Operations on bits when doing binary long division - algorithm

This is from the number theory chapter in CLRS.
We are asked to prove that binary "paper and pencil" long division a/b with result q and reminder r does O((1+lgq)lgb) operations on bits.
The way I see it is we do 1 subtraction of b for each bit in q. So assuming that subtracting b does lgb operations (one for each bit in b), then we have a total of O(lgblgq) operations, which is not what is requested.
If you take into account that the first subtraction operation you do might result in a 0 bit (for example, dividing 100b by 11b), then, OK, you can add 1 to lgq to compensate for this subtraction. But... the same could be said of the subtraction itself - it can take lgb operations or it can take lg(b)+1 operations depending on the numbers (in the 100b and 11b example, the second subtraction will be 100b-11b which takes 3 operations to complete).
So if we're factoring these cases, then the number of operations should be O((1+lgb)(1+lgq)).
So the question is, how can you show that the division takes O((1+lgq)lgb) operations?

When you subtract 100b-11b, you can actually ignore the leading bit in the first number, because you already know that the corresponding bit in the result will be 0. Had it been 1, you would have done a subtraction instead of a shift in the previous step. So the subtraction always considers exactly lg b bits.

Related

Hashing with the Division Method - Choosing number of slots?

So, in CLRS, there's this quote
A prime not too close to an exact power of 2 is often a good choice for m.
Several Questions...
I understand how a power of 2 will just be the lower order bits of your key...however, say you have keys from a universe of 1 to 1 million, with each key having an equal probability of being any number from universe (which I'm guessing is a common assumption about your universe if given no other data?) then wouldn't taking say the 4 lower order bits result in (2^4) lower order bit patterns that were pretty much equally likely for the keys from 1 to 1 million? How am I thinking about this incorrectly?
Why a prime number? So, if power of 2's aren't a good idea, why is a prime number a better choice as opposed to a composite number close to a power of 2 (Also why should it be close to a power of 2...lol)?
You are trying to find a hash table that works well for typical input data, and typical input data does things that you wouldn't expect from good random number generators. Very often you get formatted or semi-formatted strings which, when converted to numbers, end up as K, K+A, K+2A, K+3A,.... for some integers K and A. If K+xA and K+yA hash to the same number mod m, then (x-y)A must be 0 mod m. If m is prime, this can only happen if A = 0 mod m or if x = y mod m, so one time in m. But if m=pq and A happens to be divisible by p, then you get a collision every time x-y is divisible by q, which is more often since q < m.
I guess close to a power of 2 because it might be convenient for the memory management system to have blocks of memory of the resulting size - I really don't know. If you really care, and if you have the time, you could try different primes with some representative data and see which of them are best in practice.

