Introduction To Python Programming

The basics
Functions and modules

Numerical Python (NUMPY) : arrays
Visualising data (Matplotlib)
Numerical Methods using Python (SCIPY)
Concrete examples and exercises
Introduction to Python and some applications.
Marion Darbas
LAMFA UMR CNRS 7352, Universit de Picardie Jules Verne
CIMPA SCHOOL
NUOL University, Vientiane, Laos
January 03-13 2017
Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

The basics
Aims : Introduce the Python programming language

Basics
Useful Python librairies
Using Python for solving ordinary differential equations
(ODEs) and partial differential equations (PDEs)

The basics
Python
Open source general-purpose language

Object Oriented Procedural, Functional
Easy to interface with C/ObjC/Java/Fortran
Easy to interface with C++ (via SWIG)
Great interactive environment
Downloads : http ://www.python.org

Documentation : http ://www.python.org/doc/
Free book : http ://www.diveintopython.org

The basics
Packages for scientific computing
NUMPY module contains among other things :

a powerful N-dimensional array object
linear algebra tools
sophisticated (broadcasting) functions
tools for integrating C/C++ and Fortran code

The basics
Packages for scientific computing
SCIPY provides a multitude of numerical algorithms

Numerical integration
Interpolation
Solving ordinary differential equations, . . .
Matplotlib is a high quality plotting library.

The basics
Outline of the course
1 The basics
2 Functions and modules
3 Numerical Python (NUMPY) : arrays
4 Visualising data (Matplotlib)
5 Numerical Methods using Python (SCIPY)
6 Concrete examples and exercises

The basics
The basics
Python is an interpreted language.
Enter individual commands at the Python prompt (>>>) which are

immediately evaluated and carried out by the Python interpreter.
Collect sequences of commands into a text file and save this file as
a Python program. Extension ".py"

The basics
The basics
Python is a calculator : basic operations such as addition (+),

subtraction (-), multiplication (*), division (/) and exponentiation (**).
>>> 10 + 10000
10010
>>> 42 1.5
40.5
>>> 15./6.
2.5
>>> 23
8

The basics
The basics
Mathematical functions : mathematics module with name math.

(sin, cos, tan, exp, log, . . .)
>>> import math

>>> math.exp(1.0)
2.7182818284590451
help(math), help(math.exp) provide more information.
>>> math.pi
3.141592653589793
>>> math.e
2.7182818284590451

The basics
The basics
Variables : A variable is created through assignment.
>>> x = 0.5
>>> x2
0.25
>>> x = y = z = 0 # initialise x, y and z with 0
Notation +=
>>> x = 4
>>> x+ = 1
>>> print(x)
5
Other operators : *=, -=, /=.

The basics
The basics
The print() command

The print command is the most commonly used command to print
information.
>>> a, b = 10, 20
>>> print(a)
10
>>> print(a, b)
10 20
>>> print(The answer is, a)
The answer is 10

The basics
The basics
Different data types
>>> a = 45
>>> type(a)
int
>>> b =0 This is a string0
>>> type(b)
str
>>> c = 2 + 3j
>>> type(c)
complex
>>> d = [4, 5, 6]
>>> type(d)
list

The basics
The basics
Complex numbers
>>> x = 1 + 3j
>>> abs(x) # modulus
3.1622776601683795
>>> x.imag # imaginary part
3.0
>>> x.real # real part
1.0
>>> import cmath

>>> cmath.sqrt(x)
(1.442615274452683 + 1.0397782600555705j)
cmath means Complex MATHematics.

