Update command line args format in parallel docs section.

docs/source/parallel/parallel_demos.txt

@@ -1,284 +1,284 b''
 =================
 Parallel examples
 =================
 
 .. note::
 
     Performance numbers from ``IPython.kernel``, not newparallel.
 
 In this section we describe two more involved examples of using an IPython
 cluster to perform a parallel computation. In these examples, we will be using
 IPython's "pylab" mode, which enables interactive plotting using the
 Matplotlib package. IPython can be started in this mode by typing::
 
     ipython --pylab
 
 at the system command line.
 
 150 million digits of pi
 ========================
 
 In this example we would like to study the distribution of digits in the
 number pi (in base 10). While it is not known if pi is a normal number (a
 number is normal in base 10 if 0-9 occur with equal likelihood) numerical
 investigations suggest that it is. We will begin with a serial calculation on
 10,000 digits of pi and then perform a parallel calculation involving 150
 million digits.
 
 In both the serial and parallel calculation we will be using functions defined
 in the :file:`pidigits.py` file, which is available in the
 :file:`docs/examples/newparallel` directory of the IPython source distribution.
 These functions provide basic facilities for working with the digits of pi and
 can be loaded into IPython by putting :file:`pidigits.py` in your current
 working directory and then doing:
 
 .. sourcecode:: ipython
 
     In [1]: run pidigits.py
 
 Serial calculation
 ------------------
 
 For the serial calculation, we will use `SymPy <http://www.sympy.org>`_ to
 calculate 10,000 digits of pi and then look at the frequencies of the digits
 0-9. Out of 10,000 digits, we expect each digit to occur 1,000 times. While
 SymPy is capable of calculating many more digits of pi, our purpose here is to
 set the stage for the much larger parallel calculation.
 
 In this example, we use two functions from :file:`pidigits.py`: 
 :func:`one_digit_freqs` (which calculates how many times each digit occurs)
 and :func:`plot_one_digit_freqs` (which uses Matplotlib to plot the result).
 Here is an interactive IPython session that uses these functions with
 SymPy:
 
 .. sourcecode:: ipython
 
     In [7]: import sympy
 
     In [8]: pi = sympy.pi.evalf(40)
 
     In [9]: pi
     Out[9]: 3.141592653589793238462643383279502884197
 
     In [10]: pi = sympy.pi.evalf(10000)
 
     In [11]: digits = (d for d in str(pi)[2:])  # create a sequence of digits
 
     In [12]: run pidigits.py  # load one_digit_freqs/plot_one_digit_freqs
 
     In [13]: freqs = one_digit_freqs(digits)
 
     In [14]: plot_one_digit_freqs(freqs)
     Out[14]: [<matplotlib.lines.Line2D object at 0x18a55290>]
 
 The resulting plot of the single digit counts shows that each digit occurs
 approximately 1,000 times, but that with only 10,000 digits the
 statistical fluctuations are still rather large:
 
 .. image:: single_digits.*
 
 It is clear that to reduce the relative fluctuations in the counts, we need
 to look at many more digits of pi. That brings us to the parallel calculation.
 
 Parallel calculation
 --------------------
 
 Calculating many digits of pi is a challenging computational problem in itself. 
 Because we want to focus on the distribution of digits in this example, we
 will use pre-computed digit of pi from the website of Professor Yasumasa
 Kanada at the University of Tokyo (http://www.super-computing.org). These
 digits come in a set of text files (ftp://pi.super-computing.org/.2/pi200m/)
 that each have 10 million digits of pi.
 
 For the parallel calculation, we have copied these files to the local hard
 drives of the compute nodes. A total of 15 of these files will be used, for a
 total of 150 million digits of pi. To make things a little more interesting we
 will calculate the frequencies of all 2 digits sequences (00-99) and then plot
 the result using a 2D matrix in Matplotlib.
 
 The overall idea of the calculation is simple: each IPython engine will
 compute the two digit counts for the digits in a single file. Then in a final
 step the counts from each engine will be added up. To perform this
 calculation, we will need two top-level functions from :file:`pidigits.py`:
 
 .. literalinclude:: ../../examples/newparallel/pidigits.py
    :language: python
-   :lines: 41-56
+   :lines: 47-62
 
 We will also use the :func:`plot_two_digit_freqs` function to plot the
 results. The code to run this calculation in parallel is contained in
 :file:`docs/examples/newparallel/parallelpi.py`. This code can be run in parallel
 using IPython by following these steps:
 
 1. Use :command:`ipcluster` to start 15 engines. We used an 8 core (2 quad 
    core CPUs) cluster with hyperthreading enabled which makes the 8 cores 
    looks like 16 (1 controller + 15 engines) in the OS. However, the maximum
    speedup we can observe is still only 8x.
 2. With the file :file:`parallelpi.py` in your current working directory, open
    up IPython in pylab mode and type ``run parallelpi.py``.  This will download
    the pi files via ftp the first time you run it, if they are not
    present in the Engines' working directory.
 
 When run on our 8 core cluster, we observe a speedup of 7.7x. This is slightly
 less than linear scaling (8x) because the controller is also running on one of
 the cores.
 
 To emphasize the interactive nature of IPython, we now show how the
 calculation can also be run by simply typing the commands from
 :file:`parallelpi.py` interactively into IPython:
 
 .. sourcecode:: ipython
 
     In [1]: from IPython.parallel import Client
 
     # The Client allows us to use the engines interactively.
     # We simply pass Client the name of the cluster profile we
     # are using.
     In [2]: c = Client(profile='mycluster')
     In [3]: view = c.load_balanced_view()
 
     In [3]: c.ids
     Out[3]: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14]
 
     In [4]: run pidigits.py
 
     In [5]: filestring = 'pi200m.ascii.%(i)02dof20'
 
     # Create the list of files to process.
     In [6]: files = [filestring % {'i':i} for i in range(1,16)]
 
     In [7]: files
     Out[7]: 
     ['pi200m.ascii.01of20',
      'pi200m.ascii.02of20',
      'pi200m.ascii.03of20',
      'pi200m.ascii.04of20',
      'pi200m.ascii.05of20',
      'pi200m.ascii.06of20',
      'pi200m.ascii.07of20',
      'pi200m.ascii.08of20',
      'pi200m.ascii.09of20',
      'pi200m.ascii.10of20',
      'pi200m.ascii.11of20',
      'pi200m.ascii.12of20',
      'pi200m.ascii.13of20',
      'pi200m.ascii.14of20',
      'pi200m.ascii.15of20']
     
     # download the data files if they don't already exist:
     In [8]: v.map(fetch_pi_file, files)
     
     # This is the parallel calculation using the Client.map method 
     # which applies compute_two_digit_freqs to each file in files in parallel.
     In [9]: freqs_all = v.map(compute_two_digit_freqs, files)
 
     # Add up the frequencies from each engine.
     In [10]: freqs = reduce_freqs(freqs_all)
 
     In [11]: plot_two_digit_freqs(freqs)
     Out[11]: <matplotlib.image.AxesImage object at 0x18beb110>
 
     In [12]: plt.title('2 digit counts of 150m digits of pi')
     Out[12]: <matplotlib.text.Text object at 0x18d1f9b0>
 
 The resulting plot generated by Matplotlib is shown below. The colors indicate
 which two digit sequences are more (red) or less (blue) likely to occur in the
 first 150 million digits of pi. We clearly see that the sequence "41" is 
 most likely and that "06" and "07" are least likely. Further analysis would
 show that the relative size of the statistical fluctuations have decreased
 compared to the 10,000 digit calculation.
 
 .. image:: two_digit_counts.*
 
 
 Parallel options pricing
 ========================
 
 An option is a financial contract that gives the buyer of the contract the
 right to buy (a "call") or sell (a "put") a secondary asset (a stock for
 example) at a particular date in the future (the expiration date) for a
 pre-agreed upon price (the strike price). For this right, the buyer pays the
 seller a premium (the option price). There are a wide variety of flavors of
 options (American, European, Asian, etc.) that are useful for different
 purposes: hedging against risk, speculation, etc.
 
 Much of modern finance is driven by the need to price these contracts
 accurately based on what is known about the properties (such as volatility) of
 the underlying asset. One method of pricing options is to use a Monte Carlo
 simulation of the underlying asset price. In this example we use this approach
 to price both European and Asian (path dependent) options for various strike
 prices and volatilities.
 
 The code for this example can be found in the :file:`docs/examples/newparallel`
 directory of the IPython source. The function :func:`price_options` in
 :file:`mcpricer.py` implements the basic Monte Carlo pricing algorithm using
 the NumPy package and is shown here:
 
 .. literalinclude:: ../../examples/newparallel/mcpricer.py
    :language: python
 
 To run this code in parallel, we will use IPython's :class:`LoadBalancedView` class,
 which distributes work to the engines using dynamic load balancing. This
 view is a wrapper of the :class:`Client` class shown in
 the previous example. The parallel calculation using :class:`LoadBalancedView` can
 be found in the file :file:`mcpricer.py`. The code in this file creates a
 :class:`TaskClient` instance and then submits a set of tasks using
 :meth:`TaskClient.run` that calculate the option prices for different
 volatilities and strike prices. The results are then plotted as a 2D contour
 plot using Matplotlib.
 
 .. literalinclude:: ../../examples/newparallel/mcdriver.py
    :language: python
 
 To use this code, start an IPython cluster using :command:`ipcluster`, open
 IPython in the pylab mode with the file :file:`mcdriver.py` in your current
 working directory and then type:
 
 .. sourcecode:: ipython
 
     In [7]: run mcdriver.py
     Submitted tasks:  [0, 1, 2, ...]
 
 Once all the tasks have finished, the results can be plotted using the
 :func:`plot_options` function. Here we make contour plots of the Asian
 call and Asian put options as function of the volatility and strike price:
 
 .. sourcecode:: ipython
 
     In [8]: plot_options(sigma_vals, K_vals, prices['acall'])
 
     In [9]: plt.figure()
     Out[9]: <matplotlib.figure.Figure object at 0x18c178d0>
 
     In [10]: plot_options(sigma_vals, K_vals, prices['aput'])
 
 These results are shown in the two figures below. On a 8 core cluster the
 entire calculation (10 strike prices, 10 volatilities, 100,000 paths for each)
 took 30 seconds in parallel, giving a speedup of 7.7x, which is comparable
 to the speedup observed in our previous example.
 
 .. image:: asian_call.*
 
 .. image:: asian_put.*
 
 Conclusion
 ==========
 
 To conclude these examples, we summarize the key features of IPython's
 parallel architecture that have been demonstrated:
 
 * Serial code can be parallelized often with only a few extra lines of code.
   We have used the :class:`DirectView` and :class:`LoadBalancedView` classes
   for this purpose.
 * The resulting parallel code can be run without ever leaving the IPython's
   interactive shell.
 * Any data computed in parallel can be explored interactively through 
   visualization or further numerical calculations.
 * We have run these examples on a cluster running Windows HPC Server 2008.
   IPython's built in support for the Windows HPC job scheduler makes it
   easy to get started with IPython's parallel capabilities.
 
 .. note::
 
     The newparallel code has never been run on Windows HPC Server, so the last
     conclusion is untested.

docs/source/parallel/parallel_intro.txt

@@ -1,253 +1,253 b''
 .. _ip1par:
 
 ============================
 Overview and getting started
 ============================
 
 Introduction
 ============
 
 This section gives an overview of IPython's sophisticated and powerful
 architecture for parallel and distributed computing. This architecture
 abstracts out parallelism in a very general way, which enables IPython to
 support many different styles of parallelism including:
 
 * Single program, multiple data (SPMD) parallelism.
 * Multiple program, multiple data (MPMD) parallelism.
 * Message passing using MPI.
 * Task farming.
 * Data parallel.
 * Combinations of these approaches.
 * Custom user defined approaches.
 
 Most importantly, IPython enables all types of parallel applications to
 be developed, executed, debugged and monitored *interactively*. Hence,
 the ``I`` in IPython.  The following are some example usage cases for IPython:
 
 * Quickly parallelize algorithms that are embarrassingly parallel
   using a number of simple approaches.  Many simple things can be
   parallelized interactively in one or two lines of code.
 
 * Steer traditional MPI applications on a supercomputer from an
   IPython session on your laptop.
 
 * Analyze and visualize large datasets (that could be remote and/or
   distributed) interactively using IPython and tools like
   matplotlib/TVTK.
 
 * Develop, test and debug new parallel algorithms 
   (that may use MPI) interactively.
 
 * Tie together multiple MPI jobs running on different systems into
   one giant distributed and parallel system.
 
 * Start a parallel job on your cluster and then have a remote 	
   collaborator connect to it and pull back data into their 
   local IPython session for plotting and analysis.
 
 * Run a set of tasks on a set of CPUs using dynamic load balancing.
 
 Architecture overview
 =====================
 
 The IPython architecture consists of four components:
 
 * The IPython engine.
 * The IPython hub.
 * The IPython schedulers.
 * The controller client.
 
 These components live in the :mod:`IPython.parallel` package and are
 installed with IPython.  They do, however, have additional dependencies
 that must be installed.  For more information, see our
 :ref:`installation documentation <install_index>`.
 
 .. TODO: include zmq in install_index
 
 IPython engine
 ---------------
 
 The IPython engine is a Python instance that takes Python commands over a
 network connection. Eventually, the IPython engine will be a full IPython
 interpreter, but for now, it is a regular Python interpreter. The engine
 can also handle incoming and outgoing Python objects sent over a network
 connection.  When multiple engines are started, parallel and distributed
 computing becomes possible. An important feature of an IPython engine is
 that it blocks while user code is being executed. Read on for how the
 IPython controller solves this problem to expose a clean asynchronous API
 to the user.
 