Fastest algorithm to find factors of 2 and 5

I've read this question: Which is the fastest algorithm to find prime numbers?, but I'd like to do this only for 2 and 5 primes.
For example, the number 42000 is factorized as:
24 • 31 • 53 • 71
I'm only interested in finding the powers of 2 and 5: 4 and 3 in this example.
My naive approach is to successively divide by 2 while the remainder is 0, then successively divide by 5 while the remainder is 0.
The number of successful divisions (with zero remainder) are the powers of 2 and 5.
This involves performing (x + y + 2) divisions, where x is the power of 2 and y is the power of 5.
Is there a faster algorithm to find the powers of 2 and 5?
Following the conversation, I do think your idea is the fastest way to go, with one exception:
Division (in most cases) is expensive. On the other hand, checking the last digit of the number is (usually?) faster, so I would check the last digit (0/5 and 0/2/4/6/8) before dividing.
I am basing this off this comment by the OP:
my library is written in PHP and the number is actually stored as a string in base 10. That's not the most efficient indeed, but this is what worked best within the technical limits of the language.
If you are committed to strings-in-php, then the following pseudo-code will speed things up compared to actual general-purpose repeated modulus and division:
while the string ends in 0, but is not 0
chop a zero off the end,
increment ctr2 and ctr5
switch repeatedly depending on the last digit:
if it is a 5,
divide it by 5
increment ctr5
if it is 2, 4, 6, 8,
divide it by 2
increment ctr2
otherwise
you have finished
This does not require any modulus operations, and you can implement divide-by-5 and divide-by-2 cheaper than a general-purpose long-number division.
On the other hand, if you want performance, using string representations for unlimited-size integers is suicide. Use gmp (which has a php library) for your math, and convert to strings only when necessary.
edit:
you can gain extra efficiency (and simplify your operations) using the following pseudocode:
if the string is zero, terminate early
while the last non-zero character of the string is a '5',
add the string to itself
decrement ctr2
count the '0's at the end of the string into a ctr0
chop off ctr0 zeros from the string
ctr2 += ctr0
ctr5 += ctr0
while the last digit is 2, 4, 6, 8
divide the string by 2
increment ctr2
Chopping many 0s at once is better than looping. And mul2 beats div5 in terms of speed (it can be implemented by adding the number once).
If you have a billion digit number, you do not want to do divisions on it unless it's really necessary. If you don't have reason to believe that it is in the 1/2^1000 numbers divisible by 2^1000, then it makes sense to use much faster tests that only look at the last few digits. You can tell whether a number is divisible by 2 by looking at the last digit, whether it is divisible by 4 by looking at the last 2 digits, and by 2^n by looking at the last n digits. Similarly, you can tell whether a number is divisible by 5 by looking at the last digit, whether it is divisible by 25 by looking at the last 2 digits, and by 5^n by looking at the last n digits.
I suggest that you first count and remove the trailing 0s, then decide from the last digit whether you are testing for powers of 2 (last digit 2,4,6, or 8) or powers of 5 (last digit 5).
If you are testing for powers of 2, then take the last 2, 4, 8, 16, ... 2^i digits, and multiply this by 25, 625, ... 5^2^i, counting the trailing 0s up to 2^i (but not beyond). If you get fewer than 2^i trailing 0s, then stop.
If you are testing for powers of 5, then take the last 2, 4, 8, 16, ... 2^i digits, and multiply this by 4, 16, ... 2^2^i, counting the trailing 0s up to 2^i (but not beyond). If you get fewer than 2^i trailing 0s, then stop.
For example, suppose the number you are analyzing is 283,795,456. Multiply 56 by 25, you get 1400 which has 2 trailing 0s, continue. Multiply 5,456 by 625, you get 3,410,000, which has 4 trailing 0s, continue. Multiply 83,795,456 by 5^8=390,625, you get 32,732,600,000,000, which has 8 trailing 0s, continue. Multiply 283,795,456 by 5^16 to get 43,303,750,000,000,000,000 which has only 13 trailing 0s. That's less than 16, so stop, the power of 2 in the prime factorization is 2^13.
