Professional Documents
Culture Documents
0704 1622 PDF
0704 1622 PDF
Two key concepts underpinning quantum physics are when studying 1D quantum mechanics is a student treat-
the Schrodinger equation and the Born probability equa- ing Ψ(x) and |Ψi interchangeably, ignoring the fact that
tion. In 1930 Dirac introduced bra-ket notation for state the first is a scalar but the ket corresponds to a column
vectors and operators.1 This notation emphasized and vector. For example, they may write incorrectly
clarified the role of inner products and linear function
spaces in these two equations and is fundamental to
our modern understanding of quantum mechanics. The
Schrodinger equation tells us how the state Ψ of a particle hp||Ψi|xi = |Ψihp|xi (incorrect!) (4)
evolves in time. In bra-ket notation, it reads
Note that in the MATLAB syntax the object V (:, 3) % Three-point finite-difference representation of Laplacian
% using sparse matrices, where you save memory by only
specifies the column vector consisting of all the elements % storing non-zero matrix elements
in column 3 of matrix V . Similarly V (2, :) is the row e = ones(N,1); Lap = spdiags([e -2*e e],[-1 0 1],N,N)/dx^2;
vector consisting of all elements in row 2 of V ; V (3, 1) is % Total Hamiltonian
the element at row 3, column 1 of V ; and V (2, 1 : 3) is hbar = 1; m = 1; % constants for Hamiltonian
H = -1/2*(hbar^2/m)*Lap + spdiags(U,0,N,N);
a row vector consisting of elements V (2, 1), V (2, 2) and
V (2, 3). % Find lowest nmodes eigenvectors and eigenvalues of sparse matrix
nmodes = 3; options.disp = 0;
[V,E] = eigs(H,nmodes,’sa’,options); % find eigs
[E,ind] = sort(diag(E));% convert E to vector and sort low to high
IV. ARBITRARY POTENTIALS V = V(:,ind); % rearrange corresponding eigenvectors
When doing numerical calculations, it is important to The separable solutions |ΨE (t)i are only a subset of
minimize the effect of rounding errors by choosing units all possible solutions of Eq. (1). Fortunately, they are
such that the variables used in the simulation are of the complete set so that we can construct the general solution
4
0.15
and ρ(x) for the particle being entirely in the left well
0.1
2
ρ(x) = | |ψE0 (x)| − |ψE1 (x)| | . (12)
0.05
of the ground state |ΨE0 i and first excited state |ΨE1 i % Plot rho(x,T) and rescaled potential energy Usc
of the double well system. As snapshots of the program plot(x,rho,’o-k’, x, Usc,’.-b’); axis([-L/8 L/8 -1 6]);
lgnd_str = [repmat(’T = ’,1,1),num2str(T)];
output show in Fig. 2, the particle is initially completely text(-0.12,5.5,lgnd_str, ’FontSize’, 18);
localized in the rightmost well. However, due to E0 6= E1 , xlabel(’x [m]’, ’FontSize’, 24);
ylabel(’probability density [1/m]’,’FontSize’, 24);
the probability density
pause(0.05); % wait 0.05 seconds
1 end
ρ(x, t) = [ |ψE0 (x)|2 + |ψE1 (x)|2 +
2
2|ψE0 (x)||ψE1 (x)|cos2 ((E1 − E2 )t/~) ] (10)
5
6 6
lates and displays the time evolution of a wavepacket for
T = 0s T = 2.8189s one of two possible potentials, either U = 0 or a step
5 5 potential U = Uo Θ(x − L). The initial wavepacket is
generated as the Fast Fourier Transform of a Gaussian
probability density [1/m]
4 4
They can extract the group velocity from their numerical
3 3
simulation and observe that indeed vg = 2vw for a typ-
2 2 ical wave packet. Students can also observe that, while
vg matches the particle speed from classical mechanics,
1 1
the wavepacket spreads out as time elapses.
