2013.01.28-ece595e-l09
TRANSCRIPT
-
7/29/2019 2013.01.28-ECE595E-L09
1/18
ECE 595, Section 10Numerical Simulations
Lecture 9: Programming for Linear
Algebra
Prof. Peter Bermel
January 28, 2013
-
7/29/2019 2013.01.28-ECE595E-L09
2/18
Outline
Recap from Friday
Application Examples
Electrostatic potential (Poissons equation)
1D array of charge
2D grid of charge
Arrays of interacting spins
1D interaction along a chain
2D nearest-neighbor coupling
1/28/2013 ECE 595, Prof. Bermel
-
7/29/2019 2013.01.28-ECE595E-L09
3/18
Recap from Friday
Eigenproblem Solution Techniques
Power Methods
Inverse Iteration
Atomic Transformations
Factorization Methods
Linear algebra software packages
Basic Linear Algebra Subroutines (BLAS)
Linear Algebra Package (LAPACK)
1/28/2013 ECE 595, Prof. Bermel
-
7/29/2019 2013.01.28-ECE595E-L09
4/18
Electrostatic Potential Example
Consider an array of charges governed by
Gauss law:
=
/
Using the definition of potential yields
Poissons equation:
2
=
/
Consider solving this equation for an arbitrary
set of charges what to do?
1/28/2013 ECE 595, Prof. Bermel
-
7/29/2019 2013.01.28-ECE595E-L09
5/18
Electrostatic Potential Example
Strictly speaking, continuous fields have an
uncountable number of possible values, and
cannot be evaluated numerically
Key is to transform from continuous fieldvalues to those on a grid
Increase resolution as needed
1/28/2013 ECE 595, Prof. Bermel
1 2 3 45 6 7 8 910
-
7/29/2019 2013.01.28-ECE595E-L09
6/18
Electrostatic Potential Solution: 1D
Approximate Laplacian in 1D with:
2 1 2 + +12 Where h is the grid spacing
Sets up the linear algebra problem:2 1 01 2 10 1 2
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
0 0 0
2 1 0
1 2 10 1 2
123
21
=2
123
21
1/28/2013 ECE 595, Prof. Bermel
-
7/29/2019 2013.01.28-ECE595E-L09
7/18
Electrostatic Potential Solution: 1D
In MATLAB, use:
1/28/2013 ECE 595, Prof. Bermel
-
7/29/2019 2013.01.28-ECE595E-L09
8/18
Electrostatic Potential Solution: 1D
Use backslash operator to solve linear algebra
problems of the form =
1/28/2013 ECE 595, Prof. Bermel
Charge distribution
Electric potential
-
7/29/2019 2013.01.28-ECE595E-L09
9/18
Electrostatic Potential Solution: 2D
Connections in two directions creates a total of 5 non-vanishingdiagonals in our linear algebra problem:
4 1 01
4 1
0 1 4
1 0 0
0 1 0
0 0 1 1 0 0
0 1 0
0 0 1
4 1 0
1 4 10 1 4
123
21
= 2
123
21
1/28/2013 ECE 595, Prof. Bermel
-
7/29/2019 2013.01.28-ECE595E-L09
10/18
Electrostatic Potential Solution: 2D
1/28/2013 ECE 595, Prof. Bermel
Charge distribution in 2D
(7x7 grid)Electrostatic potential in 2D
(7x7 grid)
-
7/29/2019 2013.01.28-ECE595E-L09
11/18
Spin Array Example
Consider an array of spins , coupled by anexchange interaction
Ising-model Hamiltonian is given by:
=
+
The brackets often are interpreted to meannearest-neighbor interactions only
1/28/2013 ECE 595, Prof. Bermel
-
7/29/2019 2013.01.28-ECE595E-L09
12/18
Spin Array Solution: 1D
Convert Hamiltonian into matrix, assuming nearest neighborinteraction only:
=
0 0
0 0 0
0 0 0
0 0 0
0 0 00 0 0
0 0 0
0 0
Use Schrodinger equation to set up eigenproblem:
= Choose basis spinor wavefunction :
= 1, 2, ,
1/28/2013 ECE 595, Prof. Bermel
-
7/29/2019 2013.01.28-ECE595E-L09
13/18
Spin Array Solution: 1D
In MATLAB, use the following:
1/28/2013 ECE 595, Prof. Bermel
-
7/29/2019 2013.01.28-ECE595E-L09
14/18
Spin Array Solution: 1D
1/28/2013 ECE 595, Prof. Bermel
-
7/29/2019 2013.01.28-ECE595E-L09
15/18
Spin Array Solution: 2D
Convert Hamiltonian into matrix, assuming nearest neighborinteractions in 2 directions:
=
0 0
0 00 00 0
0 00 00 0
0 0
Use Schrodinger equation to set up eigenproblem:
= Choose basis spinor wavefunction :
= 1, 2, ,
1/28/2013 ECE 595, Prof. Bermel
-
7/29/2019 2013.01.28-ECE595E-L09
16/18
Spin Array Solution: 2D
In MATLAB, use the following:
1/28/2013 ECE 595, Prof. Bermel
-
7/29/2019 2013.01.28-ECE595E-L09
17/18
Spin Array Solution: 2D
1/28/2013 ECE 595, Prof. Bermel
-
7/29/2019 2013.01.28-ECE595E-L09
18/18
Next Class
Is on Wednesday, Jan. 30
Will discuss numerical tools for
simulating eigenproblems further
1/28/2013 ECE 595, Prof. Bermel