I hope that for larger multiplications you are implementing an n log n algorithm for multiplying n digit numbers, but even if you aren't, this technique should outperform anything involving division on typical large numbers.
Let's look at the average-case time complexity of various algorithms, assuming that each n-digit number is equally likely.
Addition or subtraction of two n-digit numbers takes theta(n) steps.
Dividing an n-digit number by a small number like 5 takes theta(n) steps. Dividing by the base is O(1).
Dividing an n-digit number by another large number takes theta(n log n) steps using the FFT, or theta(n^2) by a naive algorithm. The same is true for multiplication.
The algorithm of repeatedly dividing a base 10 number by 2 has an average case time complexity of theta(n): It takes theta(n) time for the first division, and on average, you need to do only O(1) divisions.
Computing a large power of 2 with at least n digits takes theta(n log n) by repeated squaring, or theta(n^2) with simple multiplication. Performing Euclid's algorithm to compute the GCD takes an average of theta(n) steps. Although divisions take theta(n log n) time, most of the steps can be done as repeated subtractions and it takes only theta(n) time to do those. It takes O(n^2 log log n) to perform Euclid's algorithm this way. Other improvements might bring this down to theta(n^2).
Checking the last digit for divisibility by 2 or 5 before performing a more expensive calculation is good, but it only results in a constant factor improvement. Applying the original algorithm after this still takes theta(n) steps on average.
Checking the last d digits for divisibility by 2^d or 5^d takes O(d^2) time, O(d log d) with the FFT. It is very likely that we only need to do this when d is small. The fraction of n-digit numbers divisible by 2^d is 1/2^d. So, the average time spent on these checks is O(sum(d^2 / 2^d)) and that sum is bounded independent of n, so it takes theta(1) time on average. When you use the last digits to check for divisibility, you usually don't have to do any operations on close to n digits.
depends on whether you're starting with a native binary number or some bigint string -
chopping off very long chains of trailing edge zeros in bigint strings are a lot easier than trying to extract powers of 2 and 5 separately - e.g. 23456789 x 10^66
23456789000000000000000000000000000000000000000000000000000000000000000000
This particular integer, on the surface, is 244-bits in total, requiring a 177-bit-wide mantissa (178-bit precision minus 1-bit implicit) to handle it losslessly, so even newer data types such as uint128 types won't suffice :
11010100011010101100101010010000110000101000100001000110100101
01011011111101001110100110111100001001010000110111110101101101
01001000011001110110010011010100001001101000010000110100000000
0000000000000000000000000000000000000000000000000000000000
The sequential approach is to spend 132 loop cycles in a bigint package to get them out ::
129 63 2932098625
130 64 586419725
131 65 117283945
132 66 23456789
133 2^66 x
5^66 x
23456789
But once you can quickly realize there's a chain of 66 trailing zeros, the bigint package becomes fully optional, since the residual digits is less than 24.5-bits in total width:
2^66
5^66
23456789
I think your algorithm will be the fastest. But I have a couple of suggestions.
One alternative is based on the greatest common divisor. Take the gcd of your input number with the smallest power of 2 greater than your input number; that will give you all the factors of 2. Divide by the gcd, then repeat the same operation with 5; that will give you all the factors of 5. Divide again by the gcd, and the remainder tells you if there are any other factors.
Another alternative is based on binary search. Split the binary representation of your input number in half; if the right half is 0, move left, otherwise move right. Once you have the factors of 2, divide, then apply the same algorithm on the remainder using powers of 5.
I'll leave it to you to implement and time these algorithms. But my gut feeling is that repeated division will be hard to beat.
I just read your comment that your input number is stored in base 10. In that case, divide repeatedly by 10 as long as the remainder is 0; that gives factors of both 2 and 5. Then apply your algorithm on the reduced number.

