Cimrman Python Fem 04-03-2007

Introduction
Our choice
Example Problem
Final slide :-)
Python + FEM
Introduction to SFE
Robert Cimrman
Department of Mechanics & New Technologies Research Centre University of West Bohemia Plze n, Czech Republic
April 3, 2007, Plze n
1/22
Introduction
Our choice
Example Problem
Final slide :-)
Outline
1
Introduction Programming Languages Programming Techniques Our choice Mixing Languages Best of Both Worlds Python Software Dependencies Example Problem Continuous Formulation Weak Formulation Problem Description File Final slide :-) 2/22
Introduction
Our choice
Example Problem
Final slide :-)
Programming Languages
1
compiled (fortran, C, C++, Java, . . . ) Cons

(often) complicated build process, recompile after any change low-level lots of lines to get basic stu done code size maintenance problems static!
Pros
speed, speed, speed, . . . , did I say speed? large code base (legacy codes), tradition
interpreted or scripting (sh, tcl, matlab, perl, ruby, python, . . . ) Cons

many are relatively new lack of speed, or even utterly slow!
Pros
no compiling (very) high-level a few of lines to get (complex) stu done code size easy maintenance dynamic! (often) large code base
3/22
Introduction
Our choice
Example Problem
Final slide :-)
1

Pros

Pros
3/22
Introduction
Our choice
Example Problem
Final slide :-)
1

Pros

Pros
3/22
Introduction
Our choice
Example Problem
Final slide :-)
1

Pros

Pros
3/22
Introduction
Our choice
Example Problem
Final slide :-)
1

Pros

Pros
3/22
Introduction
Our choice
Example Problem
Final slide :-)
1

Pros

Pros
3/22
Introduction
Our choice
Example Problem
Final slide :-)
Example Python Code
4/22
Introduction
Our choice
Example Problem
Final slide :-)
Example C Code
c h a r s [ ] = s e l tudy , mel dudy . a s e l o p r a v d u t u d y . ; i n t s t r f i n d ( char s t r , char t g t ) { int tlen = s t r l e n ( tgt ); i n t max = s t r l e n ( s t r ) t l e n ; register int i ; f o r ( i = 0 ; i < max ; i ++) { i f ( s t r n c m p (& s t r [ i ] , t g t , t l e n ) == 0) return i ; } r e t u r n 1; } int i1 ; i 1 = s t r f i n d ( s , tudy ) ; p r i n t f ( %d \ n , i 1 ) ;
5/22
Introduction
Our choice
Example Problem
Final slide :-)
Programming Techniques
Random remarks: spaghetti code modular programming OOP design patterns (gang-of-four, GO4) post-OOP dynamic languages . . . all used, but I try to move down the list! :-)
6/22
Introduction
Our choice
Example Problem
Final slide :-)
Programming Techniques
Random remarks: spaghetti code modular programming OOP design patterns (gang-of-four, GO4) post-OOP dynamic languages . . . all used, but I try to move down the list! :-)
6/22
Introduction
Our choice
Example Problem
Final slide :-)
Mixing Languages Best of Both Worlds

Low level: C FE matrix evaluations costly mesh-related functions (listing faces, edges, surface generation, . . . ) ... can use also fortran
+
High level: Python www.python.org batteries included logic of the code conguration les! problem input les! 7/22
Introduction
Our choice
Example Problem
Final slide :-)
Mixing Languages Best of Both Worlds

