How to describe the intersection of two sets? The following is a homework problem:

Let 
  $$\begin{align*}
W_1 &= \{(a_1, a_2, a_3) \in\mathbb{R}^3 \mid a_1 = 3  a_2\text{ and }a_3 = -a_2\}\\
W_2 &= \{(a_1, a_2, a_3) \in \mathbb{R}^3 \mid a_1 - 4 a_2 - a_3 = 0\}
\end{align*}$$
  Describe the intersection of W1 and W2 and observe that it is a subspace.

I realize that the intersection is 
$$\{ (a_1, a_2, a_3) \in \mathbb{R}^3 \mid a_1 = 3  a_2\text{ and }a_3 = -a_2\text{ and }a_1 - 4  a_2 - a_3 = 0 \}.$$ 
Further, I've observed that this seems to just be the set, $W_1$, based on the conditions.  I just kind of haphazardly messed with the formulas to try to figure stuff out with no real methodology. 
What is the method to solving these kinds of problems?  How do I solve this and similar problems?
 A: Assume that $(a_1,a_2,a_3)\in W_1\cap W_2$. As you note, you  must have $a_1=3a_2$ and $a_3=-a_2$, in order for the point to be in $W_1$. And you must have $a_1-4a_2-a_3=0$ in order to be in $W_2$.
That means that the point must satisfy the following three equations:
$$\begin{align*}
a_1 -3a_2 &= 0\\
a_2 + a_3 &= 0\\
a_1 - 4a_2 - a_3 &= 0
\end{align*}$$
So the intersection consists precisely of all solutions to this system of linear equations. 
So the problem devolves into solving a system of linear equations. There are plenty of ways of doing this; e.g., Gaussian Elimination, Back Substitution, etc.
Here, we obtain that if $a_1 = 3a_2$ and $a_2=-a_3$, then necessarily $a_1-4a_2 - a_3 = 0$, so that the third equation is a consequence of the first two, which is what you observed: it turns out that $W_1\subseteq W_2$, so $W_1\cap W_2 = W_1$.
In general, if your sets are given as solution sets of equations, then finding the intersection is equivalent to solving a system of equations; whether this is easy or not will depend on the equations. And for even more generality, it may be extremely hard to say what the intersection is explicitly; for instance, trying to figure out exactly what integers are in 
$$\{ p\in\mathbb{Z}\mid p\text{ is prime}\}\cap \{n\in\mathbb{Z}\mid n=p+2\text{ for some prime }p\}$$
is the same as trying to figure out exactly what the twin primes are, and it is still unknown whether there are infinitely many of them or not....