 IPython controller
 ------------------
 
 The IPython controller processes provide an interface for working with a set of engines.
 At a general level, the controller is a collection of processes to which IPython engines
 and clients can connect. The controller is composed of a :class:`Hub` and a collection of
 :class:`Schedulers`. These Schedulers are typically run in separate processes but on the
 same machine as the Hub, but can be run anywhere from local threads or on remote machines.
 
 The controller also provides a single point of contact for users who wish to
 utilize the engines connected to the controller. There are different ways of
 working with a controller. In IPython, all of these models are implemented via
 the client's :meth:`.View.apply` method, with various arguments, or
 constructing :class:`.View` objects to represent subsets of engines. The two
 primary models for interacting with engines are:
 
 * A **Direct** interface, where engines are addressed explicitly.
 * A **LoadBalanced** interface, where the Scheduler is trusted with assigning work to
   appropriate engines.
 
 Advanced users can readily extend the View models to enable other
 styles of parallelism. 
 
 .. note:: 
 
     A single controller and set of engines can be used with multiple models
     simultaneously. This opens the door for lots of interesting things.
 
 
 The Hub
 *******
 
 The center of an IPython cluster is the Hub. This is the process that keeps
 track of engine connections, schedulers, clients, as well as all task requests and
 results. The primary role of the Hub is to facilitate queries of the cluster state, and
 minimize the necessary information required to establish the many connections involved in
 connecting new clients and engines.
 
 
 Schedulers
 **********
 
 All actions that can be performed on the engine go through a Scheduler. While the engines
 themselves block when user code is run, the schedulers hide that from the user to provide
 a fully asynchronous interface to a set of engines.
 
 
 IPython client and views
 ------------------------
 
 There is one primary object, the :class:`~.parallel.Client`, for connecting to a cluster.
 For each execution model, there is a corresponding :class:`~.parallel.View`. These views
 allow users to interact with a set of engines through the interface. Here are the two default
 views:
 
 * The :class:`DirectView` class for explicit addressing.
 * The :class:`LoadBalancedView` class for destination-agnostic scheduling.
 
 Security
 --------
 
 IPython uses ZeroMQ for networking, which has provided many advantages, but
 one of the setbacks is its utter lack of security [ZeroMQ]_. By default, no IPython
 connections are encrypted, but open ports only listen on localhost. The only
 source of security for IPython is via ssh-tunnel. IPython supports both shell
 (`openssh`) and `paramiko` based tunnels for connections.  There is a key necessary
 to submit requests, but due to the lack of encryption, it does not provide
 significant security if loopback traffic is compromised.
 
 In our architecture, the controller is the only process that listens on
 network ports, and is thus the main point of vulnerability. The standard model
 for secure connections is to designate that the controller listen on
 localhost, and use ssh-tunnels to connect clients and/or
 engines.
 
 To connect and authenticate to the controller an engine or client needs
 some information that the controller has stored in a JSON file.
 Thus, the JSON files need to be copied to a location where
 the clients and engines can find them. Typically, this is the
 :file:`~/.ipython/profile_default/security` directory on the host where the 
 client/engine is running (which could be a different host than the controller). 
 Once the JSON files are copied over, everything should work fine.
 
 Currently, there are two JSON files that the controller creates:
 
 ipcontroller-engine.json
     This JSON file has the information necessary for an engine to connect
     to a controller.
 
 ipcontroller-client.json
     The client's connection information.  This may not differ from the engine's,
     but since the controller may listen on different ports for clients and
     engines, it is stored separately.
 
 More details of how these JSON files are used are given below.
 
 A detailed description of the security model and its implementation in IPython
 can be found :ref:`here <parallelsecurity>`.
 
 .. warning::
 
     Even at its most secure, the Controller listens on ports on localhost, and
     every time you make a tunnel, you open a localhost port on the connecting
     machine that points to the Controller. If localhost on the Controller's
     machine, or the machine of any client or engine, is untrusted, then your
     Controller is insecure. There is no way around this with ZeroMQ.
 
 
 
 Getting Started
 ===============
 
 To use IPython for parallel computing, you need to start one instance of the
 controller and one or more instances of the engine. Initially, it is best to
 simply start a controller and engines on a single host using the
 :command:`ipcluster` command. To start a controller and 4 engines on your
 localhost, just do::
 
-    $ ipcluster start n=4
+    $ ipcluster start --n=4
 
 More details about starting the IPython controller and engines can be found
 :ref:`here <parallel_process>`
 
 Once you have started the IPython controller and one or more engines, you
 are ready to use the engines to do something useful. To make sure
 everything is working correctly, try the following commands:
 
 .. sourcecode:: ipython
 
 	In [1]: from IPython.parallel import Client
 	
 	In [2]: c = Client()
 	
 	In [4]: c.ids
 	Out[4]: set([0, 1, 2, 3])
 	
 	In [5]: c[:].apply_sync(lambda : "Hello, World")
 	Out[5]: [ 'Hello, World', 'Hello, World', 'Hello, World', 'Hello, World' ]
 
 
 When a client is created with no arguments, the client tries to find the corresponding JSON file
 in the local `~/.ipython/profile_default/security` directory. Or if you specified a profile,
 you can use that with the Client.  This should cover most cases:
 
 .. sourcecode:: ipython
 
     In [2]: c = Client(profile='myprofile')
 
 If you have put the JSON file in a different location or it has a different name, create the
 client like this:
 
 .. sourcecode:: ipython
 
     In [2]: c = Client('/path/to/my/ipcontroller-client.json')
 
 Remember, a client needs to be able to see the Hub's ports to connect. So if they are on a
 different machine, you may need to use an ssh server to tunnel access to that machine,
 then you would connect to it with:
 
 .. sourcecode:: ipython
 
     In [2]: c = Client(sshserver='myhub.example.com')
 
 Where 'myhub.example.com' is the url or IP address of the machine on
 which the Hub process is running (or another machine that has direct access to the Hub's ports).
 
 The SSH server may already be specified in ipcontroller-client.json, if the controller was
 instructed at its launch time.
 
 You are now ready to learn more about the :ref:`Direct
 <parallel_multiengine>` and :ref:`LoadBalanced <parallel_task>` interfaces to the
 controller.
 
 .. [ZeroMQ] ZeroMQ.  http://www.zeromq.org

docs/source/parallel/parallel_mpi.txt

@@ -1,156 +1,156 b''
 .. _parallelmpi:
 
 =======================
 Using MPI with IPython
 =======================
 
 .. note::
 
     Not adapted to zmq yet
     This is out of date wrt ipcluster in general as well
 
 Often, a parallel algorithm will require moving data between the engines. One
 way of accomplishing this is by doing a pull and then a push using the
 multiengine client. However, this will be slow as all the data has to go
 through the controller to the client and then back through the controller, to
 its final destination.
 
 A much better way of moving data between engines is to use a message passing
 library, such as the Message Passing Interface (MPI) [MPI]_. IPython's
 parallel computing architecture has been designed from the ground up to
 integrate with MPI. This document describes how to use MPI with IPython.
 
 Additional installation requirements
 ====================================
 
 If you want to use MPI with IPython, you will need to install:
 
 * A standard MPI implementation such as OpenMPI [OpenMPI]_ or MPICH.
 * The mpi4py [mpi4py]_ package.
 
 .. note::
 
     The mpi4py package is not a strict requirement. However, you need to
     have *some* way of calling MPI from Python. You also need some way of
     making sure that :func:`MPI_Init` is called when the IPython engines start
     up. There are a number of ways of doing this and a good number of
     associated subtleties. We highly recommend just using mpi4py as it
     takes care of most of these problems. If you want to do something
     different, let us know and we can help you get started.
 
 Starting the engines with MPI enabled
 =====================================
 
 To use code that calls MPI, there are typically two things that MPI requires.
 
 1. The process that wants to call MPI must be started using
    :command:`mpiexec` or a batch system (like PBS) that has MPI support.
 2. Once the process starts, it must call :func:`MPI_Init`.
 
 There are a couple of ways that you can start the IPython engines and get
 these things to happen.
 
 Automatic starting using :command:`mpiexec` and :command:`ipcluster`
 --------------------------------------------------------------------
 
 The easiest approach is to use the `MPIExec` Launchers in :command:`ipcluster`,
 which will first start a controller and then a set of engines using
 :command:`mpiexec`::
 
-    $ ipcluster start n=4 elauncher=MPIExecEngineSetLauncher
+    $ ipcluster start --n=4 --elauncher=MPIExecEngineSetLauncher
 
 This approach is best as interrupting :command:`ipcluster` will automatically
 stop and clean up the controller and engines.
 
 Manual starting using :command:`mpiexec`
 ----------------------------------------
 
 If you want to start the IPython engines using the :command:`mpiexec`, just
 do::
 
-    $ mpiexec n=4 ipengine mpi=mpi4py
+    $ mpiexec n=4 ipengine --mpi=mpi4py
 
 This requires that you already have a controller running and that the FURL
 files for the engines are in place. We also have built in support for
 PyTrilinos [PyTrilinos]_, which can be used (assuming is installed) by
 starting the engines with::
 
-    $ mpiexec n=4 ipengine mpi=pytrilinos
+    $ mpiexec n=4 ipengine --mpi=pytrilinos
 
 Automatic starting using PBS and :command:`ipcluster`
 ------------------------------------------------------
 
 The :command:`ipcluster` command also has built-in integration with PBS. For
 more information on this approach, see our documentation on :ref:`ipcluster
 <parallel_process>`.
 
 Actually using MPI
 ==================
 
 Once the engines are running with MPI enabled, you are ready to go. You can
 now call any code that uses MPI in the IPython engines. And, all of this can
 be done interactively. Here we show a simple example that uses mpi4py
 [mpi4py]_ version 1.1.0 or later.
 
 First, lets define a simply function that uses MPI to calculate the sum of a
 distributed array. Save the following text in a file called :file:`psum.py`:
 
 .. sourcecode:: python
 
     from mpi4py import MPI
     import numpy as np
 
     def psum(a):
         s = np.sum(a)
         rcvBuf = np.array(0.0,'d')
         MPI.COMM_WORLD.Allreduce([s, MPI.DOUBLE],
             [rcvBuf, MPI.DOUBLE],
             op=MPI.SUM)
         return rcvBuf
 
 Now, start an IPython cluster::
 
-    $ ipcluster start profile=mpi n=4
+    $ ipcluster start --profile=mpi --n=4
     
 .. note::
 
     It is assumed here that the mpi profile has been set up, as described :ref:`here
     <parallel_process>`.
 
 Finally, connect to the cluster and use this function interactively. In this
 case, we create a random array on each engine and sum up all the random arrays
 using our :func:`psum` function:
 
 .. sourcecode:: ipython
 
     In [1]: from IPython.parallel import Client
     
     In [2]: %load_ext parallel_magic
     
     In [3]: c = Client(profile='mpi')
     
     In [4]: view = c[:]
     
     In [5]: view.activate()
     
     # run the contents of the file on each engine:
     In [6]: view.run('psum.py')
     
     In [6]: px a = np.random.rand(100)
     Parallel execution on engines: [0,1,2,3]
     
     In [8]: px s = psum(a)
     Parallel execution on engines: [0,1,2,3]
     
     In [9]: view['s']
     Out[9]: [187.451545803,187.451545803,187.451545803,187.451545803]
 
 Any Python code that makes calls to MPI can be used in this manner, including
 compiled C, C++ and Fortran libraries that have been exposed to Python.
 
 .. [MPI] Message Passing Interface.  http://www-unix.mcs.anl.gov/mpi/
 .. [mpi4py] MPI for Python. mpi4py: http://mpi4py.scipy.org/
 .. [OpenMPI] Open MPI. http://www.open-mpi.org/
 .. [PyTrilinos] PyTrilinos. http://trilinos.sandia.gov/packages/pytrilinos/
 
 

docs/source/parallel/parallel_multiengine.txt

@@ -1,847 +1,847 b''
 .. _parallel_multiengine:
 
 ==========================
 IPython's Direct interface
 ==========================
 
 The direct, or multiengine, interface represents one possible way of working with a set of
 IPython engines. The basic idea behind the multiengine interface is that the
 capabilities of each engine are directly and explicitly exposed to the user.
 Thus, in the multiengine interface, each engine is given an id that is used to
 identify the engine and give it work to do. This interface is very intuitive
 and is designed with interactive usage in mind, and is the best place for
 new users of IPython to begin.
 
 Starting the IPython controller and engines
 ===========================================
 
 To follow along with this tutorial, you will need to start the IPython
 controller and four IPython engines. The simplest way of doing this is to use
 the :command:`ipcluster` command::
 
-    $ ipcluster start n=4
+    $ ipcluster start --n=4
     
 For more detailed information about starting the controller and engines, see
 our :ref:`introduction <ip1par>` to using IPython for parallel computing.
 
 Creating a ``Client`` instance
 ==============================
 
 The first step is to import the IPython :mod:`IPython.parallel`
 module and then create a :class:`.Client` instance:
 
 .. sourcecode:: ipython
 
     In [1]: from IPython.parallel import Client
     
     In [2]: rc = Client()
 
 This form assumes that the default connection information (stored in
 :file:`ipcontroller-client.json` found in :file:`IPYTHON_DIR/profile_default/security`) is
 accurate. If the controller was started on a remote machine, you must copy that connection
 file to the client machine, or enter its contents as arguments to the Client constructor:
 
 .. sourcecode:: ipython
 
     # If you have copied the json connector file from the controller:
     In [2]: rc = Client('/path/to/ipcontroller-client.json')
     # or to connect with a specific profile you have set up:
     In [3]: rc = Client(profile='mpi')
     
 
 To make sure there are engines connected to the controller, users can get a list
 of engine ids:
 
 .. sourcecode:: ipython
 
     In [3]: rc.ids
     Out[3]: [0, 1, 2, 3]
 
 Here we see that there are four engines ready to do work for us.
 