Low level: C FE matrix evaluations costly mesh-related functions (listing faces, edges, surface generation, . . . ) ... can use also fortran
+
High level: Python www.python.org batteries included logic of the code conguration les! problem input les! 7/22
Introduction
Our choice
Example Problem
Final slide :-)
Python I
Python r is a dynamic object-oriented programming language that can be used for many kinds of software development. It oers strong support for integration with other languages and tools, comes with extensive standard libraries, and can be learned in a few days. Many Python programmers report substantial productivity gains and feel the language encourages the development of higher quality, more maintainable code.
...
batteries included
8/22
Introduction
Our choice
Example Problem
Final slide :-)
Python II
1.1.16 Why is it called Python? At the same time he began implementing Python, Guido van Rossum was also reading the published scripts from Monty Pythons Flying Circus (a BBC comedy series from the seventies, in the unlikely case you didnt know). It occurred to him that he needed a name that was short, unique, and slightly mysterious, so he decided to call the language Python. 1.1.17 Do I have to like Monty Pythons Flying Circus? No, but it helps. :) . . . General Python FAQ
9/22
Introduction
Our choice
Example Problem
Final slide :-)
Python III
NASA uses Python. . .
. . . so does Rackspace, Industrial Light and Magic, AstraZeneca, Honeywell, and many others.
http://wiki.python.org/moin/NumericAndScientic
10/22
Introduction
Our choice
Example Problem
Final slide :-)
Python III
NASA uses Python. . .
. . . so does Rackspace, Industrial Light and Magic, AstraZeneca, Honeywell, and many others.
http://wiki.python.org/moin/NumericAndScientic
10/22
Introduction
Our choice
Example Problem
Final slide :-)
Software Dependencies
1
libraries & modules SciPy: free (BSD license) collection of numerical computing libraries for Python
enables Matlab-like array/matrix manipulations and indexing
UMFPACK: very fast direct solver for sparse systems

Python wrappers are part of SciPy :-)
output les (results) and data les in HDF5:

general purpose library and le format for storing scientic data, ecient storage and I/O . . . via PyTables
2
misc: Pyparsing, Matplotlib tools SWIG: automatic generation of Python-C interface code Medit: simple mesh and data viewer ParaView: VTK-based advanced mesh and data viewer in Tcl/Tk MayaVi/MayaVi2: VTK-based advanced mesh and data viewer in Python
11/22
Introduction
Our choice
Example Problem
Final slide :-)
1


2
11/22
Introduction
Our choice
Example Problem
Final slide :-)
1


2
11/22
Introduction
Our choice
Example Problem
Final slide :-)
1


2
11/22
Introduction
Our choice
Example Problem
Final slide :-)
1


2
11/22
Introduction
Our choice
Example Problem
Final slide :-)
1


2
11/22
Introduction
Our choice
Example Problem
Final slide :-)
1


2
11/22
Introduction
Our choice
Example Problem
Final slide :-)
Shape Optimization in Incompressible Flow Problems

Objective Function (u ) 2 |u |2 min
c
(1)
minimize gradients of solution (e.g. losses) in c by moving design boundary perturbation of by vector eld V (t ) = + {t V (x )}x c \ where V = 0 in (2)
D in D D
C C out
12/22
Introduction
Our choice
Example Problem
Final slide :-)

c
(1)
D in D D
C C out
12/22
Introduction
Our choice
Example Problem
Final slide :-)

c
(1)
D in D D
C C out
12/22
Introduction
Our choice
Example Problem
Final slide :-)
Appetizer
ow and spline boxes, left: initial, right: nal
connectivity of spline boxes (6 boxes in dierent colours)
13/22
Introduction
Our choice
Example Problem
Final slide :-)
Continuous Formulation
laminar steady state incompressible Navier-Stokes equations 2 u + u u + p u boundary conditions u = u 0 on in u pn + = p n on out n u = 0 on walls adjoint equations for computing the adjoint solution w : 2 w + u w u w r = c 2 u in w = 0 in w = 0 on in walls and c is the characteristic function of c (6) = = 0 0 in in (3) (4)
(5)
14/22
Introduction
Our choice
Example Problem
Final slide :-)
Continuous Formulation
laminar steady state incompressible Navier-Stokes equations 2 u + u u + p u boundary conditions u = u 0 on in u pn + = p n on out n u = 0 on walls adjoint equations for computing the adjoint solution w : 2 w + u w u w r = c 2 u in w = 0 in w = 0 on in walls and c is the characteristic function of c (6) = = 0 0 in in (3) (4)
(5)
14/22
Introduction
Our choice
Example Problem
Final slide :-)
Weak Formulation
state problem: nd u, p (from suitable function spaces) such that v : u +

