Length
June 29, 2018
Recently, at a beginning programmer’s forum, I saw a user asking about the function length that finds the length of a list. He gave a simple version of length, then used it find the lengths of two lists and determine which was shorter. He correctly realized that calculating the full lengths of both lists is inefficient if one list is much longer than the other, and he asked if there was a way to run through the lists simultaneously, stopping as soon as it is known which list is shorter.
Your task is to write length and a function to determine which of two lists is shorter; your solution must use recursion. When you are finished, you are welcome to read or run a suggested solution, or to post your own solution or discuss the exercise in the comments below.
Next Identical Popcount, Revisited
June 26, 2018
In a comment on the prior exercise, Richard A. O’Keefe said:
This is just the “next k-subset of 1..n” problem. It is possible to go fairly directly from one popc-k integer to the next. Let the least significant 1 bit be at offset p. If bit(p+1) is 0, set it to 1 and clear bit(p). If bit(p+1) is 1, clear bit(p), set bit(0), advance bits (p+1..msb) to the next integer with k-1 bits.
The algorithm at the top of the page is spectacularly inefficient. Consider the case of 32-bit integers, bit(30) is on, and all the others are off. (k=1) The (next n) algorithm loops about 2**30 times.
Dr. O’Keefe is correct. My solution isn’t very good. In fact, it’s abominable.
Your task is to write a program to solve the next-identical-popcount problem in the manner Professor O’Keefe suggests. When you are finished, you are welcome to read or run a suggested solution, or to post your own solution or discuss the exercise in the comments below.
Next Identical Popcount
June 22, 2018
Today’s exercise is an interview question:
The binary weight of a positive integer is the number of 1-bits in its binary representation. For example, the decimal number 7, which is 111 in binary, has a binary weight of 3. The smallest integer greater than 7 that has the same binary weight as 7 is 11, which is 1011 in binary.
Your task is to write a program that, given a positive integer, finds the next larger integer that has the same binary weight. When you are finished, you are welcome to read or run a suggested solution, or to post your own solution or discuss the exercise in the comments below.
Fibonacci Hash
June 19, 2018
Hash tables are ubiquitous in computing, and we’ve both used and implemented hash tables on several occasions; there is even a hash table implementation in the Standard Prelude. Most hash tables nowadays use the elements of an array as the heads of linked lists that store key/value pairs, so to access (store or fetch) a particular element the procedure is to calculate the hash value based on the key, convert the hash value to an index into the array, and scan down the appropriate linked list until you find the desired item.
The usual way to convert the hash value to an array index is to choose the array size prime, then take the hash value modulo the prime. That can be slow, because the modulo operation requires a division, which is always slow compared to other primitive arithmetic operations. Fibonacci hash replaces the division with multiplication and a shift, which often takes less than half the time.
Fibonacci hash works like this: Choose b the number of bits in an integer, usually either 32 or 64. Then compute 2b ÷ φ, where φ = (1 + √5) ÷ 2 ≈ 1.618; for b = 32 this quantity is 2654435769 and for b = 64 this quantity is 11400714819323198485. Then to compute the array index, multiply the hash value by the quantity shown above, allowing it to wrap around if the product exceeds the integer bounds, and shift right, leaving as many bits as desired. The array size should be a power of 2. Here is an implementation in C:
unsigned long long fibhash(unsigned long long h)
{
return (h * 11400714819323198485ull) >> 54
}
This returns a ten-bit value (since the 64 bit product is shifted right 54 bits) from 0 inclusive to 1024 exclusive, which is used as the index to the array. If you need more bits, just change the size of the shift.
Malte Skarupke describes the theory behind Fibonacci hash, which you should look at, because it includes some neat math; be sure not to miss the linked video from Vi Hart.
Your task is to implement Fibonacci hash in the language of your choice. When you are finished, you are welcome to read or run a suggested solution, or to post your own solution or discuss the exercise in the comments below.
Uncouple
June 15, 2018
Today’s exercise is from a programming textbook:
A couple is two adjacent identical items in a sequence. You are to remove all couples, then process the list recursively to remove any additional couples formed by the removal of the original couples. For instance, given the list {red blue green green blue red yellow}, first remove the green couple, leaving {red blue blue red yellow}, then remove the blue couple, leaving {red red yellow}, and finally remove the red couple, leaving {yellow}.
Your task is to write a program to uncouple a list. When you are finished, you are welcome to read or run a suggested solution, or to post your own solution or discuss the exercise in the comments below.
Multiples Of 5
June 12, 2018
We have today an interview question from Amazon:
Write a program to determine if an integer is a multiple of 5 in O(log n) time complexity. You cannot use the division or modulus operators.
Your task is to solve the Amazon interview question. When you are finished, you are welcome to read or run a suggested solution, or to post your own solution or discuss the exercise in the comments below.
Overlap
June 8, 2018
We have today a simple exercise to spice up your Friday lunch hour:
A range of integers is specified by its endpoints; for instance, the range [19, 25] includes the values 19, 20, 21, 22, 23, 24, and 25. The overlap of two ranges is those values that appear in both; for instance, given the ranges [19, 25] and [22, 30], the overlap is the range [22, 25]. The ranges [19, 25] and [12, 17] have no overlap.
Your task is to write a program that takes the endpoints of two ranges and returns their overlap, or reports that they have no overlap. When you are finished, you are welcome to read or run a suggested solution, or to post your own solution or discuss the exercise in the comments below.
Estimating Pi
June 5, 2018
There’s a nice piece of math (it was either on NumberPhile or 3brown1blue, but I can’t find it again, and in any event it is a well-known equality) that says if you pick two positive integers at random, the odds of them having no common divisors are 6 ÷ π² ≈ 0.61. Let’s test that.
Your task is to estimate π using the formula shown above. When you are finished, you are welcome to read or run a suggested solution, or to post your own solution or discuss the exercise in the comments below.
Exercise 7
June 1, 2018
This must be somebody’s homework:
You are given an input file containing lines with three pipe-separated fields; the first field is a student number (a positive integer), the second field is a class name (a string), and the third field is the grade the student received for the class (a non-negative integer, no greater than 100):
22|Data Structures|45 23|English|52 22|English|51 26|Data Structures|72 23|Data Structures|61 21|English|81You are to output a list of class names along with the grade earned by the lowest-numbers student in each class. For instance, given the above data, the output for Data Structures is 45 corresponding to student 22 (with student number lower than 23 or 26, who also took Data Structures) and the output for English is 81 corresponding to student 21 (with student number lower than 22 or 23, who also took English).
Your task is to write a program to produce the requested output. When you are finished, you are welcome to read or run a suggested solution, or to post your own solution or discuss the exercise in the comments below.
Programming Praxis 