This is maybe a little too long to explain on an email - but I'll try. I'll also assume you know what a Vector is, otherwise this is definitely too much for an email, I'll give this the 5 minute write up and hopefully others can correct my mistakes.
A matrix is an amazing unit of mathematics that can transform a Vector from one "space" to another. Think maya's "object" space and "world" space, thats _exactly_ what we're talking about.
A untransformed matrix at the origin would look like
1 0 0 0
0 1 0 0
0 0 1 0
0 0 0 1
This is called the identity matrix - it has no translation, rotation, scale or shearing applied to it. You can tell just by looking at it because the 4 rows are easily readable vectors in their own right.
1 0 0 0 <- X-axis vector of this "space"
0 1 0 0 <- Y-axis "" ""
0 0 1 0 <- Z-axis "" ""
0 0 0 1 <- translation of this "space"
The last column is always [0, 0, 0, 1] I think, just because they need to be for the math to work.
given the X-axis vector is <1, 0, 0> (perfectly aligned with scene-X) and the Y, Z vectors are the same - we can see this matrix is not rotated.
given the translation vector is <0, 0, 0> we can see this matrix is not translated.
given the 3 axis vectors are all exactly 1-unit long, we can see this hasnt been scaled in any axis.
given the 3 axis vectors are all at right-angles (easy to tell in this case, usually not so easy) we can see the matrix is not sheared.
So this is the matrix of a node sitting at the origin - completely untransformed.
Thats how you "read" a matrix, try moving a node around in maya and query its "worldMatrix[0]" in different positions, you'll quickly see whats going on.
Which brings us to matrix vs worldMatrix.
Matrices can be "added together" - which I believe is actually matricies being multiplied together, mathemically speaking - and is super efficient for calculating hierarchy transformations (because you can build a single "world" matrix to transform a vector in worldspace to an equivalent vector in local space.)
The "matrix" attribute is a description of the matrix relative to a node's parent (i.e. the node's local space). The "worldMatrix" as a description of the matrix relative to the scene. The worldMatrix can be computed by multiplying, in order, the local matrix of each parent of a node. The "parentMatrix" is a shortcut attribute in maya to access the worldMatrix of the node's immediate parent.
Knowing this, the "inverse" matrix is kinda what it sounds like. If a matrix is a math object to transform a vector from one space to another, the same matrix's "inverse" is a matrix to go the other way. In our world, I think, matrices will always have a possible inverse - this isn't the case in pure math world.
The details of "how" you multiply matrices, or "how" you multiply a vector by a matrix isn't important. But knowing what result to expect _is_.
I'd just start creating some Vectors, and multiplying them by a matrix that you can move around.
e.g.
vector = pymel.core.dt.Vector(1, 0, 0)
matrix = pymel.core.PyNode("someTransformNode").worldMatrix[0].get()
result = vector * matrix
Just play with it. Try using inverses too.
Finally, I'd point out there _is_ a difference between Points and Vectors. (points are actually length=4). - I think Multiplying a vector by a matrix will essentially rotate your vector in to the new space's orientation, while multiplying a Point will fully transform a translated point in to a new space.
Good luck!