 For direct execution, we will make use of a :class:`DirectView` object, which can be
 constructed via list-access to the client:
 
 .. sourcecode:: ipython
 
     In [4]: dview = rc[:] # use all engines
 
 .. seealso::
 
     For more information, see the in-depth explanation of :ref:`Views <parallel_details>`.
 
 
 Quick and easy parallelism
 ==========================
 
 In many cases, you simply want to apply a Python function to a sequence of
 objects, but *in parallel*. The client interface provides a simple way
 of accomplishing this: using the DirectView's :meth:`~DirectView.map` method.
 
 Parallel map
 ------------
 
 Python's builtin :func:`map` functions allows a function to be applied to a
 sequence element-by-element. This type of code is typically trivial to
 parallelize. In fact, since IPython's interface is all about functions anyway,
 you can just use the builtin :func:`map` with a :class:`RemoteFunction`, or a 
 DirectView's :meth:`map` method:
 
 .. sourcecode:: ipython
 
     In [62]: serial_result = map(lambda x:x**10, range(32))
     
     In [63]: parallel_result = dview.map_sync(lambda x: x**10, range(32))
 
     In [67]: serial_result==parallel_result
     Out[67]: True
 
 
 .. note::
 
     The :class:`DirectView`'s version of :meth:`map` does
     not do dynamic load balancing. For a load balanced version, use a
     :class:`LoadBalancedView`.
 
 .. seealso::
     
     :meth:`map` is implemented via :class:`ParallelFunction`.
 
 Remote function decorators
 --------------------------
 
 Remote functions are just like normal functions, but when they are called,
 they execute on one or more engines, rather than locally. IPython provides
 two decorators:
 
 .. sourcecode:: ipython
 
     In [10]: @dview.remote(block=True)
         ...: def getpid():
         ...:     import os
         ...:     return os.getpid()
         ...: 
 
     In [11]: getpid()
     Out[11]: [12345, 12346, 12347, 12348]
 
 The ``@parallel`` decorator creates parallel functions, that break up an element-wise
 operations and distribute them, reconstructing the result.
 
 .. sourcecode:: ipython
 
     In [12]: import numpy as np
     
     In [13]: A = np.random.random((64,48))
     
     In [14]: @dview.parallel(block=True)
         ...: def pmul(A,B):
         ...:     return A*B
     
     In [15]: C_local = A*A
     
     In [16]: C_remote = pmul(A,A)
     
     In [17]: (C_local == C_remote).all()
     Out[17]: True
 
 .. seealso::
 
     See the docstrings for the :func:`parallel` and :func:`remote` decorators for
     options.
 
 Calling Python functions
 ========================
 
 The most basic type of operation that can be performed on the engines is to
 execute Python code or call Python functions. Executing Python code can be
 done in blocking or non-blocking mode (non-blocking is default) using the
 :meth:`.View.execute` method, and calling functions can be done via the
 :meth:`.View.apply` method.
 
 apply
 -----
 
 The main method for doing remote execution (in fact, all methods that
 communicate with the engines are built on top of it), is :meth:`View.apply`.
 
 We strive to provide the cleanest interface we can, so `apply` has the following
 signature:
 
 .. sourcecode:: python
 
     view.apply(f, *args, **kwargs)
 
 There are various ways to call functions with IPython, and these flags are set as
 attributes of the View.  The ``DirectView`` has just two of these flags:
 
 dv.block : bool
     whether to wait for the result, or return an :class:`AsyncResult` object
     immediately
 dv.track : bool
     whether to instruct pyzmq to track when 
     This is primarily useful for non-copying sends of numpy arrays that you plan to
     edit in-place.  You need to know when it becomes safe to edit the buffer
     without corrupting the message.
 
 
 Creating a view is simple: index-access on a client creates a :class:`.DirectView`.
 
 .. sourcecode:: ipython
 
     In [4]: view = rc[1:3]
     Out[4]: <DirectView [1, 2]>
 
     In [5]: view.apply<tab>
     view.apply  view.apply_async  view.apply_sync
 
 For convenience, you can set block temporarily for a single call with the extra sync/async methods.
 
 Blocking execution
 ------------------
 
 In blocking mode, the :class:`.DirectView` object (called ``dview`` in
 these examples) submits the command to the controller, which places the
 command in the engines' queues for execution. The :meth:`apply` call then
 blocks until the engines are done executing the command:
 
 .. sourcecode:: ipython
 
     In [2]: dview = rc[:] # A DirectView of all engines
     In [3]: dview.block=True
     In [4]: dview['a'] = 5
 
     In [5]: dview['b'] = 10
 
     In [6]: dview.apply(lambda x: a+b+x, 27)
     Out[6]: [42, 42, 42, 42]
 
 You can also select blocking execution on a call-by-call basis with the :meth:`apply_sync`
 method:
 
     In [7]: dview.block=False
 
     In [8]: dview.apply_sync(lambda x: a+b+x, 27)
     Out[8]: [42, 42, 42, 42]
 
 Python commands can be executed as strings on specific engines by using a View's ``execute``
 method:
 
 .. sourcecode:: ipython
 
     In [6]: rc[::2].execute('c=a+b')
 
     In [7]: rc[1::2].execute('c=a-b')
 
     In [8]: dview['c'] # shorthand for dview.pull('c', block=True)
     Out[8]: [15, -5, 15, -5]
 
 
 Non-blocking execution
 ----------------------
 
 In non-blocking mode, :meth:`apply` submits the command to be executed and
 then returns a :class:`AsyncResult` object immediately. The
 :class:`AsyncResult` object gives you a way of getting a result at a later
 time through its :meth:`get` method.
 
 .. Note::
 
     The :class:`AsyncResult` object provides a superset of the interface in 
     :py:class:`multiprocessing.pool.AsyncResult`.  See the 
     `official Python documentation <http://docs.python.org/library/multiprocessing#multiprocessing.pool.AsyncResult>`_
     for more.
 
 
 This allows you to quickly submit long running commands without blocking your
 local Python/IPython session:
 
 .. sourcecode:: ipython
     
     # define our function
     In [6]: def wait(t):
        ...:     import time
        ...:     tic = time.time()
        ...:     time.sleep(t)
        ...:     return time.time()-tic
     
     # In non-blocking mode
     In [7]: ar = dview.apply_async(wait, 2)
 
     # Now block for the result
     In [8]: ar.get()
     Out[8]: [2.0006198883056641, 1.9997570514678955, 1.9996809959411621, 2.0003249645233154]
 
     # Again in non-blocking mode
     In [9]: ar = dview.apply_async(wait, 10)
 
     # Poll to see if the result is ready
     In [10]: ar.ready()
     Out[10]: False
     
     # ask for the result, but wait a maximum of 1 second:
     In [45]: ar.get(1)
     ---------------------------------------------------------------------------
     TimeoutError                              Traceback (most recent call last)
     /home/you/<ipython-input-45-7cd858bbb8e0> in <module>()
     ----> 1 ar.get(1)
 
     /path/to/site-packages/IPython/parallel/asyncresult.pyc in get(self, timeout)
          62                 raise self._exception
          63         else:
     ---> 64             raise error.TimeoutError("Result not ready.")
          65 
          66     def ready(self):
 
     TimeoutError: Result not ready.
 
 .. Note::
 
     Note the import inside the function. This is a common model, to ensure
     that the appropriate modules are imported where the task is run. You can
     also manually import modules into the engine(s) namespace(s) via 
     :meth:`view.execute('import numpy')`.
 
 Often, it is desirable to wait until a set of :class:`AsyncResult` objects
 are done. For this, there is a the method :meth:`wait`. This method takes a
 tuple of :class:`AsyncResult` objects (or `msg_ids` or indices to the client's History),
 and blocks until all of the associated results are ready:
 
 .. sourcecode:: ipython
 
     In [72]: dview.block=False
 
     # A trivial list of AsyncResults objects
     In [73]: pr_list = [dview.apply_async(wait, 3) for i in range(10)]
 
     # Wait until all of them are done
     In [74]: dview.wait(pr_list)
 
     # Then, their results are ready using get() or the `.r` attribute
     In [75]: pr_list[0].get()
     Out[75]: [2.9982571601867676, 2.9982588291168213, 2.9987530708312988, 2.9990990161895752]
 
 
 
 The ``block`` and ``targets`` keyword arguments and attributes
 --------------------------------------------------------------
 
 Most DirectView methods (excluding :meth:`apply` and :meth:`map`) accept ``block`` and
 ``targets`` as keyword arguments. As we have seen above, these keyword arguments control the
 blocking mode and which engines the command is applied to. The :class:`View` class also has
 :attr:`block` and :attr:`targets` attributes that control the default behavior when the keyword
 arguments are not provided. Thus the following logic is used for :attr:`block` and :attr:`targets`:
 
 * If no keyword argument is provided, the instance attributes are used.
 * Keyword argument, if provided override the instance attributes for
   the duration of a single call.
   
 The following examples demonstrate how to use the instance attributes:
 
 .. sourcecode:: ipython
 
     In [16]: dview.targets = [0,2]
     
     In [17]: dview.block = False
 
     In [18]: ar = dview.apply(lambda : 10)
 
     In [19]: ar.get()
     Out[19]: [10, 10]
 
     In [16]: dview.targets = v.client.ids # all engines (4)
     
     In [21]: dview.block = True
 
     In [22]: dview.apply(lambda : 42)
     Out[22]: [42, 42, 42, 42]
 
 The :attr:`block` and :attr:`targets` instance attributes of the
 :class:`.DirectView` also determine the behavior of the parallel magic commands.
 
 Parallel magic commands
 -----------------------
 
 .. warning::
 
     The magics have not been changed to work with the zeromq system. The
     magics do work, but *do not* print stdin/out like they used to in IPython.kernel.
 
 We provide a few IPython magic commands (``%px``, ``%autopx`` and ``%result``)
 that make it more pleasant to execute Python commands on the engines
 interactively. These are simply shortcuts to :meth:`execute` and
 :meth:`get_result` of the :class:`DirectView`. The ``%px`` magic executes a single 
 Python command on the engines specified by the :attr:`targets` attribute of the
 :class:`DirectView` instance:
 
 .. sourcecode:: ipython
 
     # load the parallel magic extension:
     In [21]: %load_ext parallelmagic
     
     # Create a DirectView for all targets
     In [22]: dv = rc[:]
     
     # Make this DirectView active for parallel magic commands
     In [23]: dv.activate()
 
     In [24]: dv.block=True
 
     In [25]: import numpy
 
     In [26]: %px import numpy
     Parallel execution on engines: [0, 1, 2, 3]
 
     In [27]: %px a = numpy.random.rand(2,2)
     Parallel execution on engines: [0, 1, 2, 3]
 
     In [28]: %px ev = numpy.linalg.eigvals(a)
     Parallel execution on engines: [0, 1, 2, 3]
 
     In [28]: dv['ev']
     Out[28]: [ array([ 1.09522024, -0.09645227]), 
                array([ 1.21435496, -0.35546712]), 
                array([ 0.72180653,  0.07133042]), 
                array([  1.46384341e+00,   1.04353244e-04])
              ]
 
 The ``%result`` magic gets the most recent result, or takes an argument
 specifying the index of the result to be requested. It is simply a shortcut to the
 :meth:`get_result` method:
 
 .. sourcecode:: ipython
     
     In [29]: dv.apply_async(lambda : ev)
     
     In [30]: %result
     Out[30]: [ [ 1.28167017  0.14197338],
                 [-0.14093616  1.27877273],
                 [-0.37023573  1.06779409],
                 [ 0.83664764 -0.25602658] ]
 
 The ``%autopx`` magic switches to a mode where everything you type is executed
 on the engines given by the :attr:`targets` attribute:
 
 .. sourcecode:: ipython
 
     In [30]: dv.block=False
 
     In [31]: %autopx
     Auto Parallel Enabled
     Type %autopx to disable
 
     In [32]: max_evals = []
     <IPython.parallel.AsyncResult object at 0x17b8a70>
 
     In [33]: for i in range(100):
        ....:     a = numpy.random.rand(10,10)
        ....:     a = a+a.transpose()
        ....:     evals = numpy.linalg.eigvals(a)
        ....:     max_evals.append(evals[0].real)
        ....:
        ....:
     <IPython.parallel.AsyncResult object at 0x17af8f0>
 
     In [34]: %autopx
     Auto Parallel Disabled
 
     In [35]: dv.block=True
 
     In [36]: px ans= "Average max eigenvalue is: %f"%(sum(max_evals)/len(max_evals))
     Parallel execution on engines: [0, 1, 2, 3]
     
     In [37]: dv['ans']
     Out[37]: [ 'Average max eigenvalue is:  10.1387247332',
                'Average max eigenvalue is:  10.2076902286',
                'Average max eigenvalue is:  10.1891484655',
                'Average max eigenvalue is:  10.1158837784',]
 
 
 Moving Python objects around
 ============================
 
 In addition to calling functions and executing code on engines, you can
 transfer Python objects to and from your IPython session and the engines. In
 IPython, these operations are called :meth:`push` (sending an object to the
 engines) and :meth:`pull` (getting an object from the engines).
 
