Financial Modelling in Python

P1: JYS

c02 JWBK378-Fletcher May 1, 2009 19:51 Printer: Yet to come

The PPF Package 7

Trying:

print black

scholes(S=42., K=40., r=0.1, sig= 0.2, T=0.5, CP=PUT)

Expecting:

0.808598915338

ok

2 items had no tests:

main

main . test

1 items passed all tests:

2 tests in

main .black scholes

2 tests in 3 items.

2 passed and 0 failed.

Test passed.

2.2.2 PyUnit

A full suite of unit tests for all modules in the ppf package is provided in the ppf.test

sub-package. The tests can be run module-by-module or, to execute all tests in one go, a driver

‘test

all.py’ is provided:

C:\Python25\Lib\site-packages\ppf\test>python test all.py --verbose

python test

all.py --verbose

test

call (test core.black scholes tests) ... ok

test

put (test core.black scholes tests) ... ok

test (test

core.libor rate tests) ... ok

.

.

.

test

upper bound (test utility.bound tests) ... ok

test

equal range (test utility.bound tests) ... ok

test

bound (test utility.bound tests) ... ok

test

bound ci (test utility.bound tests) ... ok

--------------------------------------------------------------------

Ran 51 tests in 25.375s

OK

2.3 BUILDING AND INSTALLING PPF

In this section we look at what it takes to build and install the ppf package.

2.3.1 Prerequisites and Dependencies

ppf is composed of a mixture of pure Python modules underlied by some supporting extension

modules implemented instandard C++. Accordingly, to build and install ppf requiresa modern

C++ compiler. The C++ extension modules have some library dependencies of their own,

notably the Boost C++ libraries and the Blitz++ C++ library. Instructions for downloading

P1: JYS

c02 JWBK378-Fletcher May 1, 2009 19:51 Printer: Yet to come

8 Financial Modelling in Python

and installing the Boost C++ libraries can be found at http://www.boost.org and

instructions for Blitz++ can be found at http://www.oonumerics.org. Naturally, an

installation of Python is also required. On Windows, the authors favour the freely available

ActiveState Python distribution, see http://www.activestate.com for download and

installation details. Also required on the Python side for ppf is an installation of the NumPy

package, see http://www.scipy.org for download and installation details.

2.3.2 Building the C++ Extension Modules

The ppf C++ extension modules are most conveniently built using the Boost.Build system

1

a

copy of which is included with the ppf sources. Also provided with the ppf sources for the

convenience of Windows users is a pre-built executable ‘bjam.exe’. Although these notes will

become a little Windows-centric at this point, the basic principles will hold for *NIX users also.

On Windows, the ppf package has been successfully built and tested with the Microsoft Visual

Studio C++ compiler versions 7.1, 8.0 (express edition), 9.0 (express edition), mingw/gcc-

3.4.5,

2

mingw/gcc-4.3.0 with Python versions 2.4 and 2.5, Boost versions 1.33.1, 1.34.0,

1.35, 1.36, 1.37 and Blitz++ version 0.9. The ppf package has also been built and tested on

the popular Linux-based operating system, Ubuntu-8.04.1 with Boost version 1.36.0, Blitz++

version 0.9 and gcc-4.2.3.

In the remainder of this section, without loss of generality, we will assume a Windows

operating system, Blitz++ version 0.9, the ActiveState distribution of Python version 2.5 and

Boost version 1.36.

Build Instructions

• Prerequisites

- Copy c:/path/to/ppf/ext/bjam.exe to somewhere in your %PATH%

- Install

o Blitz++-0.9

o Boost-1.36

o ActiveState Python 2.5

o NumPy for Python 2.5 (version 1.0.4 or 1.1.0)

- Edit as appropriate for your site

o c:/path/to/ppf/ext/build/user-config.jam

o c:/path/to/ppf/ext/build/site-config.jam

• Build

- c:/path/to/ppf>cd ext&&bjam [debug|release]

This will create:

o c:/path/to/ppf/ppf/math/ppf

math.pyd and

o c:/path/to/ppf/ppf/date

time/ppf date time.pyd

1

See http://www.boost.org/doc/tools/build/index.html.

2

Minimalist GNU for Windows – see http://www.mingw.org.

P1: JYS

c02 JWBK378-Fletcher May 1, 2009 19:51 Printer: Yet to come

The PPF Package 9

2.3.3 Installing the PPF Package

Assuming the steps of the previous section have been performed, installation of the ppf

package which relies on the standard Python Distutils package is very simple.

• Install

- c:/path/to/ppf>python setup.py install

which will copy the ppf package to the standard Python installation location

(c:/python25/lib/site-packages/ppf).

2.3.4 Testing a PPF Installation

The easiest way to verify a ppf installation is to run the ppf unit test suite.

• Test

- c:/python25/lib/site-packages/ppf/test>python test all.py --

verbose

P1: JYS

c02 JWBK378-Fletcher May 1, 2009 19:51 Printer: Yet to come

10