Cell array and strings

I have initialized a series of vectors named vector1, vector2, ... vector8, and am using the code below.

for k=1:numVecs
   myCell{1,k}=sprintf('vector%d',k);

     myCell{2,k}=???
  end

I want to put, in the second row, kth entry of the cell, the content of variable vectork, where k=1,2, etc. I tried:

myCell{2,k}=(strcat('vector',num2str(k)),1:8),

but MATLAB says "Expression or statement is incorrect--possibly unbalanced (, {, or [." right before the 1. Does anyone have a suggestion? Thanks in advance.

[OCDER]

The problem starts right here:

"I have initialized a series of vectors named vector1, vector2, ... vector8"

How did you do this?

[J.S.]

Simply typed at the top of my code:

vector1=[1,4,5]; vector2=[4,18,7];

....

etc. This is intentional. The user is intended to manually edit the contents of the vectors to fit the situation

[Stephen23]

"I have initialized a series of vectors named vector1, vector2, ... vector8"

If you are creating numbered variables then you are doing something wrong. Trying to access numbered variables is one way that beginners force themselves into writing slow, complex, buggy code. Read this to know why:

https://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval

Indexing is very neat, very simple, very easy to write, very easy to debug. You should be using indexing, because then your code would be simpler, more efficient, and you avoid this entire situation:

vec{1} = ...
vec{2} = ...
vec{3} = ...

Or if the vectors are the same size then put them into one matrix:

mat(:,1) = ...
mat(:,2) = ...
mat(:,3) = ...

In both cases you can trivially and efficiently access the required data using indexing.

Answer 1

[OCDER]

Take a read here: https://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval

Your vector1, vector2, etc naming scheme will lead to a lot of headaches and you'll end up recoding everything. See the simple fix below.

Instead of :

vector1 = [1, 4, 5];
vector2 = [4 18 17];

Do this:

vector{1} = [1, 4, 5];
vector{2} = [4 18 17];

To access "vector1", use "vector{1}". 

If you want a 2xN cell storing a variable name on row 1, the vector on row 2, then do this:

vecnames = strsplit(sprintf('vector%d;', 1:length(vector)), ';');
C = vertcat(vecnames(1:length(vector)), vector);
C =
  2×2 cell array
    'vector1'       'vector2'   
    [1×3 double]    [1×3 double]

[Jan]

By the way: There is a valuable but undocumented function:

vecnames = sprintfc('vector%d', 1:length(vector))

[OCDER]

Nice! This is so much better. Why this is undocumented puzzles me...

[Guillaume]

sprintfc has been superseded by compose (introduced in R2016b) which is documented and more powerful.

vecnames = compose('vector%d', 1:numel(vector))

[OCDER]

Cool, so compose is the official version of sprintfc. But compose is 20% slower as it adds more capabilities over sprintfc. This difference won't matter much for most applications though (micro optimization).

t1 = timeit(@() sprintfc('vector%d', 1:100000), 1);
t2 = timeit(@() compose('vector%d', 1:100000), 1);
t1/t2 = 
   0.798   %sprintfc is ~20% faster

[Jan]

On my R2016b:

v = 1:1e5;
t1 = timeit(@() sprintfc('vector%d', v), 1)
t2 = timeit(@() compose('vector%d', v),  1)
fmt = string('vector%d');
t3 = timeit(@() fmt.compose(v),  1)

 t1 =  0.0736
 t2 =  0.0717
 t3 =  0.0714

[OCDER]

t1 = 0.0540
t2 = 0.0702
t3 = 0.0526  

Version 2017a. Maybe compose got more lines of codes in 2017a+ ?

[Yair Altman]

Note that compose is not a full replacement for sprintfc, although it does cover the most common use-case. Specifically, sprintfc's 3rd [isImaginary] input flag, and its 2nd/3rd output args [errorMsg and isLeft] are not supported by compose.

Additional info: https://undocumentedmatlab.com/blog/sprintfc-undocumented-helper-function