The basics
The basics
Sequence type : list
A list is a sequence of objects. The objects can be of any type. For
example, integers or strings :
>>> a = [34, 12, 54]

>>> a = [dog, cat, mouse]
It is also possible to mix different types in the same list :
>>> a = [1022, elephant, 0, 4, duck]
An empty list is presented by
>>> a = [ ] #empty list

The basics
The basics
Indexing sequences
Individual objects in lists can be accessed by using the index of the object
and square brackets. Note that Python starts counting indices from zero !
>>> a = [12, 10, 18, 7, 15, 3] # Create a list

>>> print(a)
[12, 10, 18, 7, 15, 3]
>>> a[0]
12
>>> a[2]
18
>>> a[1] = 11
>>> print(a)
[12, 11, 18, 7, 15, 3]

The basics
The basics
Some operations

The basics
The basics
Some operations
You can add one object to the end of a list using the append() method
>>> a = [34, 56, 23]

>>> a.append(42)
>>> print(a)
[34, 56, 23, 42]
You can delete an object from a list by calling the remove() method
>>> a = [34, 56, 23, 42]
>>> a.remove(56)
>>> print(a)
[34, 23, 42]
c=a[:] will return a copy of a.

The basics
The basics
Slicing
A subsequence of a sequence is called a slice.
fraise[i:j] returns elements i up to j 1.

>>> fraise[2 : 4] #index of end not included
[18, 7]
The basics
The basics
Indexing versus slicing
If you are not sure what the right index is, it is always a good technique
to play around with a example of small size to test before or while you
write your program.

The basics
The basics
The range() command

A special type of list is frequently required (often together with for-loops).
A command exists to generate that list : the range(n) command
generates a list of integers starting from 0 and going up to n-1.
>>> range(10)
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
>>> range(3, 10) # start = 3
[3 , 4 , 5 , 6 , 7 , 8 , 9 ]
>>> range(3, 10, 2) # start = 3 , step = 2
[3, 5, 7, 9]
>>> range(10, 0, 1) # start = 10 , step = 1
[10, 9, 8, 7, 6, 5, 4, 3, 2, 1]

The basics
The basics
If-then-else
The if statement allows conditional execution of code.
The if-statement can also have an else branch which is executed if the
condition is wrong.
Python requires the colon symbol : after each condition.

The basics
The basics
There is the elif keyword that allows checking for several (exclusive)
possibilities.
Comparison operators

The basics
The basics
For loop
The for-loop allows to iterate over a sequence.
This could be a string or a list, for example.
The output is [1.0, 0.5, 0.33333333333333331, 0.25].

The basics
The basics
While loop
The while keyword allows to repeat an operation while a condition is true.
To create the list [1, 1/2, 1/3, 1/4], we write
Remark 1 : It is important not to forget the instruction n+ = 1.

Remark 2 : The command break allows to quit a loop.
The basics

Functions allow us to group a number of statements into a logical block.
We communicate with a function through a clearly defined interface,
providing certain parameters to the function and receiving some
information back.
Ex. : we define a function that computes the square of a given variable.

The basics
The listing of a file that shows how the function square.py can be used :
This program will produce the following output

The basics
In the following example, the function derivatives computes an

approximation of the first and second derivatives of a function f at point
x using the formula
f (x + h) f (x h) f (x + h) 2f (x) + f (x h)
f 0 (x) ' , f (x) '
2h h2

The basics
To approximate the values of the first and second derivatives of the

function x 7 cos(x) at x = /2, we write
df, ddf = derivatives(math.cos, math.pi/2, 1.0e 5)

print(First derivative, df)
print(Second derivative, ddf)
Another example : the input is a list

The basics

A module is a collection of functions. Pythons standard library contains
a vast collection of modules.
Importing modules
We introduce the name math into the namespace in which the import
command was issued. More than one module can be imported in the
same statement.
We import the sin function from the math module, but the name math is
not introduced into the current namespace. It only introduces the name
sin.

The basics
The NumPy package (read as NUMerical PYthon) provides access to a

new data structure called arrays which allows efficient vector and matrix
operations. It also provides a number of linear algebra operations (such
as solving of systems of linear equations, computation of eigenvectors
and eigenvalues).
An array appears to be very similar to a list but an array can keep only
elements of the same type (whereas a list can mix different kinds of
objects).

