1 

2 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 
3 
% 
4 
% Copyright (c) 20032009 by University of Queensland 
5 
% Earth Systems Science Computational Center (ESSCC) 
6 
% http://www.uq.edu.au/esscc 
7 
% 
8 
% Primary Business: Queensland, Australia 
9 
% Licensed under the Open Software License version 3.0 
10 
% http://www.opensource.org/licenses/osl3.0.php 
11 
% 
12 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 
13 

14 
\section{One Dimensional Heat Diffusion in an Iron Rod} 
15 
%\label{Sec:1DHDv0} 
16 
We will start by examining a simple one dimensional heat diffusion example. While this exact problem is not strictly relevant to earth sciences; it will provide a good launch pad to build our knowledge of \ESCRIPT and how to solve simple partial differential equations (PDEs) \footnote{In case you wonder what a 
17 
\textit{partial differential equation} wikipedia provides a 
18 
comprehensive introducton at 
19 
\url{http://en.wikipedia.org/wiki/Partial_differential_equation}, 
20 
but things should become a bit clearer when you read further in the cookbook.} 
21 

22 

23 

24 
Start by imagining we have a simple cold iron bar at a constant temperature of zero. 
25 
\TODO{Add a diagram to explain the setup.} 
26 
The bar is perfectly insulated on all sides and at one end we will apply a heating element of some description. Intuition tells us that as heat is applied, that energy will disperse through the bar with time until the bar reaches the same temperature as the heat source. At this point the temperature in the bar will be constant and the same as the heat source. 
27 

28 
We can model this problem using the one dimensional heat diffusion equation. A detailed discussion on how the heat diffusion equation is derived can be found at 
29 
\url{http://online.redwoods.edu/instruct/darnold/DEProj/sp02/AbeRichards/paper.pdf} 
30 
Heat diffusion equation is defined as: 
31 
\begin{equation} 
32 
\rho c\hackscore p \frac{\partial T}{\partial t}  \kappa \frac{\partial^{2} T}{\partial x^{2}} = q\hackscore H 
33 
\label{eqn:hd} 
34 
\end{equation} 
35 
where $\rho$ is the material density, $c\hackscore p$ is the specific heat and $\kappa$ is the temperature diffusion constant. All of these values are readily available for most materials or can be established through predefined experimentation techniques. \TODO{Give some numbers for eg. granite}. 
36 
The heatsource is on the right hand side of \eqref{eqn:hd} as $q_{H}$, this could be a constant or defined by an expression \TODO{Give an example}. There are also two partial derivatives in \eqref{eqn:hd}, $\frac{\partial T}{\partial t}$ describes the change in temperature with time while $\frac{\partial T}{\partial x}$ describes the spatial change to temperature. There is only a single spatial dimension to our problem, and so our temperature solution $T$ is only dependent on the time $t$ and our position along the iron bar $x$. 
37 

38 
\TODO{Explain the concept of time discritzation} 
39 

40 
To solve this equation we will write a simple python script which uses \ESCRIPT and 
41 
\FINLEY of the \ESYS module. At this point we assume that you have some basic 
42 
understanding of the python programming language. There are in fact a large number of 
43 
python tutorial available online, for instance 
44 
\begin{itemize} 
45 
\item This is a very crisp introduction \url{http://hetland.org/writing/instantpython.html}. It covers everthing you need to get started with \ESCRIPT. 
46 
\item A nice and easy to follow introduction: \url{http://www.sthurlow.com/python/} 
47 
\item An other crip tutorial \url{http://www.zetcode.com/tutorials/pythontutorial/}. 
48 
\item A very comprehensive tutorial from the python authors: \url{http://www.python.org/doc/2.5.2/tut/tut.html}. It covers much more than what you will ever need for \ESCRIPT. 
49 
\item an other comprehensive tutorial: \url{http://www.tutorialspoint.com/python/index.htm} 
50 
\end{itemize} 
51 
In the following we will develop the script to solve the heat equation stepbystep. 
52 
The first step is to import the necessary libraries. 
53 
\begin{verbatim} 
54 
from esys.escript import * 
55 
from esys.escript.linearPDEs import SingleLinearPDE 
56 
from esys.finley import Rectangle 
57 
import os 
58 
\end{verbatim} 
59 
It is generally a good idea to import all of \verb escript , although if you know the packages you need you can specify them individually. The function \verbLinearPDE has been imported individually for ease of use later in the script. \verbRectangle is going to be our type of domain and the \verbos package is needed to handle file outputs once our PDE has been solved. 
60 

61 
Once our libraries dependancies have been established, defining the problem specific variables is the next step. In general the number of variables needed will vary between problems. These variables belong to two categories. They are either directly related to the PDE and can be used as inputs into the escript solver, or they are script variables used to control internal functions and iterations in our problem. For this PDE there are a number of constants which will need values. Firstly, the domain upon which we wish to solve our problem needs to be defined. There are many different types of domains in escript. We will demonstrate a few in later tutorials but for our iron rod we will simply use a rectangular domain. 
62 

