Consider Preallocating for Speed

I have the following code and I am being informed that I can "preallocate for speed". When attempting to set up an array which I think is the solution to this issue, I am told "Index exceeds the number of array elements (1)".

What is the correct array to use for the code?

A= 2;         %area (m^2)
T = 0.1;      %tau, integral time constant (min)
h=0.001;      %stepsize
t_final= 100; %the total time (min)
N=t_final/h;

Kc1 = 1;          %p

roportional gain, m^2/min
t1(1)=0;          %Kc=1 at 0 minutes
x1(1)=0;          %deviation of liquid height (difference between actual height and desired height)
v1(1)=2;          %derivative of change in height with respect to time
for i=1:N
t1(i+1)=t1(i)+h;
x1(i+1)=x1(i)+h*(v1(i));
v1(i+1)=v1(i)+h*((-Kc1*v1(i)/A)-(Kc1/(T*A))*x1(i));
end

Kc2 = 2;
t2(1)=0;
v2(1)=2;
x2(1)=0;
for i=1:N
t2(i+1)=t2(i)+h;
x2(i+1)=x2(i)+h*(v2(i));
v2(i+1)=v2(i)+h*((-Kc2.*v2(i)/A)-(Kc2./(T*A))*x2(i));
end

Kc3 = 3;
t3(1)=0;
v3(1)=2;
x3(1)=0;
for i=1:N
t3(i+1)=t3(i)+h;
x3(i+1)=x3(i)+h*(v3(i));
v3(i+1)=v3(i)+h*((-Kc3.*v3(i)/A)-(Kc3./(T*A))*x3(i));
end

Kc4 = 4;
t4(1)=0;
v4(1)=2;
x4(1)=0;
for i=1:N
t4(i+1)=t4(i)+h;
x4(i+1)=x4(i)+h*(v4(i));
v4(i+1)=v4(i)+h*((-Kc4.*v4(i)/A)-(Kc4./(T*A))*x4(i));
end

Kc5 = 5;
t5(1)=0;
v5(1)=2;
x5(1)=0;
for i=1:N
t5(i+1)=t5(i)+h;
x5(i+1)=x5(i)+h*(v5(i));
v5(i+1)=v5(i)+h*((-Kc5.*v5(i)/A)-(Kc5./(T*A))*x5(i));
end

Kc75 = 7.5;
t6(1)=0;
v6(1)=2;
x6(1)=0;
for i=1:N
t6(i+1)=t6(i)+h;
x6(i+1)=x6(i)+h*(v6(i));
v6(i+1)=v6(i)+h*((-Kc75.*v6(i)/A)-(Kc75./(T*A))*x6(i));
end

Kc10 = 10;
t7(1)=0;
v7(1)=2;
x7(1)=0;
for i=1:N
t7(i+1)=t7(i)+h;
x7(i+1)=x7(i)+h*(v7(i));
v7(i+1)=v7(i)+h*((-Kc10.*v7(i)/A)-(Kc10./(T*A))*x7(i));
end

Kc20 = 20;
t8(1)=0;
v8(1)=2;
x8(1)=0;
for i=1:N
t8(i+1)=t8(i)+h;
x8(i+1)=x8(i)+h*(v8(i));
v8(i+1)=v8(i)+h*((-Kc20.*v8(i)/A)-(Kc20./(T*A))*x8(i));
end

[Stephen23]

Duplicating code like that is a very strong code smell

https://en.wikipedia.org/wiki/Code_smell

and seems to be caused by forcing meta-data into the variable names (Kc1, Kc2, Kc5, Kc10, Kc20).

To simplify your code keep the meta-data in a separate array and use indexing rather than separate variables. Then you can trivially avoid duplicating code like that.

Answer 1

[Daniel Bengtson]

You just need to initialize your vectors before entering the for loops. Something like the below sample would satisfy the preallocation for the kc1 case. You'd have to do the same for the other loops as well.

A= 2;         %area (m^2)
T = 0.1;      %tau, integral time constant (min)
h=0.001;      %stepsize
t_final= 100; %the total time (min)
N=t_final/h;
Kc1 = 1;          %proportional gain, m^2/min

%% preallocate vectors of zeros
t1 = zeros(N+1,1);
x1 = zeros(N+1,1);
v1 = zeros(N+1,1);


t1(1)=0;          %Kc=1 at 0 minutes
x1(1)=0;          %deviation of liquid height (difference between actual height and desired height)
v1(1)=2;          %derivative of change in height with respect to time
for i=1:N
t1(i+1)=t1(i)+h;
x1(i+1)=x1(i)+h*(v1(i));
v1(i+1)=v1(i)+h*((-Kc1*v1(i)/A)-(Kc1/(T*A))*x1(i));
end