The basics

Vectors (1d-arrays)
Conversion of a list into an array using numpy.array :
Creation of vector with zeros
Creation of a vector using ArrayRange :

The basics

Once the array is established, we can set and retrieve individual values.
Calculations are performed on every element in the vector with a single

statement :

The basics

Matrices (2d-arrays)
There are two ways to create a 2d-array :
By converting a list of lists into an array :
Using the "zeros method" (ex. : matrix with 5 rows and 4 columns)

The basics

Individual elements can be accessed and set using this syntax
Create the unity matrix A M3 (C)
>>> A = N.eye(3) # numpy .eye

The basics

Standard linear algebra operations
Matrix-vector product
Resolution of the linear system Ax = b
Computation of the eigenvalues and eigenvectors of a matrix A

>>> evalues, evectors = LA.eig(A)

The basics

Standard linear algebra operations
Numpy provides the routine polyfit(x,y,n) which takes two lists x
and y of same length and a desired order n of the polynomial that
will be determined to fit the data in the least-square sense.
Numpy creates a polynomial function p from coefficients :

p = numpy.poly1d(z)
The basics
The Python library Matplotlib is a python 2D plotting library which

produces publication quality figures.
To plot the graph of a function f : [a, b] 7 R, Python needs a finite
number of points (xi , f (xi )). To generate the points (xi ), two commands
exist (NUMPY module)
ba
linspace(a,b,n) generates a list of n points xi = a + i ,
n1
i = 0, . . . , n 1 (x0 = a, xn1 = b).
b a
arange(a,b,h) generates a list of n = E + 1 elements
h
xi = a + ih, i = 0, . . . , n 1.

The basics
An example

The basics

Some commands
figure(figsize=(5,5)) sets the figure size to 5 by 5.

plot(x,y1,label=sin(x)) The label keyword denotes the
name of this line.
plot(x,y,b-,linewidth=2) will plot a blue solid line with two
pixel thickness 2.
axis([-2,2,-1,1]). This fixes the displayed area to go from
xmin= 2 to xmax= 2 in x-direction, and from ymin= 1 to
ymax=1 in y-direction.
grid() This command will display a grid on the backdrop.
xlabel(...), ylabel(...) allow labelling the axes.

The basics

Some options of Pylab
The full list of options can be found when typing help(pylab.plot)

The basics

2D
Consider the function f (x, y ) = (1 x)2 + (y x 2 )2 .
Mesh the rectangle [1, 1] [1, 2]
Display the contour lines of the function f

The basics

SCIPY package (SCIentific PYthon) provides a multitude of numerical
algorithms. SCIPY is built on NUMPY.
Solving ordinary differential equations
To solve an ODE of the type
y 0 (t) = f (t, y (t))
with a given y (t0 ) = y0 , we use SCIPYs odeint function.

Example : find y (t) for t [0, 2] solution to
y 0 (t) = 2ty (t)
with y (0) = 1.
2
Exact solution : y (t) = e t .

The basics

The basics

The basics
Root finding : Find x such that f (x) = 0.

Dichotomy method or Bisect algorithm
Assume f (a)f (b) 0.
Initialization : n = 0, x0 = a, x1 = b, x2 = (x0 + x1 )/2
While (|f (x2 )| ) and (n itermax)
If f (x0 )f (x2 ) 0 set x1 = x2 , else set x0 = x2
x2 = (x0 + x1 )/2
n=n+1

The basics

Suppose we need to compute the roots of f (x) = x 2 3. Here is a
program that determines this root numerically
This produces the following output :

The basics

A more efficient algorithm than the bisection one is implemented in the
general purpose fsolve() function for root finding of (multidimensional)
functions. This algorithm needs only one starting point close to the
suspected location of the root. Ex : f (x) = x 3 2x 2 .
This produces the following output :

The basics
Exercise 1
1 Open a new file with name testodeint.py file.
2 Write a program that computes the solution y (t) of this ODE using
the odeint algorithm
y 0 (t) = exp(t)(10 sin(10t) + cos(10t))
from t = 0 to t = 10. The initial value is y (0) = 1.

3 Display the solution graphically at t = 0, t = 0.01, . . . , t = 10.

The basics

The basics
Solving ordinary differential equations
Cauchy problem
Consider the system of ODEs of first order :
find u : t [t 0 , T ] 7 u(t) Rd
0
u (t) = f (t, u(t)), t 0 t T ,
(1)
u(t 0 ) = u 0 ,
with t 0 , T > 0, f : [t 0 , T ] Rd Rd a regular function and u 0 Rd .