 Basic push and pull
 -------------------
 
 Here are some examples of how you use :meth:`push` and :meth:`pull`:
 
 .. sourcecode:: ipython
 
     In [38]: dview.push(dict(a=1.03234,b=3453))
     Out[38]: [None,None,None,None]
 
     In [39]: dview.pull('a')
     Out[39]: [ 1.03234, 1.03234, 1.03234, 1.03234]
 
     In [40]: dview.pull('b', targets=0)
     Out[40]: 3453
 
     In [41]: dview.pull(('a','b'))
     Out[41]: [ [1.03234, 3453], [1.03234, 3453], [1.03234, 3453], [1.03234, 3453] ]
     
     In [43]: dview.push(dict(c='speed'))
     Out[43]: [None,None,None,None]
 
 In non-blocking mode :meth:`push` and :meth:`pull` also return
 :class:`AsyncResult` objects:
 
 .. sourcecode:: ipython
 
     In [48]: ar = dview.pull('a', block=False)
 
     In [49]: ar.get()
     Out[49]: [1.03234, 1.03234, 1.03234, 1.03234]
 
 
 Dictionary interface
 --------------------
 
 Since a Python namespace is just a :class:`dict`, :class:`DirectView` objects provide
 dictionary-style access by key and methods such as :meth:`get` and
 :meth:`update` for convenience. This make the remote namespaces of the engines
 appear as a local dictionary. Underneath, these methods call :meth:`apply`:
 
 .. sourcecode:: ipython
 
     In [51]: dview['a']=['foo','bar']
 
     In [52]: dview['a']
     Out[52]: [ ['foo', 'bar'], ['foo', 'bar'], ['foo', 'bar'], ['foo', 'bar'] ]
 
 Scatter and gather
 ------------------
 
 Sometimes it is useful to partition a sequence and push the partitions to
 different engines. In MPI language, this is know as scatter/gather and we
 follow that terminology. However, it is important to remember that in
 IPython's :class:`Client` class, :meth:`scatter` is from the
 interactive IPython session to the engines and :meth:`gather` is from the
 engines back to the interactive IPython session. For scatter/gather operations
 between engines, MPI should be used:
 
 .. sourcecode:: ipython
 
     In [58]: dview.scatter('a',range(16))
     Out[58]: [None,None,None,None]
 
     In [59]: dview['a']
     Out[59]: [ [0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11], [12, 13, 14, 15] ]
 
     In [60]: dview.gather('a')
     Out[60]: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15]
 
 Other things to look at
 =======================
 
 How to do parallel list comprehensions
 --------------------------------------
 
 In many cases list comprehensions are nicer than using the map function. While
 we don't have fully parallel list comprehensions, it is simple to get the
 basic effect using :meth:`scatter` and :meth:`gather`:
 
 .. sourcecode:: ipython
 
     In [66]: dview.scatter('x',range(64))
 
     In [67]: %px y = [i**10 for i in x]
     Parallel execution on engines: [0, 1, 2, 3]
     Out[67]:
 
     In [68]: y = dview.gather('y')
 
     In [69]: print y
     [0, 1, 1024, 59049, 1048576, 9765625, 60466176, 282475249, 1073741824,...]
 
 Remote imports
 --------------
 
 Sometimes you will want to import packages both in your interactive session
 and on your remote engines.  This can be done with the :class:`ContextManager`
 created by a DirectView's :meth:`sync_imports` method:
 
 .. sourcecode:: ipython
 
     In [69]: with dview.sync_imports():
         ...:     import numpy
     importing numpy on engine(s)
 
 Any imports made inside the block will also be performed on the view's engines.
 sync_imports also takes a `local` boolean flag that defaults to True, which specifies
 whether the local imports should also be performed.  However, support for `local=False`
 has not been implemented, so only packages that can be imported locally will work
 this way.
 
 You can also specify imports via the ``@require`` decorator.  This is a decorator
 designed for use in Dependencies, but can be used to handle remote imports as well.
 Modules or module names passed to ``@require`` will be imported before the decorated
 function is called.  If they cannot be imported, the decorated function will never
 execution, and will fail with an UnmetDependencyError.
 
 .. sourcecode:: ipython
 
     In [69]: from IPython.parallel import require
 
     In [70]: @requre('re'):
         ...: def findall(pat, x):
         ...:     # re is guaranteed to be available
         ...:     return re.findall(pat, x)
           
     # you can also pass modules themselves, that you already have locally:
     In [71]: @requre(time):
         ...: def wait(t):
         ...:     time.sleep(t)
         ...:     return t
 
 .. _parallel_exceptions:
 
 Parallel exceptions
 -------------------
 
 In the multiengine interface, parallel commands can raise Python exceptions,
 just like serial commands. But, it is a little subtle, because a single
 parallel command can actually raise multiple exceptions (one for each engine
 the command was run on). To express this idea, we have a
 :exc:`CompositeError` exception class that will be raised in most cases. The
 :exc:`CompositeError` class is a special type of exception that wraps one or
 more other types of exceptions. Here is how it works:
 
 .. sourcecode:: ipython
 
     In [76]: dview.block=True
 
     In [77]: dview.execute('1/0')
     ---------------------------------------------------------------------------
     CompositeError                            Traceback (most recent call last)
     /home/user/<ipython-input-10-5d56b303a66c> in <module>()
     ----> 1 dview.execute('1/0')
 
     /path/to/site-packages/IPython/parallel/client/view.pyc in execute(self, code, targets, block)
         591                 default: self.block
         592         """
     --> 593         return self._really_apply(util._execute, args=(code,), block=block, targets=targets)
         594 
         595     def run(self, filename, targets=None, block=None):
 
     /home/user/<string> in _really_apply(self, f, args, kwargs, targets, block, track)
 
     /path/to/site-packages/IPython/parallel/client/view.pyc in sync_results(f, self, *args, **kwargs)
          55 def sync_results(f, self, *args, **kwargs):
          56     """sync relevant results from self.client to our results attribute."""
     ---> 57     ret = f(self, *args, **kwargs)
          58     delta = self.outstanding.difference(self.client.outstanding)
          59     completed = self.outstanding.intersection(delta)
 
     /home/user/<string> in _really_apply(self, f, args, kwargs, targets, block, track)
 
     /path/to/site-packages/IPython/parallel/client/view.pyc in save_ids(f, self, *args, **kwargs)
          44     n_previous = len(self.client.history)
          45     try:
     ---> 46         ret = f(self, *args, **kwargs)
          47     finally:
          48         nmsgs = len(self.client.history) - n_previous
 
     /path/to/site-packages/IPython/parallel/client/view.pyc in _really_apply(self, f, args, kwargs, targets, block, track)
         529         if block:
         530             try:
     --> 531                 return ar.get()
         532             except KeyboardInterrupt:
         533                 pass
 
     /path/to/site-packages/IPython/parallel/client/asyncresult.pyc in get(self, timeout)
         101                 return self._result
         102             else:
     --> 103                 raise self._exception
         104         else:
         105             raise error.TimeoutError("Result not ready.")
 
     CompositeError: one or more exceptions from call to method: _execute
     [0:apply]: ZeroDivisionError: integer division or modulo by zero
     [1:apply]: ZeroDivisionError: integer division or modulo by zero
     [2:apply]: ZeroDivisionError: integer division or modulo by zero
     [3:apply]: ZeroDivisionError: integer division or modulo by zero
 
 Notice how the error message printed when :exc:`CompositeError` is raised has
 information about the individual exceptions that were raised on each engine.
 If you want, you can even raise one of these original exceptions:
 
 .. sourcecode:: ipython
 
     In [80]: try:
        ....:     dview.execute('1/0')
        ....: except parallel.error.CompositeError, e:
        ....:     e.raise_exception()
        ....:     
        ....:     
     ---------------------------------------------------------------------------
     RemoteError                               Traceback (most recent call last)
     /home/user/<ipython-input-17-8597e7e39858> in <module>()
           2     dview.execute('1/0')
           3 except CompositeError as e:
     ----> 4     e.raise_exception()
 
     /path/to/site-packages/IPython/parallel/error.pyc in raise_exception(self, excid)
         266             raise IndexError("an exception with index %i does not exist"%excid)
         267         else:
     --> 268             raise RemoteError(en, ev, etb, ei)
         269 
         270 
 
     RemoteError: ZeroDivisionError(integer division or modulo by zero)
     Traceback (most recent call last):
       File "/path/to/site-packages/IPython/parallel/engine/streamkernel.py", line 330, in apply_request
         exec code in working,working
       File "<string>", line 1, in <module>
       File "/path/to/site-packages/IPython/parallel/util.py", line 354, in _execute
         exec code in globals()
       File "<string>", line 1, in <module>
     ZeroDivisionError: integer division or modulo by zero
 
 If you are working in IPython, you can simple type ``%debug`` after one of
 these :exc:`CompositeError` exceptions is raised, and inspect the exception
 instance:
 
 .. sourcecode:: ipython
 
     In [81]: dview.execute('1/0')
     ---------------------------------------------------------------------------
     CompositeError                            Traceback (most recent call last)
     /home/user/<ipython-input-10-5d56b303a66c> in <module>()
     ----> 1 dview.execute('1/0')
 
     /path/to/site-packages/IPython/parallel/client/view.pyc in execute(self, code, targets, block)
         591                 default: self.block
         592         """
     --> 593         return self._really_apply(util._execute, args=(code,), block=block, targets=targets)
         594 
         595     def run(self, filename, targets=None, block=None):
 
     /home/user/<string> in _really_apply(self, f, args, kwargs, targets, block, track)
 
     /path/to/site-packages/IPython/parallel/client/view.pyc in sync_results(f, self, *args, **kwargs)
          55 def sync_results(f, self, *args, **kwargs):
          56     """sync relevant results from self.client to our results attribute."""
     ---> 57     ret = f(self, *args, **kwargs)
          58     delta = self.outstanding.difference(self.client.outstanding)
          59     completed = self.outstanding.intersection(delta)
 
     /home/user/<string> in _really_apply(self, f, args, kwargs, targets, block, track)
 
     /path/to/site-packages/IPython/parallel/client/view.pyc in save_ids(f, self, *args, **kwargs)
          44     n_previous = len(self.client.history)
          45     try:
     ---> 46         ret = f(self, *args, **kwargs)
          47     finally:
          48         nmsgs = len(self.client.history) - n_previous
 
     /path/to/site-packages/IPython/parallel/client/view.pyc in _really_apply(self, f, args, kwargs, targets, block, track)
         529         if block:
         530             try:
     --> 531                 return ar.get()
         532             except KeyboardInterrupt:
         533                 pass
 
     /path/to/site-packages/IPython/parallel/client/asyncresult.pyc in get(self, timeout)
         101                 return self._result
         102             else:
     --> 103                 raise self._exception
         104         else:
         105             raise error.TimeoutError("Result not ready.")
 
     CompositeError: one or more exceptions from call to method: _execute
     [0:apply]: ZeroDivisionError: integer division or modulo by zero
     [1:apply]: ZeroDivisionError: integer division or modulo by zero
     [2:apply]: ZeroDivisionError: integer division or modulo by zero
     [3:apply]: ZeroDivisionError: integer division or modulo by zero
     
     In [82]: %debug
     > /path/to/site-packages/IPython/parallel/client/asyncresult.py(103)get()
         102             else:
     --> 103                 raise self._exception
         104         else:
     
     # With the debugger running, self._exception is the exceptions instance.  We can tab complete
     # on it and see the extra methods that are available.
     ipdb> self._exception.<tab>
     e.__class__         e.__getitem__       e.__new__           e.__setstate__      e.args
     e.__delattr__       e.__getslice__      e.__reduce__        e.__str__           e.elist
     e.__dict__          e.__hash__          e.__reduce_ex__     e.__weakref__       e.message
     e.__doc__           e.__init__          e.__repr__          e._get_engine_str   e.print_tracebacks
     e.__getattribute__  e.__module__        e.__setattr__       e._get_traceback    e.raise_exception
     ipdb> self._exception.print_tracebacks()
     [0:apply]: 
     Traceback (most recent call last):
       File "/path/to/site-packages/IPython/parallel/engine/streamkernel.py", line 330, in apply_request
         exec code in working,working
       File "<string>", line 1, in <module>
       File "/path/to/site-packages/IPython/parallel/util.py", line 354, in _execute
         exec code in globals()
       File "<string>", line 1, in <module>
     ZeroDivisionError: integer division or modulo by zero
 
 
     [1:apply]: 
     Traceback (most recent call last):
       File "/path/to/site-packages/IPython/parallel/engine/streamkernel.py", line 330, in apply_request
         exec code in working,working
       File "<string>", line 1, in <module>
       File "/path/to/site-packages/IPython/parallel/util.py", line 354, in _execute
         exec code in globals()
       File "<string>", line 1, in <module>
     ZeroDivisionError: integer division or modulo by zero
 
 
     [2:apply]: 
     Traceback (most recent call last):
       File "/path/to/site-packages/IPython/parallel/engine/streamkernel.py", line 330, in apply_request
         exec code in working,working
       File "<string>", line 1, in <module>
       File "/path/to/site-packages/IPython/parallel/util.py", line 354, in _execute
         exec code in globals()
       File "<string>", line 1, in <module>
     ZeroDivisionError: integer division or modulo by zero
 
 
     [3:apply]: 
     Traceback (most recent call last):
       File "/path/to/site-packages/IPython/parallel/engine/streamkernel.py", line 330, in apply_request
         exec code in working,working
       File "<string>", line 1, in <module>
       File "/path/to/site-packages/IPython/parallel/util.py", line 354, in _execute
         exec code in globals()
       File "<string>", line 1, in <module>
     ZeroDivisionError: integer division or modulo by zero
 
 
 All of this same error handling magic even works in non-blocking mode:
 
 .. sourcecode:: ipython
 
     In [83]: dview.block=False
 
     In [84]: ar = dview.execute('1/0')
 
     In [85]: ar.get()
     ---------------------------------------------------------------------------
     CompositeError                            Traceback (most recent call last)
     /home/user/<ipython-input-21-8531eb3d26fb> in <module>()
     ----> 1 ar.get()
 
     /path/to/site-packages/IPython/parallel/client/asyncresult.pyc in get(self, timeout)
         101                 return self._result
         102             else:
     --> 103                 raise self._exception
         104         else:
         105             raise error.TimeoutError("Result not ready.")
 