(u u ) v
p v =
out
p v n
v V0 ,
q u = 0 q Q
(7) adjoint equations: nd w, r (from suitable function spaces) such that v : w +

(v u ) w +
(u v ) w +
r v (8)
=
c
v : u
v V0 ,
q w = 0 q Q
+ proper BC . . .
15/22
Introduction
Our choice
Example Problem
Final slide :-)
Weak Formulation

(u u ) v
p v =
out
p v n
v V0 ,
q u = 0 q Q

(v u ) w +
(u v ) w +
r v (8)
=
c
v : u
v V0 ,
q w = 0 q Q
+ proper BC . . .
15/22
Introduction
Our choice
Example Problem
Final slide :-)
Weak Formulation

(u u ) v
p v =
out
p v n
v V0 ,
q u = 0 q Q

(v u ) w +
(u v ) w +
r v (8)
=
c
v : u
v V0 ,
q w = 0 q Q
+ proper BC . . .
15/22
Introduction
Our choice
Example Problem
Final slide :-)
Problem Description File I

a regular python le all python machinery is accessible! requires/recognizes several keywords to dene
mesh regions subdomains, boundaries, characteristic functions, . . . boundary conditions Dirichlet elds eld approximation per subdomain variables unknown elds, test elds, parameters equations materials constitutive relations, e.g. newton uid, elastic solid, . . . solver parameters nonlinear solver, ow solver, optimization solver, . . . ...
16/22
Introduction
Our choice
Example Problem
Final slide :-)
Problem Description File I

a regular python le all python machinery is accessible! requires/recognizes several keywords to dene
mesh regions subdomains, boundaries, characteristic functions, . . . boundary conditions Dirichlet elds eld approximation per subdomain variables unknown elds, test elds, parameters equations materials constitutive relations, e.g. newton uid, elastic solid, . . . solver parameters nonlinear solver, ow solver, optimization solver, . . . ...
16/22
Introduction
Our choice
Example Problem
Final slide :-)
Problem Description File II

mesh
f i l e N a m e m e s h = s i m p l e . mesh
regions
region 1000 = { name : Omega , s e l e c t : elements of group 6 , } region 0 = { name : W a l l s , s e l e c t : n o d e s o f s u r f a c e n r . O u t l e t , } region 1 = { name : I n l e t , s e l e c t : n o d e s by c i n c ( x , y , z , 0 ) , } region 2 = { name : O u t l e t , s e l e c t : n o d e s by c i n c ( x , y , z , 1 ) , } region 100 = { name : Omega C , # C o n t r o l domain . s e l e c t : n o d e s i n ( x > 25.9999 e 3) & ( x < 18.990 e 3) , } region 101 = { name : Omega D , # D e s i g n domain . s e l e c t : a l l e r . Omega C , }
17/22
Introduction
Our choice
Example Problem
Final slide :-)
Problem Description File III

elds
field 1 = { name : 3 v e l o c i t y , dim : ( 3 , 1 ) , f l a g s : ( BQP , ) , b a s e s : ( ( Omega , 3 4 P1B ) , ) } field 2 = { name : p r e s s u r e , dim : ( 1 , 1 ) , flags : ( ) , b a s e s : ( ( Omega , 3 4 P1 ) , ) }
variables
variables u w 0 w v p r q Nu } = : : : : : : : : { ( ( ( ( ( ( ( ( field field field field field field field field , , , , , , , , parameter , 3 v e l o c i t y , ( 3 , 4 , 5 ) ) , parameter , 3 v e l o c i t y , ( 3 , 4 , 5 ) ) , unknown , 3 v e l o c i t y , ( 3 , 4 , 5 ) , 0 ) , test , 3 v e l o c i t y , ( 3 , 4 , 5 ) , w ) , parameter , p r e s s u r e , ( 9 , ) ) , unknown , p r e s s u r e , ( 9 , ) , 1 ) , test , pressure , (9 ,) , r ) , meshVelocity , pressure , ( 9 , ) ) ,
18/22
Introduction
Our choice
Example Problem
Final slide :-)
Problem Description File III

