Here is an algorithm to output a subset of activities S such that no activities in S overlap and profit(S) is maximum.

Define $p[i]$ to be the largest index $j$ such that $a_j$ does not overlap $a_i$.

$\#$ Return maximum possible profit from $a_1,...,a_n$

def rec_opt(n):
    if n == 0:
        return 0
        return max(rec_opt(n-1), w_n + rec_opt(p[n]))

It doesn't mention what $w_n$ is in the slide, but I think it is the profit of the activity $a_n$.

Consider a set of $n$ activities where there is no overlap at all. Then $T(n) = 2T(n − 1) + c$. This recurrence has an exponential closed-form, $O(2^n)$.

  • "return max(rec_opt(n-1), w_n + rec_opt(p[n]))": I think this line takes time $2T(n-1)$.

  • "if n == 0": this line takes time $c$.

Am I correct?

One more question, how can we get $O(2^n)$ from $T(n) = 2T(n − 1) + c$?

  • 1
    $\begingroup$ Please ask only one question per post. $\endgroup$ – Yuval Filmus Dec 16 '16 at 9:34
  • $\begingroup$ I think they are related, so I put them together. $\endgroup$ – yashirq Dec 16 '16 at 9:43

You are asking several questions. I will only answer the second one.

Suppose that $T(n) = 2T(n-1) + c$, and that $T(0) = a$. Define $S(n) = T(n) + c$. Then $S(n) = T(n)+c = 2T(n-1)+2c = 2S(n-1)$, and so $S(n) = 2^nS(0) = 2^n(a+c)$. We conclude that $T(n) = 2^n(a+c)-c = 2^na+(2^n-1)c$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.