# Determining properties of linear code from generator matrix

I have a problem in error-correcting codes.

Say we have a generator matrix of a linear binary code

$$g=\begin{pmatrix} 10011 \\ 01101 \end{pmatrix}$$

Q1: How many different codeword do we have? How many code bits can the code correct?

Say the generator matrix of a $$(6,2)$$ binary code is $$g=\begin{pmatrix} 101001\\ 011111 \end{pmatrix}$$

Q2: How many errors can it guarantee to detect?

My try:

For Q1:

The way I find the number of codeword is count the number of rows in $$g$$ and raise it as a power to 2. So the answer is $$2^2=4$$.

For detecting and correcting I’m not sure, should I find the hamming distance and subtract 1?

I hope someone can explain to me how I can find the number bits I can detect and correct.

If the minimal distance is $$d$$, then
• Up to $$d-1$$ errors can be detected. This means that if a sent codeword is modified by flipping at least one and up to $$d-1$$ bits, then the receiver can tell that some bits were flipped.
Indeed, suppose that $$w$$ is the codeword that was sent, and that $$w'$$ is obtained from $$w$$ by flipping at least one and up to $$d-1$$ bits. Since $$1 \leq d_H(w,w') < d$$ (where $$d_H$$ is Hamming distance), we see that $$w'$$ is not a codeword (since the distance between any two different codewords is always at least $$d$$), and so we can detect that errors have occurred.
• Up to $$\lfloor \frac{d-1}{2} \rfloor$$ can be corrected. This means that if a sent codeword is modified by flipping at most $$\lfloor \frac{d-1}{2} \rfloor$$ bits, then the original codeword can be recovered.
Indeed, suppose that $$w$$ is the codeword that was sent, and that $$w'$$ is obtained from $$w$$ by flipping up to $$\lfloor \frac{d-1}{2} \rfloor$$ bits. I claim that $$w$$ is the unique codeword at distance at most $$\lfloor \frac{d-1}{2} \rfloor$$ from $$w'$$. Indeed, if $$z$$ were another such codeword then $$d_H(w,z) \leq d_H(w,w') + d_H(w',z) \leq 2 \lfloor \tfrac{d-1}{2} \rfloor < d,$$ contrary to the definition of minimal distance.