The primary topics in this part of the specialization are: greedy algorithms (scheduling, minimum spanning trees, clustering, Huffman codes) and dynamic programming (knapsack, sequence alignment, optimal search trees).

Loading...

From the course by 斯坦福大学

Greedy Algorithms, Minimum Spanning Trees, and Dynamic Programming

435 ratings

The primary topics in this part of the specialization are: greedy algorithms (scheduling, minimum spanning trees, clustering, Huffman codes) and dynamic programming (knapsack, sequence alignment, optimal search trees).

From the lesson

Week 3

Huffman codes; introduction to dynamic programming.

- Tim RoughgardenProfessor

Computer Science

Hey, so guess what? We just designed our first dynamic

programming algorithm. That linear time algorithm for computing

the max weight independence set in a path graph is indeed an instantiation of the

general dynamic programming paradigm. Now I've deferred articulating the

general principles of that paradigm until now because I think they are best

understood through concrete examples. Now that we have one to relate them to,

let me tell you about these guiding principles.

We will in the coming lectures see many more examples.

So the key that unlocks the potential of the dynamic programming paradigm for

solving a problem is to identify a suitable collection of sub-problems.

And these, sub-problems have to satisfy a number of properties.

In our algorithm for computing max weight independent sets and path graphs, we had

N plus one sub problems, one for each prefix of the graph.

So formally, our Ithi sub problem in our algorithm, it was to compute the max

weight independent set of G sub I, of the path graph consisting only of the first I

vertices. So the first property that you want your

collection of subproblems to possess is it shouldn't be too big.

It shouldn't have too many different subproblems.

The reason being is, in the best case scenario, you're going to be spending

constant time solving each of those subproblems, so the number of subproblems

is a lower bound than the running time of your algorithm.

Now, in the Maximum Independent Set example, we did great.

We had merely a linear number of subproblems, and we did indeed get away

with a mere constant work for each of those subproblems, giving us our linear

running time bound overall. The second property you want and this

one's really the kicker, is there should be a notion of smaller subproblems and

larger subproblems. In the context of independence sets of

path graphs this was really easy to understand.

The subproblems were prefixes of the original graph and the more vertices you

had, the bigger the subproblem. So in general, in dynamic programming,

you systematically solve all of the subproblems beginning with the smallest

ones and moving on to larger and larger subproblems.

And for this to work, it better be the case that, at a given subproblem.

Given the solutions to all of the smaller sub problems it's easier to confer what

the solution to the current sub problem is.

That is solutions to previous sub problems are sufficient to quickly and

correctly compute the solution to the current sub problem.

The way this relationship between larger and smaller subproblems is usually

expressed is via recurrence and it states what the optimal solution to a given

subproblem is as a function of the optimal solutions to smaller subproblems.

And this is exactly how things played out in our independent set algorithm.

We did indeed have a recurrence. It just said, that the optimal value, the

maxwood independence head value for G sub I.

Was the better of two candidates. And we justified this using our thought

experiment. Either you just inherit the maximum

independence set value from the preceding sub problem from the I-1 sub problem.

Or you take the optimal solution from two sub problems back from GI-2.

And you extend it by the current vertex, V sub I.

That is, you add the [INAUDIBLE] vertices weight to the weight of the optimal

solution from two sub problems back. So this is a pattern we're going to see

over and over again. We'll define subproblems for various

computational problems. And we'll use re, recurrence to express

how the optimal solution of a given subproblem depends only on the solutions

to smaller subproblems. So just like in our independent set

example once you have such a recurrence it naturally leads to a table filling

algorithm where each entry in your table corresponds to the optimal solution to

one sub-problem and you use your recurrence to just fill it in moving from

the smaller sub-problems to the larger ones.

So the third property, you probably won't have to worry about much.

Usually this just takes care of itself. But needless to say, after you've done

the work of solving all of your sub problems, you better be able to answer

the original question. This property is usually automatically

satisfied because in most cases, not all, but in most cases the original problem is

simply the biggest of all of your subproblems.

Notice this is exactly how things worked in the independent sets.

Our biggest subproblem G sub N was just the original graph.

So once we fill up the whole table, boom.

Waiting for us in the final entry was the desired solution to the original problem.

So I realize, you know this is a little abstract at the moment.

We've only have one concrete example to relate to these abstract concepts.

I encourage you to revisit this again after we see more examples and we will

see many more examples. Something that all of the forthcoming

examples should make clear is the power and flexibility of a dynamic programming

paradigm. This is really just a technique that you

have got to know. Now when you're trying to devise your own

dynamic programming algorithms, the key, the heart of the matter is to figure out

what the right sub problems are. If you nail the sub problems usually

everything else falls into place in a fairly formulaic way.

Now if you've got a black belt in dynamic programming you might be able to just

stare at a problem. And intuitively know what the right

collection of subproblems are. And then, boom, you're off to the races.

But, of course, you know? For white belts in dynamic programming,

there's still a lot of training to be done.

So rather, in the forthcoming examples. Rather than just plucking the subproblems

from the sky. We're going to go through the same kind

of process that we did for independent sets.

And try to figure out how you would ever come up with these subproblems in the

first place? By reasoning about the structure of

optimal solutions. That's a process you should be able to

mimic in your own attempts at applying this paradigm to problems that come up in

your own projects. So, perhaps you were hoping that once you

saw the ingredients of dynamic programming, all would become clearer why

on earth it's called dynamic programming and probably it's not.

So, this is an anachronistic use of the word

programming. It doesn't mean coding in the way I'm

sure almost all of you think of it. It's the same anachronism in phrases like

mathematical or linear programming. It more refers to a planning process, but

you know for the full story let's go ahead and turn to Richard Bellman

himself. He's more or less the inventor of dynamic

programming, you will see his Bellman-Ford Algorithm a little bit later

in the course. So he answers this question in his

autobiography and he's says, he talks about when he invented it in the 1950's

and he says those were not good years for mathematical research.

He was working at a place called Rand, he says we had a very interesting gentleman

in Washington named Wilson who was the Secretary of Defense.

And he actually had a pathological fear and hatred of the word research.

I'm not using the term lightly. I'm using it precisely.

His face would suffuse, he would turn red, and he would get violent if people

used the term research in his presence. You can imagine how we felt then about

the term mathematical. So the Rand Corporation was employed by

the Air Force, and the Air Force had Wilson as its boss, essentially.

Hence, I felt I had to do something to shield Wilson and the Air Force from the

fact that I was really doing mathematics inside the Rand Corporation.

What title, what name could I choose? In the first place I was interested in

planning and decision making, but planning, it's not a good word for

various reasons. I decided therefore to use the word,

Programming. Dynamic has a very interesting property

as an adjective, in that it's poss, impossible to use the word dynamic in a

pejorative sense. Try thinking of some combination that

will possibly give it a pejorative meaning.

It's impossible. Thus I thought dynamic programming was a

good name. It was something not even a congressman

could object to so I used it as an umbrella for my activities.

Coursera provides universal access to the world’s best education,
partnering with top universities and organizations to offer courses online.