     CompositeError: one or more exceptions from call to method: _execute
     [0:apply]: ZeroDivisionError: integer division or modulo by zero
     [1:apply]: ZeroDivisionError: integer division or modulo by zero
     [2:apply]: ZeroDivisionError: integer division or modulo by zero
     [3:apply]: ZeroDivisionError: integer division or modulo by zero
 

docs/source/parallel/parallel_process.txt

@@ -1,691 +1,691 b''
 .. _parallel_process:
 
 ===========================================
 Starting the IPython controller and engines
 ===========================================
 
 To use IPython for parallel computing, you need to start one instance of
 the controller and one or more instances of the engine. The controller
 and each engine can run on different machines or on the same machine.
 Because of this, there are many different possibilities.
 
 Broadly speaking, there are two ways of going about starting a controller and engines:
 
 * In an automated manner using the :command:`ipcluster` command.
 * In a more manual way using the :command:`ipcontroller` and 
   :command:`ipengine` commands.
 
 This document describes both of these methods. We recommend that new users
 start with the :command:`ipcluster` command as it simplifies many common usage
 cases.
 
 General considerations
 ======================
 
 Before delving into the details about how you can start a controller and
 engines using the various methods, we outline some of the general issues that
 come up when starting the controller and engines. These things come up no
 matter which method you use to start your IPython cluster.
 
 If you are running engines on multiple machines, you will likely need to instruct the
 controller to listen for connections on an external interface. This can be done by specifying
 the ``ip`` argument on the command-line, or the ``HubFactory.ip`` configurable in
 :file:`ipcontroller_config.py`.
 
 If your machines are on a trusted network, you can safely instruct the controller to listen
 on all public interfaces with::
 
-    $> ipcontroller ip=*
+    $> ipcontroller --ip=*
 
 Or you can set the same behavior as the default by adding the following line to your :file:`ipcontroller_config.py`:
 
 .. sourcecode:: python
 
     c.HubFactory.ip = '*'
 
 .. note::
 
     Due to the lack of security in ZeroMQ, the controller will only listen for connections on
     localhost by default. If you see Timeout errors on engines or clients, then the first
     thing you should check is the ip address the controller is listening on, and make sure
     that it is visible from the timing out machine.
     
 .. seealso::
 
     Our `notes <parallel_security>`_ on security in the new parallel computing code.
 
 Let's say that you want to start the controller on ``host0`` and engines on
 hosts ``host1``-``hostn``. The following steps are then required:
 
 1. Start the controller on ``host0`` by running :command:`ipcontroller` on
    ``host0``.  The controller must be instructed to listen on an interface visible
    to the engine machines, via the ``ip`` command-line argument or ``HubFactory.ip``
    in :file:`ipcontroller_config.py`.
 2. Move the JSON file (:file:`ipcontroller-engine.json`) created by the 
    controller from ``host0`` to hosts ``host1``-``hostn``.
 3. Start the engines on hosts ``host1``-``hostn`` by running 
    :command:`ipengine`.  This command has to be told where the JSON file 
    (:file:`ipcontroller-engine.json`) is located.
 
 At this point, the controller and engines will be connected. By default, the JSON files
 created by the controller are put into the :file:`~/.ipython/profile_default/security`
 directory. If the engines share a filesystem with the controller, step 2 can be skipped as
 the engines will automatically look at that location.
 
 The final step required to actually use the running controller from a client is to move
 the JSON file :file:`ipcontroller-client.json` from ``host0`` to any host where clients
 will be run. If these file are put into the :file:`~/.ipython/profile_default/security`
 directory of the client's host, they will be found automatically. Otherwise, the full path
 to them has to be passed to the client's constructor.
 
 Using :command:`ipcluster`
 ===========================
 
 The :command:`ipcluster` command provides a simple way of starting a
 controller and engines in the following situations:
 
 1. When the controller and engines are all run on localhost. This is useful
    for testing or running on a multicore computer.
 2. When engines are started using the :command:`mpiexec` command that comes 
    with most MPI [MPI]_ implementations
 3. When engines are started using the PBS [PBS]_ batch system 
    (or other `qsub` systems, such as SGE).
 4. When the controller is started on localhost and the engines are started on 
    remote nodes using :command:`ssh`.
 5. When engines are started using the Windows HPC Server batch system.
 
 .. note::
 
     Currently :command:`ipcluster` requires that the
     :file:`~/.ipython/profile_<name>/security` directory live on a shared filesystem that is
     seen by both the controller and engines. If you don't have a shared file
     system you will need to use :command:`ipcontroller` and
     :command:`ipengine` directly.
 
 Under the hood, :command:`ipcluster` just uses :command:`ipcontroller`
 and :command:`ipengine` to perform the steps described above.
 
 The simplest way to use ipcluster requires no configuration, and will
 launch a controller and a number of engines on the local machine. For instance, 
 to start one controller and 4 engines on localhost, just do::
 
-    $ ipcluster start n=4
+    $ ipcluster start --n=4
 
 To see other command line options, do::
 
     $ ipcluster -h
 
 
 Configuring an IPython cluster
 ==============================
 
 Cluster configurations are stored as `profiles`.  You can create a new profile with::
 
-    $ ipython profile create --parallel profile=myprofile
+    $ ipython profile create --parallel --profile=myprofile
 
 This will create the directory :file:`IPYTHONDIR/profile_myprofile`, and populate it
 with the default configuration files for the three IPython cluster commands. Once
 you edit those files, you can continue to call ipcluster/ipcontroller/ipengine
 with no arguments beyond ``profile=myprofile``, and any configuration will be maintained.
 
 There is no limit to the number of profiles you can have, so you can maintain a profile for each
 of your common use cases. The default profile will be used whenever the
 profile argument is not specified, so edit :file:`IPYTHONDIR/profile_default/*_config.py` to
 represent your most common use case.
 
 The configuration files are loaded with commented-out settings and explanations,
 which should cover most of the available possibilities.
 
 Using various batch systems with :command:`ipcluster`
 -----------------------------------------------------
 
 :command:`ipcluster` has a notion of Launchers that can start controllers
 and engines with various remote execution schemes.  Currently supported
 models include :command:`ssh`, :command:`mpiexec`, PBS-style (Torque, SGE),
 and Windows HPC Server.
 
 .. note::
 
     The Launchers and configuration are designed in such a way that advanced
     users can subclass and configure them to fit their own system that we
     have not yet supported (such as Condor)
 
 Using :command:`ipcluster` in mpiexec/mpirun mode
 --------------------------------------------------
 
 
 The mpiexec/mpirun mode is useful if you:
 
 1. Have MPI installed.
 2. Your systems are configured to use the :command:`mpiexec` or
    :command:`mpirun` commands to start MPI processes.
 
 If these are satisfied, you can create a new profile::
 
-    $ ipython profile create --parallel profile=mpi
+    $ ipython profile create --parallel --profile=mpi
 
 and edit the file :file:`IPYTHONDIR/profile_mpi/ipcluster_config.py`.
 
 There, instruct ipcluster to use the MPIExec launchers by adding the lines:
 
 .. sourcecode:: python
 
     c.IPClusterEngines.engine_launcher = 'IPython.parallel.apps.launcher.MPIExecEngineSetLauncher'
 
 If the default MPI configuration is correct, then you can now start your cluster, with::
 
-    $ ipcluster start n=4 profile=mpi
+    $ ipcluster start --n=4 --profile=mpi
 
 This does the following:
 
 1. Starts the IPython controller on current host.
 2. Uses :command:`mpiexec` to start 4 engines.
 
 If you have a reason to also start the Controller with mpi, you can specify:
 
 .. sourcecode:: python
 
     c.IPClusterStart.controller_launcher = 'IPython.parallel.apps.launcher.MPIExecControllerLauncher'
 
 .. note::
 
     The Controller *will not* be in the same MPI universe as the engines, so there is not
     much reason to do this unless sysadmins demand it.
 
 On newer MPI implementations (such as OpenMPI), this will work even if you
 don't make any calls to MPI or call :func:`MPI_Init`. However, older MPI
 implementations actually require each process to call :func:`MPI_Init` upon
 starting. The easiest way of having this done is to install the mpi4py
 [mpi4py]_ package and then specify the ``c.MPI.use`` option in :file:`ipengine_config.py`:
 
 .. sourcecode:: python
 
     c.MPI.use = 'mpi4py'
 
 Unfortunately, even this won't work for some MPI implementations. If you are
 having problems with this, you will likely have to use a custom Python
 executable that itself calls :func:`MPI_Init` at the appropriate time.
 Fortunately, mpi4py comes with such a custom Python executable that is easy to
 install and use. However, this custom Python executable approach will not work
 with :command:`ipcluster` currently.
 
 More details on using MPI with IPython can be found :ref:`here <parallelmpi>`.
 
 
 Using :command:`ipcluster` in PBS mode
 ---------------------------------------
 
 The PBS mode uses the Portable Batch System (PBS) to start the engines. 
 
 As usual, we will start by creating a fresh profile::
 
-    $ ipython profile create --parallel profile=pbs
+    $ ipython profile create --parallel --profile=pbs
 
 And in :file:`ipcluster_config.py`, we will select the PBS launchers for the controller
 and engines:
 
 .. sourcecode:: python
 
     c.IPClusterStart.controller_launcher = \
         'IPython.parallel.apps.launcher.PBSControllerLauncher'
     c.IPClusterEngines.engine_launcher = \
         'IPython.parallel.apps.launcher.PBSEngineSetLauncher'
 
 .. note::
 
     Note that the configurable is IPClusterEngines for the engine launcher, and
     IPClusterStart for the controller launcher. This is because the start command is a
     subclass of the engine command, adding a controller launcher. Since it is a subclass,
     any configuration made in IPClusterEngines is inherited by IPClusterStart unless it is
     overridden.
 
 IPython does provide simple default batch templates for PBS and SGE, but you may need
 to specify your own. Here is a sample PBS script template:
 
 .. sourcecode:: bash
 
     #PBS -N ipython
     #PBS -j oe
     #PBS -l walltime=00:10:00
     #PBS -l nodes={n/4}:ppn=4
     #PBS -q {queue}
 
     cd $PBS_O_WORKDIR
     export PATH=$HOME/usr/local/bin
     export PYTHONPATH=$HOME/usr/local/lib/python2.7/site-packages
-    /usr/local/bin/mpiexec -n {n} ipengine profile_dir={profile_dir}
+    /usr/local/bin/mpiexec -n {n} ipengine --profile_dir={profile_dir}
 
 There are a few important points about this template:
 
 1. This template will be rendered at runtime using IPython's :class:`EvalFormatter`.
    This is simply a subclass of :class:`string.Formatter` that allows simple expressions
    on keys.
 
 2. Instead of putting in the actual number of engines, use the notation
    ``{n}`` to indicate the number of engines to be started. You can also use
    expressions like ``{n/4}`` in the template to indicate the number of nodes.
    There will always be ``{n}`` and ``{profile_dir}`` variables passed to the formatter.
    These allow the batch system to know how many engines, and where the configuration
    files reside. The same is true for the batch queue, with the template variable
    ``{queue}``.
 
 3. Any options to :command:`ipengine` can be given in the batch script
    template, or in :file:`ipengine_config.py`.
 
 4. Depending on the configuration of you system, you may have to set
    environment variables in the script template.
 
 The controller template should be similar, but simpler:
 
 .. sourcecode:: bash
 
     #PBS -N ipython
     #PBS -j oe
     #PBS -l walltime=00:10:00
     #PBS -l nodes=1:ppn=4
     #PBS -q {queue}
 
     cd $PBS_O_WORKDIR
     export PATH=$HOME/usr/local/bin
     export PYTHONPATH=$HOME/usr/local/lib/python2.7/site-packages
-    ipcontroller profile_dir={profile_dir}
+    ipcontroller --profile_dir={profile_dir}
 
 
 Once you have created these scripts, save them with names like
 :file:`pbs.engine.template`. Now you can load them into the :file:`ipcluster_config` with:
 
 .. sourcecode:: python
 
     c.PBSEngineSetLauncher.batch_template_file = "pbs.engine.template"
     
     c.PBSControllerLauncher.batch_template_file = "pbs.controller.template"
         
 
 Alternately, you can just define the templates as strings inside :file:`ipcluster_config`.
 
 Whether you are using your own templates or our defaults, the extra configurables available are
 the number of engines to launch (``{n}``, and the batch system queue to which the jobs are to be
 submitted (``{queue}``)). These are configurables, and can be specified in
 :file:`ipcluster_config`:
 
 .. sourcecode:: python
 
     c.PBSLauncher.queue = 'veryshort.q'
     c.IPClusterEngines.n = 64
 
 Note that assuming you are running PBS on a multi-node cluster, the Controller's default behavior
 of listening only on localhost is likely too restrictive.  In this case, also assuming the
 nodes are safely behind a firewall, you can simply instruct the Controller to listen for
 connections on all its interfaces, by adding in :file:`ipcontroller_config`:
 
 .. sourcecode:: python
 
     c.HubFactory.ip = '*'
 
 You can now run the cluster with::
 
-    $ ipcluster start profile=pbs n=128
+    $ ipcluster start --profile=pbs --n=128
 
 Additional configuration options can be found in the PBS section of :file:`ipcluster_config`.
 
 .. note::
 
     Due to the flexibility of configuration, the PBS launchers work with simple changes
     to the template for other :command:`qsub`-using systems, such as Sun Grid Engine,
     and with further configuration in similar batch systems like Condor.
 
 
 Using :command:`ipcluster` in SSH mode
 ---------------------------------------
 
 
 The SSH mode uses :command:`ssh` to execute :command:`ipengine` on remote
 nodes and :command:`ipcontroller` can be run remotely as well, or on localhost.
 