63 
Using a rectangular domain simplifies a \textit{3D} object into a single dimension. The iron rod will have a lengthways cross section that looks like a rectangle. As a result we do not need to model the volume of the rod because a cylinder is symmetrical about its center. There are four arguments we must consider when we decide to create a rectangular domain, the model length, width and step size in each direction. When defining the size of our problem it will help us determine appropriate values for our domain arguments. If we make our dimensions large but our step sizes very small we will to a point, increase the accuracy of our solution. Unfortunately we also increase the number of calculations that must be solved per time step. This means more computational time is required to produce a solution. In our \textit{1D} problem we will define our bar as being 1 metre long. An appropriate \verbndx would be 1 to 10\% of the length. Our \verbndy need only be 1, uhis is because our problem stipulates no partial derivatives in the $y$ direction. This means the temperature does not vary with $y$. Thus the domain perameters can be defined as follows: 
64 
\TODO{Use the Unit module!} 
65 
\begin{verbatim} 
66 
#Domain related. 
67 
mx = 1 #meters  model lenght 
68 
my = .1 #meters  model width 
69 
ndx = 100 # steps in x direction 
70 
ndy = 1 # steps in y direction 
71 
\end{verbatim} 
72 
The material constants and the temperature variables must also be defined. For the iron rod in the model they are defined as: 
73 
\begin{verbatim} 
74 
#PDE related 
75 
q=473. #Kelvin  our heat source temperature 
76 
Tref = 273. # Kelvin  starting temp of iron bar 
77 
rho = 7874. #kg/m^{3} density of iron 
78 
cp = 449. #j/Kg.K 
79 
rhocp = rho*cp 
80 
eta = 0 #radiation condition 
81 
kappa = 68. #temperature diffusion constant 
82 
\end{verbatim} 
83 
\TODO{remove radiation condition and if required introduce in a second example} 
84 
Finally, to control our script we will have to specify our timing controls and where we would like to save the output from the solver. This is simple enough: 
85 
\begin{verbatim} 
86 
#Script/Iteration Related 
87 
t=0 #our start time, usually zero 
88 
tend=5.*60. #seconds  time to end simulation 
89 
outputs = 200 # number of time steps required. 
90 
h=(tendt)/outputs #size of time step 
91 
i=0 #loop counter 
92 
#the folder to put our outputs in, leave blank "" for script path 
93 
#note this folder path must exist to work 
94 
save_path = "data/onedheatdiff001" 
95 
\end{verbatim} 
96 
Now that we know our inputs we will build a domain using the \verb Rectangle() function from \verb Finley . The four arguments allow us to define our domain \verb rod as: 
97 
\begin{verbatim} 
98 
rod = Rectangle(l0=mx,l1=my,n0=ndx,n1=ndy) 
99 
\end{verbatim} 
100 
In this form \verb rod does not represent any discrete points, but rather an area of \verb ndx*ndy cells that fit into a rectangular space with opposing vertices at the origin and the point \verb [mx,my] . The reason for this is so that the discrete points used in the solution can be specified by the user. That is, are discrete points used at the corners of each cell, the middle point of a cell or halfway along each side of the cell etc. Fortunately \verb escript offers an easy way to extract finite points from the domain \verbrod using the domain property function \verbgetX() . This function uses the vertices of each cell as finite points to solve in the solution. If we let \verbx be these finite points, then: 
101 
\begin{verbatim} 
102 
x = rod.getX() 
103 
\end{verbatim} 
104 
With a domain and all our required variables established, it is now possible to set up our PDE so that it can be solved by escript. The first step is to define the type of PDE that we are trying to solve in each time step. In this example it is a single linear PDE (in comparison to a system of PDEs which will discussed later) 
105 
we can define it by: 
106 
\begin{verbatim} 
107 
mypde=LinearSinglePDE(rod) 
108 
\end{verbatim} 
109 
In the next step we need to define the coefficients of the PDE. The linear 
110 
PDEs in \ESCRIPT provide a general interface to do this. Here we will only discuss a simplified form that suits our heat diffusion problem and refer to the \ESCRIPT user's guide for the general case. This simpler form 
111 
\footnote{In the form of the \ESCRIPT users guide which uses the Einstein convention 
112 
this equation is written as 
113 
$(A\hackscore{jl} u\hackscore{,l})\hackscore{,j}+D u =Y$} 
114 
is described by 
115 
\begin{equation}\label{eqn:commonform nabla} 
116 
\nabla.(A.\nabla u) + Du = f 
117 
\end{equation} 
118 
where $A$, $D$ and $f$ are known values. 
119 
The symbol $\nabla$ which is called the \textit{Nabla operator} or \textit{del operator} represents 
120 
the spatial derivative of what ever stands next to the right from it. Lets assume for a moment that we deal with a onedimensional problem then 
121 
\begin{equation} 
122 
\nabla = \frac{\partial}{\partial x} 
123 
\end{equation} 
124 
and we can write equation \ref{eqn:commonform nabla} as 
125 
\begin{equation}\label{eqn:commonform} 
126 
A\frac{\partial^{2}u}{\partial x^{2}} + Du = f 
127 
\end{equation} 
128 
if $A$ is constant. This is exactly the equation we need to solve 
129 
in each time step as described in equation~\ref{XXXXX}. When comparing equations \eqref{eqn:hd} and \eqref{eqn:commonform} we see that; 
130 
\begin{equation} 
131 
A = \kappa; D = \frac{\rho c \hackscore{p}}{h}; f = q \hackscore{H} 
132 
\end{equation} 
133 

