"clear all" is not translated correctly #10
Description
MATLAB input
# Clear all variables
clear all
# Show numbers in a long, yet intuitive form
format long g
# Number of collocation points
N = 1000;
NN = 100000;
# Integration interval
xx0 = 0;
xxf = 50;
# Vector
n = 0:N;
# Extrema grid
x = -cos(pi*n'/N);
# Transformation to [xx0, xxf]
xtrans = (xxf-xx0)/2*(x+1);
# linearly spaced grid
xlin = linspace(-1,1,NN+1)';
Tlin = cos(acos(xlin)*n);
xtranslin = (xlin+1)*(xxf-xx0)/2;
# Chebyshev polys of the first kind
T = cos(acos(x)*n);
# Now for arrays that do not include the endpoints
xsub = x(2:N);
Tsub = T(2:N,:);
Usub = diag(1./sqrt(1-xsub.^2))*sin(acos(xsub)*n);
dTsub = Usub*diag(n);
# Add the endpoints
dT = [-(n.^2).*(-1).^(n); dTsub ; (n).^2];
# LHS without a (coefficient vector)
H = 2/(xxf-xx0)*dT;
# RHS, essentially what we're trying to integrate
F = gsimp(xtrans);
# Initial condition; with definite integrals y(xx0) = 0
H(1,:) = T(1,:);
F(1) = 0;
# coefficient vector
a = H\F;
dya = T\F;
# solution to ODE vector
y = T*a;
# definite integral
y = y-y(1);
# y on the linear grid
ylin = Tlin*a;
dylin = Tlin*dya;
# exact indefinite solution per Wolfram Alpha
#yexact = -1/10*exp(-xtranslin).*(-2*sin(2*xtranslin)+cos(2*xtranslin)+5);
#yexact = real(yexact);
# exact definite solution
#yexact = yexact-yexact(1);
# error to Chebyshev approximation
#err = abs(yexact-ylin);
#errrms = sqrt(err'*err/(NN+1))
# y approximated by quadratic integration function
#yquad = quad("g", 0, inf);
# y solved from ODE
lsode_options("absolute tolerance", 1e-18);
yode = lsode("g", 0, xtranslin);
# error in approximation
#errquad = abs(yexact(end)-yquad)
diffodech = abs(yode-ylin);
errodech = sqrt(diffodech'*diffodech/(NN+1));
figure(1)
semilogy(xtranslin,diffodech)
figure(2)
plot(xtranslin,ylin,'r',xtranslin(2:end),dylin(2:end),'g')Julia output
using PyPlot
# Clear all variables
clear all()
# Show numbers in a long; yet intuitive form
format long g
# Number of collocation points
N = 1000
NN = 100000
# Integration interval
xx0 = 0
xxf = 50
# Vector
n = 0:N
# Extrema grid()
x = -cos(pi*n'/N)
# Transformation to [xx0, xxf]
xtrans = (xxf-xx0)/2*(x+1)
# linearly spaced grid()
xlin = range(-1,1,length=NN+1)'
Tlin = cos(acos(xlin)*n)
xtranslin = (xlin+1)*(xxf-xx0)/2
# Chebyshev polys of the first kind
T = cos(acos(x)*n)
# Now for arrays that do not include the endpoints
xsub = x[2:N]
Tsub = T[2:N,:]
Usub = diag(1./sqrt(1-xsub.^2))*sin(acos(xsub)*n)
dTsub = Usub*diag(n)
# Add the endpoints
dT = [-(n.^2).*(-1).^(n) dTsub [n].^2]'
# LHS without a [coefficient vector]
H = 2/(xxf-xx0)*dT
# RHS; essentially what we're trying to integrate
F = gsimp[xtrans]
# Initial condition; with definite integrals y[xx0] = 0
H[1,:] = T[1,:]
F[1] = 0
# coefficient vector
a = H\F
dya = T\F
# solution to ODE vector
y = T*a
# definite integral
y = y-y[1]
# y on the linear grid()
ylin = Tlin*a
dylin = Tlin*dya
# exact indefinite solution per Wolfram Alpha
#yexact = -1/10*exp(-xtranslin).*(-2*sin(2*xtranslin)+cos(2*xtranslin)+5)
#yexact = real(yexact)
# exact definite solution
#yexact = yexact-yexact[1]
# error to Chebyshev approximation
#err = abs(yexact-ylin)
#errrms = sqrt(err'*err/(NN+1))
# y approximated by quadratic integration function
#yquad = quad("g", 0, inf)
# y solved from ODE
lsode_options["absolute tolerance", 1e-18]
yode = lsode["g", 0, xtranslin]
# error in approximation
#errquad = abs(yexact[end]-yquad)
diffodech = abs(yode-ylin)
errodech = sqrt(diffodech'*diffodech/(NN+1))
figure(1)
semilogy(xtranslin,diffodech)
figure(2)
plot(xtranslin,ylin,"r',xtranslin[2:end],dylin[2:end],'g")which returns the error (when this code is saved as "definite-integration.jl" and loaded with include("definite-integration.jl")):
ERROR: LoadError: syntax: extra token "all" after end of expression
Stacktrace:
[1] include(::String) at ./client.jl:403
[2] top-level scope at none:0
in expression starting at /data/GitHub/mine/math/julia-scripts/definite-integration.jl:4As the error line is line 4, for which the corresponding input code was MATLAB-compatible, this seems like a valid bug report regardless of whether this translator should be GNU Octave compatible too.
Expected Julia output
(Not sure, I'm a Julia beginner, was hoping this would help me learn Julia).