elds
field 1 = { name : 3 v e l o c i t y , dim : ( 3 , 1 ) , f l a g s : ( BQP , ) , b a s e s : ( ( Omega , 3 4 P1B ) , ) } field 2 = { name : p r e s s u r e , dim : ( 1 , 1 ) , flags : ( ) , b a s e s : ( ( Omega , 3 4 P1 ) , ) }
variables
variables u w 0 w v p r q Nu } = : : : : : : : : { ( ( ( ( ( ( ( ( field field field field field field field field , , , , , , , , parameter , 3 v e l o c i t y , ( 3 , 4 , 5 ) ) , parameter , 3 v e l o c i t y , ( 3 , 4 , 5 ) ) , unknown , 3 v e l o c i t y , ( 3 , 4 , 5 ) , 0 ) , test , 3 v e l o c i t y , ( 3 , 4 , 5 ) , w ) , parameter , p r e s s u r e , ( 9 , ) ) , unknown , p r e s s u r e , ( 9 , ) , 1 ) , test , pressure , (9 ,) , r ) , meshVelocity , pressure , ( 9 , ) ) ,
18/22
Introduction
Our choice
Example Problem
Final slide :-)
Problem Description File IV

Dirichlet BC
ebc = { Walls : ( ( Walls , ( 3 , 4 , 5 ) , 0 . 0 ) , ) , Inlet : (( VelocityInlet x , (4 ,) , 1.0) , ( VelocityInlet yz , (3 ,5) , 0.0)) , }
state equations (compare with (7))

equations = { balance : + d w d i v g r a d . Omega ( f l u i d , v , w ) + d w c o n v e c t . Omega ( v , w ) d w g r a d . Omega ( v , r ) = d w s u r f a c e l t r . Outlet ( bpress , v ) , incompressibility : d w d i v . Omega ( q , w ) = 0 , }
19/22
Introduction
Our choice
Example Problem
Final slide :-)
Problem Description File IV

Dirichlet BC
ebc = { Walls : ( ( Walls , ( 3 , 4 , 5 ) , 0 . 0 ) , ) , Inlet : (( VelocityInlet x , (4 ,) , 1.0) , ( VelocityInlet yz , (3 ,5) , 0.0)) , }
state equations (compare with (7))

equations = { balance : + d w d i v g r a d . Omega ( f l u i d , v , w ) + d w c o n v e c t . Omega ( v , w ) d w g r a d . Omega ( v , r ) = d w s u r f a c e l t r . Outlet ( bpress , v ) , incompressibility : d w d i v . Omega ( q , w ) = 0 , }
19/22
Introduction
Our choice
Example Problem
Final slide :-)
Problem Description File V

adjoint equations (compare with (8))
equations adjoint = { balance : + d w d i v g r a d . Omega ( f l u i d , v , w ) + d w a d j c o n v e c t 1 . Omega ( v , w , u ) + d w a d j c o n v e c t 2 . Omega ( v , w , u ) + d w g r a d . Omega ( v , r ) = d w a d j d i v g r a d 2 . Omega C ( one , f l u i d , v , u ) , incompressibility : d w d i v . Omega ( q , w ) = 0 , }
materials
material 2 = { name : f l u i d , mode : h e r e , r e g i o n : Omega , v i s c o s i t y : 1 . 2 5 e 1, } material 100 = { name : b p r e s s , mode : h e r e , r e g i o n : Omega , val : 0.0 , }
20/22
Introduction
Our choice
Example Problem
Final slide :-)
Problem Description File V

adjoint equations (compare with (8))
equations adjoint = { balance : + d w d i v g r a d . Omega ( f l u i d , v , w ) + d w a d j c o n v e c t 1 . Omega ( v , w , u ) + d w a d j c o n v e c t 2 . Omega ( v , w , u ) + d w g r a d . Omega ( v , r ) = d w a d j d i v g r a d 2 . Omega C ( one , f l u i d , v , u ) , incompressibility : d w d i v . Omega ( q , w ) = 0 , }
materials
material 2 = { name : f l u i d , mode : h e r e , r e g i o n : Omega , v i s c o s i t y : 1 . 2 5 e 1, } material 100 = { name : b p r e s s , mode : h e r e , r e g i o n : Omega , val : 0.0 , }
20/22
Introduction
Our choice
Example Problem
Final slide :-)
Yes, the nal slide!

