a communicator for a subset of the processes The method should be called by all processes simultaneously. The return value will be the new communicator on those processes listed in C{ranks} and C{None} for the rest. @param ranks: a list of ranks, one for each process that should belong to the new communicator @type ranks: C{list} of C{int} @returns: a new communicator containing a subset of the processes participating in the original one N((Rtranks((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pytsubsetDscCsdS(sÓ Send data to another process (blocking) @param data: the data to be sent @type data: C{str} or C{Numeric.array}. Array arguments must have contiguous storage. General object arrays are not allowed. @param destination: the rank of the destination process @type destination: C{int} @param tag: Identifier @type tag: C{int} N((Rtdatat destinationttag((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pytsendTs cCsdS(s@ Send data to another process (non-blocking) @param data: the data to be sent @type data: C{str} or C{Numeric.array}. Array arguments must have contiguous storage. General object arrays are not allowed. @param destination: the rank of the destination process @type destination: C{int} @param tag: Identifier @type tag: C{int} @returns: MPI request object (used to wait for completion) @rtype: L{MPIRequest} N((RR R R ((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pytnonblockingSendcscCsdS(sE Receive data from another process (blocking) @param data: either a contiguous array object, or a one-letter typecode (in practice, one would use Numeric.Int, Numeric.Float, etc.). If an array, the data is copied to the array, whic must have the right shape. If a typecode, an array of that type is created and used as the buffer for incoming data. @type data: C{Numeric.array} or C{str} @param source: the rank of the process from which data is accepted. C{None} means data is accepted from any process. @type source: C{int} or C{NoneType} @param tag: Identifier that acts as a filter; only messages with a matching tag are received. A value of C{None} means that any tag will match. @type tag: C{int} or C{NoneType} @returns: a tuple containing four elements: the array containing the data, the source process rank (an C{int}), the message tag (an C{int}), and the number of elements that were received (an C{int}). @rtype: C{tuple} N((RR tsourceR ((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pytreceivetscCsdS(s Receive string data from another process (blocking) @param source: the rank of the process from which data is accepted. C{None} means data is accepted from any process. @type source: C{int} or C{NoneType} @param tag: Identifier that acts as a filter; only messages with a matching tag are received. A value of C{None} means that any tag will match. @type tag: C{int} or C{NoneType} @returns: a tuple containing three elements: the string containing the data, the source process rank (an C{int}), the message tag (an C{int}). @rtype: C{tuple} N((RRR ((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyt receiveStringscCsdS(s€ Receive data from another process (non-blocking) @param data: a contiguous array object to which the incoming data is copied. It must have the right shape. @type data: C{Numeric.array} @param source: the rank of the process from which data is accepted. C{None} means data is accepted from any process. @type source: C{int} or C{NoneType} @param tag: Identifier that acts as a filter; only messages with a matching tag are received. A value of C{None} means that any tag will match. @type tag: C{int} or C{NoneType} @returns: MPI request object (used to wait for completion and obtain the received data) @rtype: L{MPIRequest} N((RR RR ((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pytnonblockingReceive¢scCsdS(sÔ Check for incoming messages @param source: the rank of the process from which messages are accepted. C{None} means data is accepted from any process. @type source: C{int} or C{NoneType} @param tag: Identifier that acts as a filter; only messages with a matching tag are considered. A value of C{None} means that any tag will match. @type tag: C{int} or C{NoneType} @returns: C{None} if no messages are available, otherwise a tuple containing the source rank and the tag @rtype: C{NoneType} or C{tuple} N((RRR ((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pytnonblockingProbe·scCsdS(sB Send data to all processes @param array: an array containing the data to be sent on the sending process and serving as a buffer for the incoming data on all processes. The shape and type of the array must be the same on all processes. @type array: Numeric.array @param root: the rank of the sending process @type root: C{int} @note: The data is sent to all processes, including the sending one. N((Rtarraytroot((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyt broadcastÉscCsdS(sç Distribute data from each processpr to all other processesors @param send: an array of identical shape and type on all processes. It contains on each process the data that is sent. @type send: C{Numeric.array} @param receive: an array whose type is the same as for the send array and which has an additional dimension (the first one) whose length is the number of processes. After the call, the value of receive[i] is equal to the contents of the array send in process i. @type receive: C{Numeric.array} N((RR R((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pytshareÙscCsdS(sŒ Wait until all processes in the communicator have called the same method, then all processes continue. N((R((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pytbarrierêscCsdS(s] Abort all processes associated with the communicator. For emergency use only. @param error_code: error code passed back to the calling program (usually a shell) under most Unix implementations of MPI @type error_code: C{int} N((Rt error_code((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pytabortñs cCsdS(s“ Combine data from all processes and send result to one @param sendbuffer: an array holding the data