 .. note::
 
     When using this mode it highly recommended that you have set up SSH keys
     and are using ssh-agent [SSH]_ for password-less logins.
 
 As usual, we start by creating a clean profile::
 
-    $ ipython profile create --parallel profile=ssh
+    $ ipython profile create --parallel --profile=ssh
 
 To use this mode, select the SSH launchers in :file:`ipcluster_config.py`:
 
 .. sourcecode:: python
 
     c.IPClusterEngines.engine_launcher = \
         'IPython.parallel.apps.launcher.SSHEngineSetLauncher'
     # and if the Controller is also to be remote:
     c.IPClusterStart.controller_launcher = \
         'IPython.parallel.apps.launcher.SSHControllerLauncher'
     
 
 The controller's remote location and configuration can be specified:
 
 .. sourcecode:: python
 
     # Set the user and hostname for the controller
     # c.SSHControllerLauncher.hostname = 'controller.example.com'
     # c.SSHControllerLauncher.user = os.environ.get('USER','username')
 
     # Set the arguments to be passed to ipcontroller
     # note that remotely launched ipcontroller will not get the contents of 
     # the local ipcontroller_config.py unless it resides on the *remote host*
     # in the location specified by the `profile_dir` argument.
-    # c.SSHControllerLauncher.program_args = ['--reuse', 'ip=*', 'profile_dir=/path/to/cd']
+    # c.SSHControllerLauncher.program_args = ['--reuse', '--ip=*', '--profile_dir=/path/to/cd']
 
 .. note::
 
     SSH mode does not do any file movement, so you will need to distribute configuration
     files manually.  To aid in this, the `reuse_files` flag defaults to True for ssh-launched
     Controllers, so you will only need to do this once, unless you override this flag back
     to False.
 
 Engines are specified in a dictionary, by hostname and the number of engines to be run
 on that host.
 
 .. sourcecode:: python
 
     c.SSHEngineSetLauncher.engines = { 'host1.example.com' : 2,
                 'host2.example.com' : 5,
-                'host3.example.com' : (1, ['profile_dir=/home/different/location']),
+                'host3.example.com' : (1, ['--profile_dir=/home/different/location']),
                 'host4.example.com' : 8 }
 
 * The `engines` dict, where the keys are the host we want to run engines on and
   the value is the number of engines to run on that host.
 * on host3, the value is a tuple, where the number of engines is first, and the arguments
   to be passed to :command:`ipengine` are the second element.
 
 For engines without explicitly specified arguments, the default arguments are set in
 a single location:
 
 .. sourcecode:: python
 
-    c.SSHEngineSetLauncher.engine_args = ['profile_dir=/path/to/profile_ssh']
+    c.SSHEngineSetLauncher.engine_args = ['--profile_dir=/path/to/profile_ssh']
 
 Current limitations of the SSH mode of :command:`ipcluster` are:
 
 * Untested on Windows. Would require a working :command:`ssh` on Windows.
   Also, we are using shell scripts to setup and execute commands on remote
   hosts.
 * No file movement - This is a regression from 0.10, which moved connection files
   around with scp. This will be improved, but not before 0.11 release.
 
 Using the :command:`ipcontroller` and :command:`ipengine` commands
 ====================================================================
 
 It is also possible to use the :command:`ipcontroller` and :command:`ipengine`
 commands to start your controller and engines. This approach gives you full
 control over all aspects of the startup process.
 
 Starting the controller and engine on your local machine
 --------------------------------------------------------
 
 To use :command:`ipcontroller` and :command:`ipengine` to start things on your
 local machine, do the following.
 
 First start the controller::
 
     $ ipcontroller
     
 Next, start however many instances of the engine you want using (repeatedly)
 the command::
 
     $ ipengine
 
 The engines should start and automatically connect to the controller using the
 JSON files in :file:`~/.ipython/profile_default/security`. You are now ready to use the
 controller and engines from IPython.
 
 .. warning:: 
     
     The order of the above operations may be important.  You *must*
     start the controller before the engines, unless you are reusing connection
     information (via ``--reuse``), in which case ordering is not important.
 
 .. note::
 
     On some platforms (OS X), to put the controller and engine into the
     background you may need to give these commands in the form ``(ipcontroller
     &)`` and ``(ipengine &)`` (with the parentheses) for them to work
     properly.
 
 Starting the controller and engines on different hosts
 ------------------------------------------------------
 
 When the controller and engines are running on different hosts, things are
 slightly more complicated, but the underlying ideas are the same:
 
 1. Start the controller on a host using :command:`ipcontroller`. The controller must be
    instructed to listen on an interface visible to the engine machines, via the ``ip``
    command-line argument or ``HubFactory.ip`` in :file:`ipcontroller_config.py`.
 2. Copy :file:`ipcontroller-engine.json` from :file:`~/.ipython/profile_<name>/security` on
    the controller's host to the host where the engines will run.
 3. Use :command:`ipengine` on the engine's hosts to start the engines.
 
 The only thing you have to be careful of is to tell :command:`ipengine` where
 the :file:`ipcontroller-engine.json` file is located. There are two ways you
 can do this:
 
 * Put :file:`ipcontroller-engine.json` in the :file:`~/.ipython/profile_<name>/security`
   directory on the engine's host, where it will be found automatically.
-* Call :command:`ipengine` with the ``file=full_path_to_the_file``
+* Call :command:`ipengine` with the ``--file=full_path_to_the_file``
   flag.
 
 The ``file`` flag works like this::
 
-    $ ipengine file=/path/to/my/ipcontroller-engine.json
+    $ ipengine --file=/path/to/my/ipcontroller-engine.json
 
 .. note::
     
     If the controller's and engine's hosts all have a shared file system  
     (:file:`~/.ipython/profile_<name>/security` is the same on all of them), then things
     will just work!
 
 Make JSON files persistent
 --------------------------
 
 At fist glance it may seem that that managing the JSON files is a bit
 annoying. Going back to the house and key analogy, copying the JSON around
 each time you start the controller is like having to make a new key every time
 you want to unlock the door and enter your house. As with your house, you want
 to be able to create the key (or JSON file) once, and then simply use it at
 any point in the future.
 
 To do this, the only thing you have to do is specify the `--reuse` flag, so that
 the connection information in the JSON files remains accurate::
 
     $ ipcontroller --reuse
 
 Then, just copy the JSON files over the first time and you are set. You can
 start and stop the controller and engines any many times as you want in the
 future, just make sure to tell the controller to reuse the file.
 
 .. note::
 
     You may ask the question: what ports does the controller listen on if you
     don't tell is to use specific ones? The default is to use high random port
     numbers. We do this for two reasons: i) to increase security through
     obscurity and ii) to multiple controllers on a given host to start and
     automatically use different ports.
 
 Log files
 ---------
 
 All of the components of IPython have log files associated with them.
 These log files can be extremely useful in debugging problems with
 IPython and can be found in the directory :file:`~/.ipython/profile_<name>/log`.
 Sending the log files to us will often help us to debug any problems.
 
 
 Configuring `ipcontroller`
 ---------------------------
 
 The IPython Controller takes its configuration from the file :file:`ipcontroller_config.py`
 in the active profile directory.
 
 Ports and addresses
 *******************
 
 In many cases, you will want to configure the Controller's network identity.  By default,
 the Controller listens only on loopback, which is the most secure but often impractical.
 To instruct the controller to listen on a specific interface, you can set the 
 :attr:`HubFactory.ip` trait.  To listen on all interfaces, simply specify:
 
 .. sourcecode:: python
 
     c.HubFactory.ip = '*'
 
 When connecting to a Controller that is listening on loopback or behind a firewall, it may
 be necessary to specify an SSH server to use for tunnels, and the external IP of the
 Controller. If you specified that the HubFactory listen on loopback, or all interfaces,
 then IPython will try to guess the external IP. If you are on a system with VM network
 devices, or many interfaces, this guess may be incorrect. In these cases, you will want
 to specify the 'location' of the Controller. This is the IP of the machine the Controller
 is on, as seen by the clients, engines, or the SSH server used to tunnel connections.
 
 For example, to set up a cluster with a Controller on a work node, using ssh tunnels
 through the login node, an example :file:`ipcontroller_config.py` might contain:
 
 .. sourcecode:: python
 
     # allow connections on all interfaces from engines
     # engines on the same node will use loopback, while engines
     # from other nodes will use an external IP
     c.HubFactory.ip = '*'
     
     # you typically only need to specify the location when there are extra
     # interfaces that may not be visible to peer nodes (e.g. VM interfaces)
     c.HubFactory.location = '10.0.1.5'
     # or to get an automatic value, try this:
     import socket
     ex_ip = socket.gethostbyname_ex(socket.gethostname())[-1][0]
     c.HubFactory.location = ex_ip
     
     # now instruct clients to use the login node for SSH tunnels:
     c.HubFactory.ssh_server = 'login.mycluster.net'
 
 After doing this, your :file:`ipcontroller-client.json` file will look something like this:
 
 .. this can be Python, despite the fact that it's actually JSON, because it's
 .. still valid Python
 
 .. sourcecode:: python
 
     {
       "url":"tcp:\/\/*:43447",
       "exec_key":"9c7779e4-d08a-4c3b-ba8e-db1f80b562c1",
       "ssh":"login.mycluster.net",
       "location":"10.0.1.5"
     }
 
 Then this file will be all you need for a client to connect to the controller, tunneling
 SSH connections through login.mycluster.net.
 
 Database Backend
 ****************
 
 The Hub stores all messages and results passed between Clients and Engines.
 For large and/or long-running clusters, it would be unreasonable to keep all
 of this information in memory. For this reason, we have two database backends:
 [MongoDB]_ via PyMongo_, and SQLite with the stdlib :py:mod:`sqlite`.
 
 MongoDB is our design target, and the dict-like model it uses has driven our design. As far
 as we are concerned, BSON can be considered essentially the same as JSON, adding support
 for binary data and datetime objects, and any new database backend must support the same
 data types.
 
 .. seealso::
 
     MongoDB `BSON doc <http://www.mongodb.org/display/DOCS/BSON>`_
 
 To use one of these backends, you must set the :attr:`HubFactory.db_class` trait:
 
 .. sourcecode:: python
 
     # for a simple dict-based in-memory implementation, use dictdb
     # This is the default and the fastest, since it doesn't involve the filesystem
     c.HubFactory.db_class = 'IPython.parallel.controller.dictdb.DictDB'
     
     # To use MongoDB:
     c.HubFactory.db_class = 'IPython.parallel.controller.mongodb.MongoDB'
     
     # and SQLite:
     c.HubFactory.db_class = 'IPython.parallel.controller.sqlitedb.SQLiteDB'
 
 When using the proper databases, you can actually allow for tasks to persist from
 one session to the next by specifying the MongoDB database or SQLite table in
 which tasks are to be stored.  The default is to use a table named for the Hub's Session,
 which is a UUID, and thus different every time.
 
 .. sourcecode:: python
 
     # To keep persistant task history in MongoDB:
     c.MongoDB.database = 'tasks'
     
     # and in SQLite:
     c.SQLiteDB.table = 'tasks'
 
 
 Since MongoDB servers can be running remotely or configured to listen on a particular port,
 you can specify any arguments you may need to the PyMongo `Connection 
 <http://api.mongodb.org/python/1.9/api/pymongo/connection.html#pymongo.connection.Connection>`_:
 
 .. sourcecode:: python
 
     # positional args to pymongo.Connection
     c.MongoDB.connection_args = []
     
     # keyword args to pymongo.Connection
     c.MongoDB.connection_kwargs = {}
 
 .. _MongoDB: http://www.mongodb.org
 .. _PyMongo: http://api.mongodb.org/python/1.9/
 
 Configuring `ipengine`
 -----------------------
 
 The IPython Engine takes its configuration from the file :file:`ipengine_config.py`
 
 The Engine itself also has some amount of configuration. Most of this
 has to do with initializing MPI or connecting to the controller.
 
 To instruct the Engine to initialize with an MPI environment set up by
 mpi4py, add:
 
 .. sourcecode:: python
 
     c.MPI.use = 'mpi4py'
 
 In this case, the Engine will use our default mpi4py init script to set up
 the MPI environment prior to exection.  We have default init scripts for
 mpi4py and pytrilinos.  If you want to specify your own code to be run
 at the beginning, specify `c.MPI.init_script`.
 
 You can also specify a file or python command to be run at startup of the
 Engine:
 
 .. sourcecode:: python
 
     c.IPEngineApp.startup_script = u'/path/to/my/startup.py'
     
     c.IPEngineApp.startup_command = 'import numpy, scipy, mpi4py'
 
 These commands/files will be run again, after each
 
 It's also useful on systems with shared filesystems to run the engines
 in some scratch directory.  This can be set with:
 
 .. sourcecode:: python
 
     c.IPEngineApp.work_dir = u'/path/to/scratch/'
 
 
 
 .. [MongoDB] MongoDB database http://www.mongodb.org
 
 .. [PBS] Portable Batch System http://www.openpbs.org
 
 .. [SSH] SSH-Agent http://en.wikipedia.org/wiki/ssh-agent

docs/source/parallel/parallel_task.txt

@@ -1,442 +1,442 b''
 .. _parallel_task:
 
 ==========================
 The IPython task interface
 ==========================
 
 The task interface to the cluster presents the engines as a fault tolerant,
 dynamic load-balanced system of workers. Unlike the multiengine interface, in
 the task interface the user have no direct access to individual engines. By
 allowing the IPython scheduler to assign work, this interface is simultaneously
 simpler and more powerful.
 