What is done basic FE element engine
approximations up to P2 on simplexes (possibly with bubble) Q1 tensor-product approximation on rectangles
What is not done general FE engine, with symbolic evaluation (` a la SyFi - merge it?) documentation, unit tests organization of solvers
fast problem-specic solvers
handling of elds and variables

boundary conditions, DOF handling
parallelization of both assembling and solving

use PETSC?
FE assembling equations, terms, regions materials, material caches generic solvers (Newton iteration, direct linear system solution)
What will not be done (?) GUI real symbolic parsing/evaluation of equations 21/22
Introduction
Our choice
Example Problem
Final slide :-)
Yes, the nal slide!



use PETSC?
Introduction
Our choice
Example Problem
Final slide :-)
Yes, the nal slide!



use PETSC?
Introduction
Our choice
Example Problem
Final slide :-)
This is not a slide!
22/22

Cimrman Python Fem 04-03-2007

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Cimrman Python Fem 04-03-2007

Загружено:

Авторское право:

Доступные форматы

Introduction

Final slide :-)

April 3, 2007, Plze n

Final slide :-)

Final slide :-)

compiled (fortran, C, C++, Java, . . . ) Cons

interpreted or scripting (sh, tcl, matlab, perl, ruby, python, . . . ) Cons

Final slide :-)

compiled (fortran, C, C++, Java, . . . ) Cons

interpreted or scripting (sh, tcl, matlab, perl, ruby, python, . . . ) Cons

Final slide :-)

compiled (fortran, C, C++, Java, . . . ) Cons

interpreted or scripting (sh, tcl, matlab, perl, ruby, python, . . . ) Cons

Final slide :-)

compiled (fortran, C, C++, Java, . . . ) Cons

interpreted or scripting (sh, tcl, matlab, perl, ruby, python, . . . ) Cons

Final slide :-)

compiled (fortran, C, C++, Java, . . . ) Cons

interpreted or scripting (sh, tcl, matlab, perl, ruby, python, . . . ) Cons

Final slide :-)

compiled (fortran, C, C++, Java, . . . ) Cons

interpreted or scripting (sh, tcl, matlab, perl, ruby, python, . . . ) Cons

Final slide :-)

Example Python Code

Final slide :-)

Final slide :-)

Final slide :-)

Final slide :-)

Mixing Languages Best of Both Worlds

Final slide :-)

Mixing Languages Best of Both Worlds

Final slide :-)

Final slide :-)

Final slide :-)

Final slide :-)

Final slide :-)

UMFPACK: very fast direct solver for sparse systems

output les (results) and data les in HDF5:

Final slide :-)

UMFPACK: very fast direct solver for sparse systems

output les (results) and data les in HDF5:

Final slide :-)

UMFPACK: very fast direct solver for sparse systems

output les (results) and data les in HDF5:

Final slide :-)

UMFPACK: very fast direct solver for sparse systems

output les (results) and data les in HDF5:

Final slide :-)

UMFPACK: very fast direct solver for sparse systems

output les (results) and data les in HDF5:

Final slide :-)

UMFPACK: very fast direct solver for sparse systems

output les (results) and data les in HDF5:

Final slide :-)

UMFPACK: very fast direct solver for sparse systems

output les (results) and data les in HDF5:

Final slide :-)

Shape Optimization in Incompressible Flow Problems

Final slide :-)

Shape Optimization in Incompressible Flow Problems

Final slide :-)

Shape Optimization in Incompressible Flow Problems

Final slide :-)

connectivity of spline boxes (6 boxes in dierent colours)