The basics
Explicit numerical schemes

Discretization of the interval I = [t 0 , T ]
T t0
1 Fix a discretization step h = , with N N .
N
2 Construct the discrete times ti = t 0 + ih, 0 i N (t0 = t 0 ,
tN = T ). We have I = 0iN [ti , ti+1 ].
Approximation
Compute the approximated values (ui )0iN , where ui Rd approaches
the value u(ti ) with u the exact solution of (1).

The basics
Forward Euler explicit scheme (order 1)
u0 = u(t 0 ) = u 0 Rd

ui+1 = ui + hf (ti , ui ), i = 0, 1, . . . , N 1.
Modified Euler explicit scheme (order 2)

0 0 d
u0 = u(t ) = u R

K1 = f (ti , ui )

h h
K2 = f (ti + , ui + K1 )

2 2
ui+1 = ui + hK2 , i = 0, 1, . . . , N 1.

The basics
Runge-Kutta method of order 4
u0 = u(t 0 ) = u 0 Rd

K1 = f (ti , ui )
h h

K2 = f (ti + 2 , ui + 2 K1 )

h h
K3 = f (ti + , ui + K2 )
K = f (t n + 2h, u n + 2hK )

4 3
h

i+1 = ui + (K1 + 2K2 + 2K3 + K4 ), i = 0, 1, . . . , N 1.
u

6

The basics
Exercise 2
1 Open a new file with name ODE1D.py file.
2 Write a program that computes the solution y (t) of this ODE using
the previous explicit numerical schemes
y 0 (t) = 2ty (t)
from t = 0 to t = 1. The initial value is y (0) = 1.

3 Display the exact solution and its approximation graphically.
4 Compute numerically the convergence order of each method and
compare with the respective theoretical orders.

The basics

Lotka-Volterra equations : predator-prey model
A pair of first-order, non-linear differential equations describes the
dynamics of biological systems in which two species interact, one as
predator and the other as prey :
0
x (t) = x(t)(a by (t))
y 0 (t) = y (t)(c + dx(t)) (2)
x(0) = x0 , y (0) = y0

with a, b, c, d , x0 , y0 non-negative constants.
x(t) is the number of preys at time t

y (t) is the number of predators at time t
There are two equilibrium points : (0, 0) (saddle point) and (d /c, a/b)
(center). Oscillatory solutions.
The basics
Exercise 3
1 Write a function which implements the forward Euler explicit
method for the resolution of a system of two ODEs of order 1.
2 Apply it for solving the Lotka-Volterra system fixing T = 10, a = 3,
b = 1, c = 2 and d = 1 and the different initial conditions
(x0 , y0 ) = (5, 1), (x0 , y0 ) = (3, 3) and (x0 , y0 ) = (1, 2).
What is the conclusion ?
3 Same questions with the RK4 scheme. What is the conclusion ?

The basics

1D finite difference method - Stationary problem
Consider the problem
u 00 (x) = cos x, /2 < x < 3/2,

(3)
u(/2) = 0, u(3/2) = 0.
Discretization of [/2, 3/2] : Constant mesh step h = /N, with

N N . We introduce the nodes (xi )i=0, ,N , xi = /2 + ih
(x0 = /2, xN = 3/2)
Finite difference scheme : We denote by ui the approximation of the
value u(xi ) where u is the exact solution of (5). The discrete
problem reads : Find uh = (ui )i=0, ,N RN+1 such that
ui1 2ui + ui+1
(
= cos ui , 1 i N 1,
h2 (4)
u0 = uN = 0.

The basics

Exercise 4
1 Let uh = (u1 , , uN1 ) RN1 . Write the discrete problem under
matricial form Ah uh = bh with Ah R(N1)(N1) , bh RN1 .
2 Write a Python function which computes the solution of this linear
system.
3 Display on the same figure the exact solution u(x) = cos x and its
approximation at a fixed N.
4 Compute for different values of N (for instance N = 5, . . . , 100) the
error ku uh k . Report it in function of the parameter N in log-log
scale.
5 Propose a method for estimating numerically the order of
convergence of the finite difference scheme. Recall that the
theoretical order equals 2.
The basics

1D finite difference method - Non-stationary problem
Consider the 1D heat equation subject to homogeneous Dirichlet
boundary conditions
t u x2 u = 0, 0 < t < T , 0 < x < 1,

u(0, x) = u0 (x), 0 < x < 1, (5)

u(t, 0) = u(t, 1) = 0, 0 < t < T ,

with u0 a given initial condition.