 Best of all, the user can use both of these interfaces running at the same time
 to take advantage of their respective strengths. When the user can break up
 the user's work into segments that do not depend on previous execution, the
 task interface is ideal. But it also has more power and flexibility, allowing
 the user to guide the distribution of jobs, without having to assign tasks to
 engines explicitly.
 
 Starting the IPython controller and engines
 ===========================================
 
 To follow along with this tutorial, you will need to start the IPython
 controller and four IPython engines. The simplest way of doing this is to use
 the :command:`ipcluster` command::
 
-	$ ipcluster start n=4
+	$ ipcluster start --n=4
 	
 For more detailed information about starting the controller and engines, see
 our :ref:`introduction <ip1par>` to using IPython for parallel computing.
 
 Creating a ``Client`` instance
 ==============================
 
 The first step is to import the IPython :mod:`IPython.parallel`
 module and then create a :class:`.Client` instance, and we will also be using
 a :class:`LoadBalancedView`, here called `lview`:
 
 .. sourcecode:: ipython
 
     In [1]: from IPython.parallel import Client
 
     In [2]: rc = Client()
     
 
 This form assumes that the controller was started on localhost with default
 configuration. If not, the location of the controller must be given as an
 argument to the constructor:
 
 .. sourcecode:: ipython
 
     # for a visible LAN controller listening on an external port:
     In [2]: rc = Client('tcp://192.168.1.16:10101')
     # or to connect with a specific profile you have set up:
     In [3]: rc = Client(profile='mpi')
 
 For load-balanced execution, we will make use of a :class:`LoadBalancedView` object, which can
 be constructed via the client's :meth:`load_balanced_view` method:
 
 .. sourcecode:: ipython
 
     In [4]: lview = rc.load_balanced_view() # default load-balanced view
 
 .. seealso::
 
     For more information, see the in-depth explanation of :ref:`Views <parallel_details>`.
 
 
 Quick and easy parallelism
 ==========================
 
 In many cases, you simply want to apply a Python function to a sequence of
 objects, but *in parallel*. Like the multiengine interface, these can be
 implemented via the task interface. The exact same tools can perform these
 actions in load-balanced ways as well as multiplexed ways: a parallel version
 of :func:`map` and :func:`@parallel` function decorator. If one specifies the
 argument `balanced=True`, then they are dynamically load balanced. Thus, if the
 execution time per item varies significantly, you should use the versions in
 the task interface.
 
 Parallel map
 ------------
 
 To load-balance :meth:`map`,simply use a LoadBalancedView:
 
 .. sourcecode:: ipython
     
     In [62]: lview.block = True
     
 	In [63]: serial_result = map(lambda x:x**10, range(32))
 
 	In [64]: parallel_result = lview.map(lambda x:x**10, range(32))
 
 	In [65]: serial_result==parallel_result
 	Out[65]: True
 
 Parallel function decorator
 ---------------------------
 
 Parallel functions are just like normal function, but they can be called on
 sequences and *in parallel*. The multiengine interface provides a decorator
 that turns any Python function into a parallel function:
 
 .. sourcecode:: ipython
 
     In [10]: @lview.parallel()
        ....: def f(x):
        ....:     return 10.0*x**4
        ....:
 
     In [11]: f.map(range(32))    # this is done in parallel
     Out[11]: [0.0,10.0,160.0,...]
 
 .. _parallel_dependencies:
 
 Dependencies
 ============
 
 Often, pure atomic load-balancing is too primitive for your work. In these cases, you
 may want to associate some kind of `Dependency` that describes when, where, or whether
 a task can be run.  In IPython, we provide two types of dependencies: 
 `Functional Dependencies`_ and `Graph Dependencies`_
 
 .. note::
 
     It is important to note that the pure ZeroMQ scheduler does not support dependencies,
     and you will see errors or warnings if you try to use dependencies with the pure
     scheduler.
 
 Functional Dependencies
 -----------------------
 
 Functional dependencies are used to determine whether a given engine is capable of running
 a particular task.  This is implemented via a special :class:`Exception` class, 
 :class:`UnmetDependency`, found in `IPython.parallel.error`.  Its use is very simple: 
 if a task fails with an UnmetDependency exception, then the scheduler, instead of relaying
 the error up to the client like any other error, catches the error, and submits the task
 to a different engine.  This will repeat indefinitely, and a task will never be submitted
 to a given engine a second time.
 
 You can manually raise the :class:`UnmetDependency` yourself, but IPython has provided 
 some decorators for facilitating this behavior.
 
 There are two decorators and a class used for functional dependencies:
 
 .. sourcecode:: ipython
 
     In [9]: from IPython.parallel import depend, require, dependent
 
 @require
 ********
 
 The simplest sort of dependency is requiring that a Python module is available. The
 ``@require`` decorator lets you define a function that will only run on engines where names
 you specify are importable:
 
 .. sourcecode:: ipython
 
     In [10]: @require('numpy', 'zmq')
         ...: def myfunc():
         ...:     return dostuff()
 
 Now, any time you apply :func:`myfunc`, the task will only run on a machine that has
 numpy and pyzmq available, and when :func:`myfunc` is called, numpy and zmq will be imported.
 
 @depend
 *******
 
 The ``@depend`` decorator lets you decorate any function with any *other* function to
 evaluate the dependency. The dependency function will be called at the start of the task,
 and if it returns ``False``, then the dependency will be considered unmet, and the task
 will be assigned to another engine. If the dependency returns *anything other than
 ``False``*, the rest of the task will continue.
 
 .. sourcecode:: ipython
 
     In [10]: def platform_specific(plat):
         ...:    import sys
         ...:    return sys.platform == plat
     
     In [11]: @depend(platform_specific, 'darwin')
         ...: def mactask():
         ...:    do_mac_stuff()
 
     In [12]: @depend(platform_specific, 'nt')
         ...: def wintask():
         ...:    do_windows_stuff()
     
 In this case, any time you apply ``mytask``, it will only run on an OSX machine.
 ``@depend`` is just like ``apply``, in that it has a ``@depend(f,*args,**kwargs)``
 signature.
 
 dependents
 **********
 
 You don't have to use the decorators on your tasks, if for instance you may want
 to run tasks with a single function but varying dependencies, you can directly construct
 the :class:`dependent` object that the decorators use:
 
 .. sourcecode::ipython
 
     In [13]: def mytask(*args):
         ...:    dostuff()
     
     In [14]: mactask = dependent(mytask, platform_specific, 'darwin')
     # this is the same as decorating the declaration of mytask with @depend
     # but you can do it again:
     
     In [15]: wintask = dependent(mytask, platform_specific, 'nt')
     
     # in general:
     In [16]: t = dependent(f, g, *dargs, **dkwargs)
     
     # is equivalent to:
     In [17]: @depend(g, *dargs, **dkwargs)
         ...: def t(a,b,c):
         ...:     # contents of f
 
 Graph Dependencies
 ------------------
 
 Sometimes you want to restrict the time and/or location to run a given task as a function
 of the time and/or location of other tasks. This is implemented via a subclass of
 :class:`set`, called a :class:`Dependency`. A Dependency is just a set of `msg_ids`
 corresponding to tasks, and a few attributes to guide how to decide when the Dependency
 has been met.
 
 The switches we provide for interpreting whether a given dependency set has been met:
 
 any|all
     Whether the dependency is considered met if *any* of the dependencies are done, or
     only after *all* of them have finished.  This is set by a Dependency's :attr:`all`
     boolean attribute, which defaults to ``True``.
 
 success [default: True]
     Whether to consider tasks that succeeded as fulfilling dependencies.
 
 failure [default : False]
     Whether to consider tasks that failed as fulfilling dependencies.
     using `failure=True,success=False` is useful for setting up cleanup tasks, to be run
     only when tasks have failed.
 
 Sometimes you want to run a task after another, but only if that task succeeded. In this case,
 ``success`` should be ``True`` and ``failure`` should be ``False``. However sometimes you may
 not care whether the task succeeds, and always want the second task to run, in which case you
 should use `success=failure=True`. The default behavior is to only use successes.
 
 There are other switches for interpretation that are made at the *task* level.  These are
 specified via keyword arguments to the client's :meth:`apply` method.
 
 after,follow
     You may want to run a task *after* a given set of dependencies have been run and/or
     run it *where* another set of dependencies are met. To support this, every task has an
     `after` dependency to restrict time, and a `follow` dependency to restrict
     destination.
 
 timeout
     You may also want to set a time-limit for how long the scheduler should wait before a
     task's dependencies are met. This is done via a `timeout`, which defaults to 0, which
     indicates that the task should never timeout. If the timeout is reached, and the
     scheduler still hasn't been able to assign the task to an engine, the task will fail
     with a :class:`DependencyTimeout`.
 
 .. note::
 
     Dependencies only work within the task scheduler. You cannot instruct a load-balanced
     task to run after a job submitted via the MUX interface.
 
 The simplest form of Dependencies is with `all=True,success=True,failure=False`. In these cases,
 you can skip using Dependency objects, and just pass msg_ids or AsyncResult objects as the
 `follow` and `after` keywords to :meth:`client.apply`:
 
 .. sourcecode:: ipython
 
     In [14]: client.block=False
 
     In [15]: ar = lview.apply(f, args, kwargs)
 
     In [16]: ar2 = lview.apply(f2)
 
     In [17]: ar3 = lview.apply_with_flags(f3, after=[ar,ar2])
 
     In [17]: ar4 = lview.apply_with_flags(f3, follow=[ar], timeout=2.5)
 
 
 .. seealso::
 
     Some parallel workloads can be described as a `Directed Acyclic Graph
     <http://en.wikipedia.org/wiki/Directed_acyclic_graph>`_, or DAG. See :ref:`DAG
     Dependencies <dag_dependencies>` for an example demonstrating how to use map a NetworkX DAG
     onto task dependencies.
 
 
 
 
 Impossible Dependencies
 ***********************
 
 The schedulers do perform some analysis on graph dependencies to determine whether they
 are not possible to be met. If the scheduler does discover that a dependency cannot be
 met, then the task will fail with an :class:`ImpossibleDependency` error. This way, if the
 scheduler realized that a task can never be run, it won't sit indefinitely in the
 scheduler clogging the pipeline.
 
 The basic cases that are checked:
 
 * depending on nonexistent messages
 * `follow` dependencies were run on more than one machine and `all=True`
 * any dependencies failed and `all=True,success=True,failures=False`
 * all dependencies failed and `all=False,success=True,failure=False`
 
 .. warning::
 
     This analysis has not been proven to be rigorous, so it is likely possible for tasks
     to become impossible to run in obscure situations, so a timeout may be a good choice.
 
 
 Retries and Resubmit
 ====================
 
 Retries
 -------
 
 Another flag for tasks is `retries`.  This is an integer, specifying how many times
 a task should be resubmitted after failure.  This is useful for tasks that should still run
 if their engine was shutdown, or may have some statistical chance of failing.  The default
 is to not retry tasks.
 
 Resubmit
 --------
 
 Sometimes you may want to re-run a task. This could be because it failed for some reason, and
 you have fixed the error, or because you want to restore the cluster to an interrupted state.
 For this, the :class:`Client` has a :meth:`rc.resubmit` method.  This simply takes one or more
 msg_ids, and returns an :class:`AsyncHubResult` for the result(s).  You cannot resubmit
 a task that is pending - only those that have finished, either successful or unsuccessful.
 
 .. _parallel_schedulers:
 
 Schedulers
 ==========
 
 There are a variety of valid ways to determine where jobs should be assigned in a 
 load-balancing situation.  In IPython, we support several standard schemes, and
 even make it easy to define your own.  The scheme can be selected via the ``scheme``
 argument to :command:`ipcontroller`, or in the :attr:`TaskScheduler.schemename` attribute
 of a controller config object.
 
 The built-in routing schemes:
 
 To select one of these schemes, simply do::
 
-    $ ipcontroller scheme=<schemename>
+    $ ipcontroller --scheme=<schemename>
     for instance:
-    $ ipcontroller scheme=lru
+    $ ipcontroller --scheme=lru
 
 lru: Least Recently Used
 
     Always assign work to the least-recently-used engine.  A close relative of
     round-robin, it will be fair with respect to the number of tasks, agnostic 
     with respect to runtime of each task.
 
 plainrandom: Plain Random
 
     Randomly picks an engine on which to run.
 
 twobin: Two-Bin Random
 
     **Requires numpy**
     
     Pick two engines at random, and use the LRU of the two. This is known to be better
     than plain random in many cases, but requires a small amount of computation.
 
 leastload: Least Load
 
     **This is the default scheme**
 
     Always assign tasks to the engine with the fewest outstanding tasks (LRU breaks tie).
 
 weighted: Weighted Two-Bin Random
 
     **Requires numpy**
     
     Pick two engines at random using the number of outstanding tasks as inverse weights,
     and use the one with the lower load.
 
 
 Pure ZMQ Scheduler
 ------------------
 
 For maximum throughput, the 'pure' scheme is not Python at all, but a C-level
 :class:`MonitoredQueue` from PyZMQ, which uses a ZeroMQ ``XREQ`` socket to perform all
 load-balancing. This scheduler does not support any of the advanced features of the Python
 :class:`.Scheduler`.
 
 Disabled features when using the ZMQ Scheduler:
 
 * Engine unregistration
     Task farming will be disabled if an engine unregisters.  
     Further, if an engine is unregistered during computation, the scheduler may not recover.
 * Dependencies
     Since there is no Python logic inside the Scheduler, routing decisions cannot be made
     based on message content.
 * Early destination notification
     The Python schedulers know which engine gets which task, and notify the Hub.  This 
     allows graceful handling of Engines coming and going.  There is no way to know
     where ZeroMQ messages have gone, so there is no way to know what tasks are on which
     engine until they *finish*.  This makes recovery from engine shutdown very difficult.
 
