Complexity Explorer Santa Fe Institute

Introduction to Information Theory

Lead instructor: Seth Lloyd

Your progress is not being saved! Enroll now or log in to track your progress or submit homework.
Please return to the Courses page to find current offerings.

1.3 Using Bits to Count and Label » Quiz 1 Solutions

Question 1:

Remember that 1 bit can by itself represent 2 states (which can be labelled as 0 and 1).

2 bits can represent 4 states, corresponding to the pairs (0, 0), (0, 1), (1, 0), and (1, 1).

We express this more conveniently by using exponentiation. The total number of states that a sequence of n bits can represent is 2n , i.e., the number of states of a single bit (2) raised to the number of bits (n).

It follows that the number of states that 3 bits can represent is =8. Similarly, the number of states that 5 bits can represent is = 32.
 

Question 2:

We know that a single bit can represent two states, two bits can represent four states, and three bits can represent 8 states. Thus, the answer to question 2 is 3. (see the answer to Question 1 for more details)

More generally, the number of states that n bits can represent is 2n. Given any number of states S, the number of bits needed to represent S states is specified by the number n which solves 2n = S. We can use the “logarithm” function, which is the opposite of exponentiation, to find the n that correspond to a given S.

The logarithm of some number S in “base” b is defined to be that number n which satisfies bn = S, and can be written as n = logb S.  Thus, the logarithm in base 2 of 8 is 3, which we write as log2 8 = 3, the logarithm of 32 in base 2 is 5, which we write as log2 32 = 5, and so on.
 

Question 3:

The correct answer is 300.  There are two ways to arrive at this answer.

First, note that the number of bits necessary to represent some number of states S is log2 S.  By using a digital calculator, we can compute that log2 1090 ≈ 298 ≈ 300 (this assumes the calculator can handle very large numbers like 1090).

There is another way to arrive at this answer, however. Remember the exponentiation rule that (ca)b = ca*b.  Then, note that we can write 1090 = (103)30. Using the fact that 103 ≈ 210, we can combine to write 1090 = (103)30 ≈ (210)30 = 210*30 = 2300. Finally, we know that 2300 states can be represented by 300 bits.

 

Question 4:

As stated, a single nucleotide can represent 4 states, written A, C, G, or T. Two nucleotides can represent 16 states: (A,A), (A,C), (A,G), (A,T), (C,A), (C,C), (C,G), (C,T), (G,A), (G,C), (G,G), (G,T), (T,A), (T,C), (T,G), and (A,T).  Using the same reasoning as in the answer to Question 1, the number of sequences that can be represented by n nucleotides is 4n, i.e., the number of states of a single nucleotide (4) raised to the number of nucleotides (n).

Since 3 billion is equal to 3,000,000,000, the number of sequences that can be represented by a genome with 3 billion nucleotides is 43,000,000,000.