Time discretization : Discrete times tn = nt , n = 0, . . . , N where
t = T /N, N N .
Space discretization : Nodes xj = jx , j = 0, . . . , J + 1, where
x = 1/(J + 1), J N.
Approximation : We denote by ujn the approximation of the exact
value u(tn , xj ), and by u n = (ujn )1jJ the approximation at time t n .

The basics

Explicit scheme
n+1
u ujn n
uj+1 2ujn + uj1
n
j , 1 n N, 1 j J,

t 2x (6)
u 0 = u0 (xj ), 1 j J,
jn

n
u0 = uJ+1 = 0, 0 n N,
The scheme writes equivalently under the form
t
ujn+1 = ujn + (uj+1
n
2ujn + uj1
n
) = 0, := .
2x
At each step n, u n is known and we compute u n+1 such that
u n+1 = Au n
with A MJ (R) a tridiagonal matrix.
The basics
The -scheme, [0, 1]
n+1 n+1
uj ujn uj+1 2ujn+1 + uj1
n+1 n
uj+1 2ujn + uj1
n
= 0,

(1 )

t 2x 2x
1 n N, 1 j J,
0
u = u (x ), 1 j J,

0 j
jn

n
u0 = uJ+1 = 0, 0 n N.

The basics

Exercise 5
Consider the initial condition
1 1
u0 (x) = 2x, 0 x , and u0 (x) = 2 2x, x 1.
2 2
The exact solution is

X 2 8 n
u(t, x) = cn e (n) t sin(nx), cn = sin( ).
n=1
n2 2 2
1 Implement the explicit scheme.

2 Discuss the convergence of the numerical method in function of the
parameter .
3 Same questions with the Crank-Nicholson scheme ( = 1/2).

Introduction To Python Programming

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Introduction To Python Programming

Hochgeladen von

Copyright:

Verfügbare Formate

The basics

Functions and modules

Introduction to Python and some applications.

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

Aims : Introduce the Python programming language

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

Open source general-purpose language

Downloads : http ://www.python.org

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

Packages for scientific computing

NUMPY module contains among other things :

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

Packages for scientific computing

SCIPY provides a multitude of numerical algorithms

Matplotlib is a high quality plotting library.

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

Outline of the course

2 Functions and modules

3 Numerical Python (NUMPY) : arrays

4 Visualising data (Matplotlib)

5 Numerical Methods using Python (SCIPY)

6 Concrete examples and exercises

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

Python is an interpreted language.

Enter individual commands at the Python prompt (>>>) which are

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

Python is a calculator : basic operations such as addition (+),

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

Mathematical functions : mathematics module with name math.

>>> import math

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

Variables : A variable is created through assignment.

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

The print() command

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

Different data types

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

>>> import cmath

cmath means Complex MATHematics.

>>> a = [34, 12, 54]

It is also possible to mix different types in the same list :

>>> a = [1022, elephant, 0, 4, duck]

An empty list is presented by

>>> a = [ ] #empty list

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

>>> a = [12, 10, 18, 7, 15, 3] # Create a list

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

>>> a = [34, 56, 23]

c=a[:] will return a copy of a.

fraise[i:j] returns elements i up to j 1.

Indexing versus slicing

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

The range() command

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

Python requires the colon symbol : after each condition.

Marion Darbas (LAMFA) CIMPA SCHOOL - Vientiane - January 03-13 2017

This could be a string or a list, for example.

The output is [1.0, 0.5, 0.33333333333333331, 0.25].

To create the list [1, 1/2, 1/3, 1/4], we write

Remark 1 : It is important not to forget the instruction n+ = 1.

Functions and modules

Ex. : we define a function that computes the square of a given variable.