 
 .. note::
 
     TODO: performance comparisons
 
 
 
 
 More details
 ============
 
 The :class:`LoadBalancedView` has many more powerful features that allow quite a bit
 of flexibility in how tasks are defined and run. The next places to look are
 in the following classes:
 
 * :class:`~IPython.parallel.client.view.LoadBalancedView`
 * :class:`~IPython.parallel.client.asyncresult.AsyncResult`
 * :meth:`~IPython.parallel.client.view.LoadBalancedView.apply`
 * :mod:`~IPython.parallel.controller.dependency`
 
 The following is an overview of how to use these classes together:
 
 1. Create a :class:`Client` and :class:`LoadBalancedView`
 2. Define some functions to be run as tasks
 3. Submit your tasks to using the :meth:`apply` method of your
    :class:`LoadBalancedView` instance.
 4. Use :meth:`Client.get_result` to get the results of the
    tasks, or use the :meth:`AsyncResult.get` method of the results to wait
    for and then receive the results.
 
 .. seealso::
 
     A demo of :ref:`DAG Dependencies <dag_dependencies>` with NetworkX and IPython.

docs/source/parallel/parallel_winhpc.txt

@@ -1,334 +1,334 b''
 ============================================
 Getting started with Windows HPC Server 2008
 ============================================
 
 .. note::
 
     Not adapted to zmq yet
 
 Introduction
 ============
 
 The Python programming language is an increasingly popular language for
 numerical computing. This is due to a unique combination of factors. First,
 Python is a high-level and *interactive* language that is well matched to
 interactive numerical work. Second, it is easy (often times trivial) to
 integrate legacy C/C++/Fortran code into Python. Third, a large number of
 high-quality open source projects provide all the needed building blocks for
 numerical computing: numerical arrays (NumPy), algorithms (SciPy), 2D/3D
 Visualization (Matplotlib, Mayavi, Chaco), Symbolic Mathematics (Sage, Sympy)
 and others.
 
 The IPython project is a core part of this open-source toolchain and is
 focused on creating a comprehensive environment for interactive and
 exploratory computing in the Python programming language. It enables all of
 the above tools to be used interactively and consists of two main components:
 
 * An enhanced interactive Python shell with support for interactive plotting
   and visualization.
 * An architecture for interactive parallel computing.
 
 With these components, it is possible to perform all aspects of a parallel
 computation interactively. This type of workflow is particularly relevant in
 scientific and numerical computing where algorithms, code and data are
 continually evolving as the user/developer explores a problem. The broad
 treads in computing (commodity clusters, multicore, cloud computing, etc.)
 make these capabilities of IPython particularly relevant.
 
 While IPython is a cross platform tool, it has particularly strong support for
 Windows based compute clusters running Windows HPC Server 2008. This document
 describes how to get started with IPython on Windows HPC Server 2008. The
 content and emphasis here is practical: installing IPython, configuring
 IPython to use the Windows job scheduler and running example parallel programs
 interactively. A more complete description of IPython's parallel computing
 capabilities can be found in IPython's online documentation
 (http://ipython.org/documentation.html).
 
 Setting up your Windows cluster
 ===============================
 
 This document assumes that you already have a cluster running Windows
 HPC Server 2008. Here is a broad overview of what is involved with setting up
 such a cluster:
 
 1. Install Windows Server 2008 on the head and compute nodes in the cluster.
 2. Setup the network configuration on each host. Each host should have a 
    static IP address.
 3. On the head node, activate the "Active Directory Domain Services" role 
    and make the head node the domain controller.
 4. Join the compute nodes to the newly created Active Directory (AD) domain.
 5. Setup user accounts in the domain with shared home directories.
 6. Install the HPC Pack 2008 on the head node to create a cluster.
 7. Install the HPC Pack 2008 on the compute nodes.
 
 More details about installing and configuring Windows HPC Server 2008 can be
 found on the Windows HPC Home Page (http://www.microsoft.com/hpc). Regardless
 of what steps you follow to set up your cluster, the remainder of this
 document will assume that:
 
 * There are domain users that can log on to the AD domain and submit jobs
   to the cluster scheduler.
 * These domain users have shared home directories. While shared home
   directories are not required to use IPython, they make it much easier to
   use IPython.
 
 Installation of IPython and its dependencies
 ============================================
 
 IPython and all of its dependencies are freely available and open source.
 These packages provide a powerful and cost-effective approach to numerical and
 scientific computing on Windows. The following dependencies are needed to run
 IPython on Windows:
 
 * Python 2.6 or 2.7 (http://www.python.org)
 * pywin32 (http://sourceforge.net/projects/pywin32/)
 * PyReadline (https://launchpad.net/pyreadline)
 * pyzmq (http://github.com/zeromq/pyzmq/downloads)
 * IPython (http://ipython.org)
 
 In addition, the following dependencies are needed to run the demos described
 in this document.
 
 * NumPy and SciPy (http://www.scipy.org)
 * Matplotlib (http://matplotlib.sourceforge.net/)
 
 The easiest way of obtaining these dependencies is through the Enthought
 Python Distribution (EPD) (http://www.enthought.com/products/epd.php). EPD is
 produced by Enthought, Inc. and contains all of these packages and others in a
 single installer and is available free for academic users. While it is also
 possible to download and install each package individually, this is a tedious
 process. Thus, we highly recommend using EPD to install these packages on
 Windows.
 
 Regardless of how you install the dependencies, here are the steps you will
 need to follow:
 
 1. Install all of the packages listed above, either individually or using EPD
    on the head node, compute nodes and user workstations.
 
 2. Make sure that :file:`C:\\Python27` and :file:`C:\\Python27\\Scripts` are
    in the system :envvar:`%PATH%` variable on each node.
 
 3. Install the latest development version of IPython. This can be done by
    downloading the the development version from the IPython website
    (http://ipython.org) and following the installation instructions.
 
 Further details about installing IPython or its dependencies can be found in
 the online IPython documentation (http://ipython.org/documentation.html)
 Once you are finished with the installation, you can try IPython out by
 opening a Windows Command Prompt and typing ``ipython``. This will
 start IPython's interactive shell and you should see something like the
 following screenshot:
 
 .. image:: ipython_shell.*
 
 Starting an IPython cluster
 ===========================
 
 To use IPython's parallel computing capabilities, you will need to start an
 IPython cluster. An IPython cluster consists of one controller and multiple
 engines:
 
 IPython controller
     The IPython controller manages the engines and acts as a gateway between 
     the engines and the client, which runs in the user's interactive IPython
     session. The controller is started using the :command:`ipcontroller`
     command.
 
 IPython engine
     IPython engines run a user's Python code in parallel on the compute nodes. 
     Engines are starting using the :command:`ipengine` command.
 
 Once these processes are started, a user can run Python code interactively and
 in parallel on the engines from within the IPython shell using an appropriate
 client. This includes the ability to interact with, plot and visualize data
 from the engines.
 
 IPython has a command line program called :command:`ipcluster` that automates
 all aspects of starting the controller and engines on the compute nodes.
 :command:`ipcluster` has full support for the Windows HPC job scheduler,
 meaning that :command:`ipcluster` can use this job scheduler to start the
 controller and engines. In our experience, the Windows HPC job scheduler is
 particularly well suited for interactive applications, such as IPython. Once
 :command:`ipcluster` is configured properly, a user can start an IPython
 cluster from their local workstation almost instantly, without having to log
 on to the head node (as is typically required by Unix based job schedulers).
 This enables a user to move seamlessly between serial and parallel
 computations.
 
 In this section we show how to use :command:`ipcluster` to start an IPython
 cluster using the Windows HPC Server 2008 job scheduler. To make sure that
 :command:`ipcluster` is installed and working properly, you should first try
 to start an IPython cluster on your local host. To do this, open a Windows
 Command Prompt and type the following command::
 
     ipcluster start n=2
 
 You should see a number of messages printed to the screen, ending with 
 "IPython cluster: started". The result should look something like the following
 screenshot:
 
 .. image:: ipcluster_start.*
 
 At this point, the controller and two engines are running on your local host.
 This configuration is useful for testing and for situations where you want to
 take advantage of multiple cores on your local computer. 
 
 Now that we have confirmed that :command:`ipcluster` is working properly, we
 describe how to configure and run an IPython cluster on an actual compute
 cluster running Windows HPC Server 2008. Here is an outline of the needed
 steps:
 
 1. Create a cluster profile using: ``ipython profile create --parallel profile=mycluster``
 
 2. Edit configuration files in the directory :file:`.ipython\\cluster_mycluster`
 
 3. Start the cluster using: ``ipcluser start profile=mycluster n=32``
 
 Creating a cluster profile
 --------------------------
 
 In most cases, you will have to create a cluster profile to use IPython on a
 cluster. A cluster profile is a name (like "mycluster") that is associated
 with a particular cluster configuration. The profile name is used by
 :command:`ipcluster` when working with the cluster.
 
 Associated with each cluster profile is a cluster directory. This cluster
 directory is a specially named directory (typically located in the
 :file:`.ipython` subdirectory of your home directory) that contains the
 configuration files for a particular cluster profile, as well as log files and
 security keys. The naming convention for cluster directories is:
 :file:`profile_<profile name>`. Thus, the cluster directory for a profile named
 "foo" would be :file:`.ipython\\cluster_foo`.
 
 To create a new cluster profile (named "mycluster") and the associated cluster
 directory, type the following command at the Windows Command Prompt::
 
-    ipython profile create --parallel profile=mycluster
+    ipython profile create --parallel --profile=mycluster
 
 The output of this command is shown in the screenshot below. Notice how
 :command:`ipcluster` prints out the location of the newly created cluster
 directory.
 
 .. image:: ipcluster_create.*
 
 Configuring a cluster profile
 -----------------------------
 
 Next, you will need to configure the newly created cluster profile by editing
 the following configuration files in the cluster directory:
 
 * :file:`ipcluster_config.py`
 * :file:`ipcontroller_config.py`
 * :file:`ipengine_config.py`
 
 When :command:`ipcluster` is run, these configuration files are used to 
 determine how the engines and controller will be started. In most cases,
 you will only have to set a few of the attributes in these files.
 
 To configure :command:`ipcluster` to use the Windows HPC job scheduler, you
 will need to edit the following attributes in the file
 :file:`ipcluster_config.py`::
 
     # Set these at the top of the file to tell ipcluster to use the
     # Windows HPC job scheduler.
     c.IPClusterStart.controller_launcher = \
         'IPython.parallel.apps.launcher.WindowsHPCControllerLauncher'
     c.IPClusterEngines.engine_launcher = \
         'IPython.parallel.apps.launcher.WindowsHPCEngineSetLauncher'
 
     # Set these to the host name of the scheduler (head node) of your cluster.
     c.WindowsHPCControllerLauncher.scheduler = 'HEADNODE'
     c.WindowsHPCEngineSetLauncher.scheduler = 'HEADNODE'
 
 There are a number of other configuration attributes that can be set, but
 in most cases these will be sufficient to get you started.
 
 .. warning::
     If any of your configuration attributes involve specifying the location
     of shared directories or files, you must make sure that you use UNC paths
     like :file:`\\\\host\\share`. It is also important that you specify
     these paths using raw Python strings: ``r'\\host\share'`` to make sure
     that the backslashes are properly escaped.
 
 Starting the cluster profile
 ----------------------------
 
 Once a cluster profile has been configured, starting an IPython cluster using
 the profile is simple::
 
-    ipcluster start profile=mycluster n=32
+    ipcluster start --profile=mycluster --n=32
 
 The ``-n`` option tells :command:`ipcluster` how many engines to start (in
 this case 32). Stopping the cluster is as simple as typing Control-C.
 
 Using the HPC Job Manager
 -------------------------
 
 When ``ipcluster start`` is run the first time, :command:`ipcluster` creates
 two XML job description files in the cluster directory:
 
 * :file:`ipcontroller_job.xml`
 * :file:`ipengineset_job.xml`
 
 Once these files have been created, they can be imported into the HPC Job
 Manager application. Then, the controller and engines for that profile can be
 started using the HPC Job Manager directly, without using :command:`ipcluster`.
 However, anytime the cluster profile is re-configured, ``ipcluster start``
 must be run again to regenerate the XML job description files. The
 following screenshot shows what the HPC Job Manager interface looks like
 with a running IPython cluster.
 
 .. image:: hpc_job_manager.*
 
 Performing a simple interactive parallel computation
 ====================================================
 
 Once you have started your IPython cluster, you can start to use it. To do
 this, open up a new Windows Command Prompt and start up IPython's interactive
 shell by typing::
 
     ipython
 
 Then you can create a :class:`MultiEngineClient` instance for your profile and
 use the resulting instance to do a simple interactive parallel computation. In
 the code and screenshot that follows, we take a simple Python function and
 apply it to each element of an array of integers in parallel using the
 :meth:`MultiEngineClient.map` method:
 
 .. sourcecode:: ipython
 
     In [1]: from IPython.parallel import *
 
     In [2]: c = MultiEngineClient(profile='mycluster')
 
     In [3]: mec.get_ids()
     Out[3]: [0, 1, 2, 3, 4, 5, 67, 8, 9, 10, 11, 12, 13, 14]
 
     In [4]: def f(x):
        ...:     return x**10
 
     In [5]: mec.map(f, range(15))  # f is applied in parallel
     Out[5]:
     [0,
      1,
      1024,
      59049,
      1048576,
      9765625,
      60466176,
      282475249,
      1073741824,
      3486784401L,
      10000000000L,
      25937424601L,
      61917364224L,
      137858491849L,
      289254654976L]
 
 The :meth:`map` method has the same signature as Python's builtin :func:`map`
 function, but runs the calculation in parallel. More involved examples of using
 :class:`MultiEngineClient` are provided in the examples that follow.
 
 .. image:: mec_simple.*