how can i resolve the problem of using indices?

i am writing a code for unipolar RZ.Where '1' in the input is transformed to 1 and 0 over the same period of input.whereas 0 in the input remains as 0 throughout the period. The code goes like this

Ndata=8
input = randint(1, 2*Ndata); % make polar signal
for j=1:length(input)
    if input(j)==1
        output(j:(j/2))=1;
        output((j/2):((j/2)+1))=0;
    else
        output(j:(j/2))=0;
        output((j/2):((j/2)+1))=0;
    end
end

it gives me an error showing that "Warning: Integer operands are required for colon operator when used as index "??? "Subscript indices must either be real positive integers or logicals".

Kindly resolve my problem. Thanks.

[per isakson]

(j/2) cannot be used as an index, since it is not guaranteed to be an integer

Answer 1

[Walter Roberson]

As per indicates, j/2 might not be integral. In your expression

j:(j/2)

that does not matter because the rules of the colon operator increment by 1 starting from the initial value and the initial j will be integral, so the non-integral j/2 will never be generated.

On the other hand, j/2 will always be less than j unless j is negative, and you cannot have negative indices. So j:j/2 will always be empty (or invalid anyhow.) And if you switch to j/2:j then you do hit the problem of non-integral indices.

You need to be concerned with using floor() to force integral values.

Caution: randint is going away. You should replace it with

randi([0 1], 1,2*Ndata)

[s]

i mean to say that if the duration of my input bit is T and the input bit is 1.At the output the output is 1 for T/2 and 0 for the next half of the input duration.So can u tell how can i incorporate floor() in my code so as to get my output displayed as above. Is randint doing the same job as randi([0 1], 1,2*Ndata)? Thanks in advance.

[Walter Roberson]

output(1:floor(T/2)) = 1;
output(floor(T/2)+1:end) = 0;

randint is defined as generating integers 0 and 1 with equal probability, which is what randi([0 1]) would do.

[s]

Ndata=8
input = randint(1, Ndata);
   for j=1:length(input)
       if input(j)==1
           output(1:floor(j/2)) = 1
           output(floor(j/2)+1:end) = 0
       else 
           output(1:floor(j/2)) = 0
           output(floor(j/2)+1:end) = 0
       end
   end

i did this but this is not returning any output as i stated below. if the duration of my input bit is T and the input bit is 1. At the output the output is 1 for T/2 and 0 for the next half of the input duration. else if input is 0 the output is 0 for the whole duration. Kindly help me resolving this.

[s]

for suppose the input is "0 0 0 1 1 0 1 0" the output is "00 00 00 10 10 00 10 00" for the same duartion of input.

[Walter Roberson]

You need to define the relationship between a duration of "T" and an indices. If the duration is in terms of indices then:

T = 15;         %duration of pulse in bits
Ndata = 8;

output = zeros(T, Ndata);
inputvals = randint(1, Ndata);

for j = 1 : length(inputvals)
  if inputvals(j) == 1
    output(1:floor(T/2), j) = 1;
    % output(floor(j/2)+1:end, j) = 0;  %already 0 do not need to set it
  else 
    % output(:,j) = 0;                  %already 0 do not need to set it
  end
end

output = reshape(output, 1, []);

This can be further optimized:

T = 15;         %duration of pulse in bits
Ndata = 8;

output = zeros(T, Ndata);
inputvals = randi([0,1], 1, Ndata);

output(1:floor(T/2), logical(inputvals)) = 1;

output = reshape(output, 1, []);

[s]

inputvals =

     1     0     1     0     0     0     1     0

output =

     0     0     0     0     0     0     0     0
I got this ouput for T=1 and Ndata=8 using the above code.This is not the desired output.I have attached the document for your reference.I am talking about unipolar_Rz.