0 0 In Program 6, we propagate the wave function forward
via the formal solution
−1 −1
−0.6 −0.4 −0.2 0 0.2 0.4 0.6 −0.6 −0.4 −0.2 0 0.2 0.4 0.6
x [m] x [m] |Ψ(t)i = exp(−iHt/~)|Ψ(0)i, (13)
%****************************************************************
% Program 6: Wavepacket propagation using exponential of H
%****************************************************************
hbar = 1;
%***************************************************************
% Simulate rho(x,T) and plot for each T
%*************************************************************** 0.05
for t = 1:NT; % time index for loop
% calculate probability density rho(x,T)
psi = E*psi; % calculate new psi from old psi
rho = conj(psi).*psi; % rho(x,T) 0
0 10 20 30 40 50 60 70 80 90 100
plot(x,rho,’k’); % plot rho(x,T) vs. x x [m]
axis([0 L 0 0.15]); % set x,y axis parameters for plotting
xlabel(’x [m]’, ’FontSize’, 24);
ylabel(’probability density [1/m]’,’FontSize’, 24); FIG. 3: Output of Program 6 showing a wavepacket encoun-
pause(0.05); % pause between each frame displayed
tering step potential of height ∼ hEi located at x/L = 0.5 at
end different times.
% Calculate Reflection probability
R=0;
for a=1:N/2; VIII. CONCLUSIONS
R=R+rho(a);
end
R=R*dx
One benefit of incorporating numerical simulation into
the teaching of quantum mechanics, as we have men-
tioned, is the development of student intuition. Another
is showing students that non-ideal, real-world problems
can be solved using the concepts they learn in the class-
numerical simulation for hEi/Uo ≈ 1. This discrepancy room. However, our experimentation incorporating these
can be reduced significantly by increasing the number of simulations in quantum physics at WPI during the past
points in the simulation to 1600 but only at the cost of year has shown us that the most important benefit is a
significantly slowing down the speed of the computation. type of side-effect to doing numerical simulation: the ac-
For our purposes, the importance comparing the analyt- ceptance on an intuitive level by the student that func-
ical and numerical calculations is that it gives student tions are like vectors and differential operators are like
a baseline from which to form an opinion or intuition matrices. While in the present paper, we have only had
regarding the accuracy of Eq. (14). sufficient space to present the most illustrative MATLAB
7
8
!"#
∗
Electronic address: garcia@wpi.edu mechanics,” Am. J. Phys. 64, 31-34 (1996).
1
P. A. M. Dirac, The Principles of Quantum Mechanics, 1st 8
A. Goldberg, H. M. Schey, J. L. Schwartz, “Computer-
ed., (Oxford University Press, 1930). Generated Motion Pictures of One-Dimensional Quantum-
2
Born’s law is stated as a proportionality because an addi- Mechanical Transmission and Reflection Phenomena,”
tional factor is necessary depending on the units of |Ψi Am. J. Phys. 35, 177-186 (1967).
3 9
C. C. Silva and R. de Andrade Martins, “Polar and ax- C. Singh, M. Belloni, and W. Christian, “Improving stu-
ial vectors versus quaternions,” Am. J. Phys. 70, 958-963 dents’ understanding of quantum mechanics,” Physics To-
(2002). day, August 2006, p. 43.
4 10
R. W. Robinett, Quantum Mechanics: Classical Results, These MATLAB commands are explained in an exten-
Modern Systems, and Visualized Examples, (Oxford Uni- sive on-line, tutorial within MATLAB and which is also
versity Press, 1997). independently available on the MATHWORKS website,
5
H. Gould, “Computational physics and the undergraduate http://www.mathworks.com/.
curriculum,” Comput. Phys. Commun. 127, 610 (2000); 11
D. J. Griffiths, Introduction to Quantum Mechanics, 2nd
J. Tobochnik and H. Gould, “Teaching computational Ed., Prentice Hall 2003.
physics to undergraduates,” in Ann. Rev. Compu. Phys. 12
A. Garcia, Numerical Methods for Physics, 2nd Ed., (Pren-
IX, edited by D. Stauffer (World Scientific, Singapore, tice Hall, 1994).
13
2001), p. 275; H. Gould, J. Tobochnik, W. Christian, An G. Lindblad, “Quantum Mechanics
Introduction to Computer Simulation Methods: Applica- with MATLAB,” available on internet,
tions to Physical Systems, (Benjamin Cummings, Upper http://mathphys.physics.kth.se/schrodinger.html.
14
Saddle River, NJ, 2006) 3rd ed.. See the user file exchange at
6
R. Spenser, “Teaching computational physics as a labora- http://www.mathworks.com/matlabcentral/.
tory sequence,” Am. J. Phys. 73, 151-153 (2005).
7
D. Styer, “Common misconceptions regarding quantum