that each process contributes @type sendbuffer: C{Numeric.array} @param receivebuffer: an array acting as a buffer for the result of the reduction. Used only by the process whose rank is root @type receivebuffer: C{Numeric.array} @param operation: one of the operation objects: L{max}, L{min}, L{prod}, L{sum}, L{land}, L{lor}, L{lxor}, L{band}, L{bor}, L{bxor}, L{maxloc} and L{minloc} @type operation: MPIOperationObject @param root: the rank of the process that received the result @type root: C{int} N((Rt sendbuffert receivebuffert operationR((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pytreduceýscCsdS(sÛ Combine data from all processes and send result to all @param sendbuffer: an array holding the data that each process contributes @type sendbuffer: C{Numeric.array} @param receivebuffer: an array acting as a buffer for the result of the reduction @type receivebuffer: C{Numeric.array} @param operation: one of the operation objects: L{max}, L{min}, L{prod}, L{sum}, L{land}, L{lor}, L{lxor}, L{band}, L{bor}, L{bxor}, L{maxloc} and L{minloc} @type operation: MPIOperationObject N((RRRR((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyt allreducesN(RRRRR R RtNoneRRRRRRRRRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR,s        t MPIRequestcBs eZdZd„Zd„ZRS(sÍ MPI Request There is no constructor for MPI Request objects. They are returned by nonblocking send and receives, and are used to query the status of the message. cCsdS(sÑ Wait till the communication has completed. If the operation was a nonblocking send, there is no return value. If the operation was a nonblocking receive, the return value is a tuple containing four elements: the array containing the data, the source process rank (an integer), the message tag (an integer), and the number of elements that were received (an integer). N((R((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pytwait-s cCsdS(s Test if communications have completed. If the operation was a nonblocking send, it returns 0 if the operation has not completed, and 1 if it has. If the operation was a nonblocking receive, 0 is returned if the operation was not completed, and a tuple containing four elements if it was completed. The four elements are: the array containing the data, the source process rank (an integer), the message tag (an integer), and the number of elements that were received (an integer). Once a test has been successful (i.e. the operation has completed), it is no longer possible to call wait() or test() on the MPI Request object. N((R((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyttest:s(RRRR"R#(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR!$s iitMPIOperationObjectcBseZRS((RR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR$TstmaxcBseZdZRS(s;The 'maximum' operation in reduce/allreduce communications.(RRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR%WstmincBseZdZRS(s;The 'minimum' operation in reduce/allreduce communications.(RRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR&[stprodcBseZdZRS(s;The 'product' operation in reduce/allreduce communications.(RRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR'_stsumcBseZdZRS(s7The 'sum' operation in reduce/allreduce communications.(RRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR(cstlandcBseZdZRS(s?The 'logical and' operation in reduce/allreduce communications.(RRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR)gstlorcBseZdZRS(s>The 'logical or' operation in reduce/allreduce communications.(RRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR*kstlxorcBseZdZRS(s%The 'logical exclusive-or' operation.(RRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR+ostbandcBseZdZRS(s?The 'bitwise and' operation in reduce/allreduce communications.(RRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR,sstborcBseZdZRS(s>The 'bitwise or' operation in reduce/allreduce communications.(RRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR-wstbxorcBseZdZRS(s%The 'bitwise exclusive-or' operation.(RRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR.{stmaxloccBseZdZRS(s(The 'location of the maximum' operation.(RRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR/stminloccBseZdZRS(s(The 'location of the minimum' operation.(RRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR0ƒstreplacecBseZdZRS(s"The 'replace' operation. (MPI 2.0)(RRR(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR1‡s(t*(t_C_APIt_registerErrorObjecttDummyCommunicatorcBsŒeZd„Zd„Zd„Zd„Zd d d„Zd d d„Zd d d„Z d d d„Z d„Z d „Z d „Z d „ZRS( cCsd|_d|_g|_dS(Nii(tsizetranktmessages(R((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyt__init__œs  cCstƒS(N(R5(R((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR¡scCsJ|dkrtdƒ‚n|jj|tj|ddƒjƒfƒdS(Nisinvalid MPI destinationtcopyi(RR8tappendtNumericRtravel(RR R R ((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR ¤s cCs|j|||ƒtdƒS(N(R t DummyRequestR (RR R R ((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR©scCs¢|dkr'|dkr'tdƒ‚nxhtt|jƒƒD]Q}|j|\}}|dksn||kr=|j|=|d|t|ƒfSq=Wtdƒ‚dS(Nisinvalid MPI sourcesno message received(R RtrangetlenR8(RRRR titdata_tagR ((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR­s cCs1|j||ƒ\}}}}|jƒ||fS(N(Rttostring(RRR Rtlength((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR·scCst|j|||ƒƒS(N(R>R(RRRR ((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR»scCs}|dkr$|dkr$td‚nxRtt|jƒƒD];}|j|\}}|dksk||kr:d|fSq:WdS(Nisinvalid MPI source(R RR?R@R8(RRR RARBR ((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR¾s cCs|dkrtdƒ‚n|S(Nisinvalid MPI rank(R(RRR((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyRÇs cCs||dcBseZd„Zd„ZRS(cCs ||_dS(N(targ(RRE((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR9×scCs|jS(N(RE(R((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR"Ús(RRR9R"(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pyR>Õs ((((((#RtEnvironmentErrorRtsystmodulesthas_keyRR!tworldR7R6R$R%R&R'R(R)R*R+R,R-R.R/R0R1R R3tScientific_mpiR4t ImportErrort Scientific.NtNR<R5R>(((s4C:\Python27\Lib\site-packages\Scientific\MPI\core.pytsF ø*       ;