One pivot: entering column, ratio test, and row reduction

A single move of the algorithm has two decisions and one calculation. First, choose which variable to increase by scanning the top row for the largest negative number. If the cookie column holds $-3$, the most negative entry, you move along the cookie axis.

Second, decide how far to go with the ratio test: for the entering column, divide each constraint row's total by that row's positive entry in the entering column. You choose the smallest ratio, because any larger ratio would land outside the feasible region. If two ratios tie for smallest, take the row with the lower index.

Then comes the calculation, Gaussian elimination on that column. Scale the chosen row so its entry in the entering column becomes $1$, then add multiples of it to the other rows so every other entry in that column becomes $0$. Now you read coordinates straight off the tableau: a column that is a single $1$ with zeros elsewhere means that variable equals its paired total, while a column still mixing several values reads as zero for that variable. The upper-right corner tracks revenue at the current vertex.