134 
We need to revisit the general PDE~\ref{eqn:commonform nabla} under the light of a 
135 
two dimensional domain. As pointed out earlier \ESCRIPT is not designed 
136 
to solve onedimensional problems so the general PDE~\ref{eqn:commonform nabla} 
137 
needs to be read as a higher dimensional problem. In the case of 
138 
two spatial dimensions the Nable operator has in fact 
139 
two components $\nabla = (\frac{\partial}{\partial x} 
140 
\frac{\partial}{\partial y})$. If we spell out the general PDE~\ref{eqn:commonform nabla} and assume a constant coefficient $A$ it then takes the form 
141 
\begin{equation}\label{eqn:commonform2D} 
142 
A\hackscore{00}\frac{\partial^{2}u}{\partial x^{2}} 
143 
A\hackscore{01}\frac{\partial^{2}u}{\partial x\partial y} 
144 
A\hackscore{10}\frac{\partial^{2}u}{\partial y\partial x} 
145 
A\hackscore{11}\frac{\partial^{2}u}{\partial y^{2}} 
146 
+ Du = f 
147 
\end{equation} 
148 
We notice that for the higher dimensional case $A$ becomes a matrix. It is also 
149 
important to notice that the usage of the Nable operator creates 
150 
a compact formulation which is also independend from the spatial dimension. 
151 
So to make the general PDE~\ref{eqn:commonform2D} one dimensional as 
152 
shown in~\ref{eqn:commonform} we need to set 
153 
\begin{equation}\label{eqn:commonform2D} 
154 
A\hackscore{00}=A; A\hackscore{01}=A\hackscore{10}=A\hackscore{11} 
155 
\end{equation} 
156 

157 

158 
BLABLA 
159 
\TODO{explain boundray condition; ignore radiation} 
160 

161 

162 

163 

164 

165 
Because we have a symmetrical problem we will also need to set the symmetry on by: 
166 
\begin{verbatim} 
167 
myPDE.setSymmetryOn() 
168 
\end{verbatim} 
169 
To input the PDE into \esc it must be compared with the general form\footnote{Available in section ?? of the users guide to \esc }. For most simple PDEs however, the general form is over complicated and confusing. Thus for this example, we will use a simplified version 
170 

171 

172 

173 

174 
Additionally we must also consider the boundary conditions of our PDE. They take the form: 
175 
\begin{equation} 
176 
\eta \hackscore{j} A\hackscore{jl} u\hackscore{,l} + du = y 
177 
\end{equation} 
178 

179 

180 
NEED TO WORK ON THIS SECTION 
181 

182 
We now need to specify Our boundary conditions and initial values. The initial values required to solve this PDE are temperatures for each discrete point in our domain that we wish to solve for. We will set our bar to: 
183 
\begin{verbatim} 
184 
T = Tref 
185 
\end{verbatim} 
186 
Boundary conditions are a little more difficult. Fortunately the escript solver will handle our insulated boundary conditions. However, we will need to apply our heat source $q_{H}$ to the end of the bar at $x=0$ . escript makes this easy by letting us define areas in our domain. To retrieve all the finite points in our domain we will use 
187 

188 
END WORK ON THIS SECTION 
189 

190 
\section{Plot total heat} 
191 
\TODO{show the script} 
192 

193 
\TODO{explain how to use matlibplot to visualize the total heat integral(rho*c*T) over time} 
194 

195 
\section{Plot Temperature Distribution} 
196 
\TODO{explain how to use matlibplot to visualize T} 
197 

198 

199 
\TODO{Move this to the 2D section as an advanced topic} 
200 
The final stage to our problem is exporting the data we have generated and turn our data and visualisation. It is best to export the calculated solutions at each time increment. escript has the inbuilt function \verbsaveVTK() which makes this step very easy. saveVTK takes two arguments, the path and the filename. We are goind to use the \verbos.path.join command to join a subdirectory which must already exist with a file name. The string opperator \verb% allows us to increment our file names with the value \verbi. In substring \verb %03d does a number of things; 
201 
\begin{itemize} 
202 
\item \verb 0 becomes the padding number; 
203 
\item \verb 3 tells us the number of padding numbers that are required; and 
204 
\item \verb d indicates the end of the \verb % operator. 
205 
\end{itemize} 
206 
To increment a \verb %i is required directly after the operation the string is involed in. The second arugment of the \verb saveVTK function is the \verb sol=T where \verb T is the solution of our PDE for a given time steo. Thus the export command becomes: 
207 
\begin{verbatim} 
208 
saveVTK(os.path.join(save_path,"data%03d.xml") %i,sol=T) 
209 
\end{verbatim} 
210 
Visualisation is then implemented through \verb mayavi which provides a command line of gui interface to develop plots to suit our output data. For this example a surface plot is appropriate. 