Efficient way to multiply a large set of small numbers

This question was asked in an interview.
You have an array of small integers. You have to multiply all of them. You need not worry about overflow you have ample support for that. What can you do to speed up the multiplication on your machine?
Would multiple additions be better in this case?
I suggested multiplying using a divide and conquer approach but the interviewer was not impressed. What could be the best possible solution for this?
Here are some thoughts:
Divide-and-Conquer with Multithreading: Split the input apart into n different blocks of size b and recursively multiply all the numbers in each block together. Then, recursively multiply all n / b blocks back together. If you have multiple cores and can run parts of this in parallel, you could save a lot of time overall.
Word-Level Parallelism: Let's suppose that your numbers are all bounded from above by some number U, which happens to be a power of two. Now, suppose that you want to multiply together a, b, c, and d. Start off by computing (4U2a + b) × (4U2c + d) = 16U4ac + 4U2ad + 4U2bc + bd. Now, notice that this expression mod U2 is just bd. (Since bd < U2, we don't need to worry about the mod U2 step messing it up). This means that if we compute this product and take it mod U2, we get back bd. Since U2 is a power of two, this can be done with a bitmask.
Next, notice that
4U2ad + 4U2bc + bd < 4U4 + 4U4 + U2 < 9U4 < 16U4
This means that if we divide the entire expression by 16U4 and round down, we will end up getting back just ad. This division can be done with a bitshift, since 16U4 is a power of two.
Consequently, with one multiplication, you can get back the values of both ac and bd by applying a subsequent bitshift and bitmask. Once you have ac and bd, you can directly multiply them together to get back the value of abcd. Assuming that bitmasks and bitshifts are faster than multiplies, this reduces the number of multiplications necessary by 33% (two instead of three here).
Hope this helps!
Your divide and conquer suggestion was a good start. It just needed more explanation to impress.
With fast multiplication algorithms used to multiply large numbers (big-ints), it is much more efficient to multiply similar sized multiplicands than a series of mismatched sizes.
Here's an example in Clojure
; Define a vector of 100K random integers between 2 and 100 inclusive
(def xs (vec (repeatedly 100000 #(+ 2 (rand-int 99)))))
; Naive multiplication accumulating linearly through the array
(time (def a1 (apply *' xs)))
"Elapsed time: 7116.960557 msecs"
; Simple Divide and conquer algorithm
(defn d-c [v]
(let [m (quot (count v) 2)]
(if (< m 3)
(reduce *' v)
(*' (d-c (subvec v 0 m)) (d-c (subvec v m))))))
; Takes less than 1/10th the time.
(time (def a2 (d-c xs)))
"Elapsed time: 600.934268 msecs"
(= a1 a2) ;=> true (same result)
Note that this improvement does not rely on a set limit for the size of the integers in the array (100 chosen arbitrarily and to demonstrate the next algorithm), but only that they be similar in size. This is a very simple divide an conquer. As the numbers get larger and more expensive to multiply, it would make sense to invest more time in iteratively grouping them by similar size. Here I am relying on random distribution and chance that the sizes will stay similar, but it is still going to be significantly better than the naive approach even for the worst case.
As suggested by Evgeny Kluev in the comments, for a large number of small integers, there is going to be a lot of duplication, so efficient exponentiation is also better than naive multiplication. This depends a lot more on the relative parameters than the divide and conquer, that is the numbers must be sufficiently small relative to the count for enough duplicates to accumulate to bother, but certainly performs well with these parameters (100K numbers in the range 2-100).
; Hopefully an efficient implementation
(defn pow [x n] (.pow (biginteger x) ^Integer n))
; Perform pow on duplications based on frequencies
(defn exp-reduce [v] (reduce *' (map (partial apply pow) (frequencies v))))
(time (def a3 (exp-reduce xs)))
"Elapsed time: 650.211789 msecs"
Note the very simple divide and conquer performed just a wee better in this trial, but would be even relatively better if fewer duplicates were expected.
Of course we can also combine the two:
(defn exp-d-c [v] (d-c (mapv (partial apply pow) (frequencies v))))
(time (def a4 (exp-d-c xs)))
"Elapsed time: 494.394641 msecs"
(= a1 a2 a3 a4) ;=> true (all equal)
Note there are better ways to combine these two since the result of the exponentiation step is going to result in various sizes of multiplicands. The value of added complexity to do so depends on the expected number of distinct numbers in the input. In this case, there are very few distinct numbers so it wouldn't pay to add much complexity.
Note also that both of these are easily parallelized if multiple cores are available.
If many of the small integers occur multiple times, you could start by counting every unique integer. If c(n) is the number of occurrences of integer n, the product can be computed as
P = 2 ^ c(2) * 3 ^ c(3) * 4 ^ c(4) * ...
For the exponentiation steps, you can use exponentiation by squaring which can reduce the number of multiplications considerably.
If the count of numbers really is large compared to the range, then we have seen two asymptotic solutions presented to reduce the complexity considerably. One was based on successive squaring to compute c^k in O(log k) time for each number c, giving O(C mean(log k)) time if the largest number is C and k gives the exponent for each number between 1 and C. The mean(log k) term is maximized if every number appears an equal number of times, so if you have N numbers then the complexity becomes O(C log(N/C)), which is very weakly dependent on N and essentially just O(C) where C specifies the range of numbers.
The other approach we saw was sorting numbers by the number of times they appear, and keeping track of the product of leading numbers (starting with all numbers) and raising this to a power so that the least frequent number is removed from the array, and then updating the exponents on the remaining element in the array and repeating. If all numbers occur the same number of times K, then this gives O(C + log K) which is an improvement over O(C log K). However, say the kth number appears 2^k times. Then this will still give O(C^2 + C log(N/C)) time which is technically worse than the previous method O(C log(N/C)) if C > log(N/C). Thus, if you don't have good information on how evenly distributed the occurrences of each number are, you should go with the first approach, just take the appropriate power of each distinct number that appears in the product by using successive squaring, and take the product of the results. Total time O(C log (N/C)) if there are C distinct numbers and N total numbers.
To answer this question we need to interpret in some way the assumption from OP: need not worry about overflow. In larger part of this answer it is interpreted as "ignore overflows". But I start with several thoughts about other interpretation ("use multiprecision arithmetic"). In this case process of multiplication may be approximately split into 3 stages:
Multiply together small sets of small numbers to get a large set of not-so-small numbers. Some of the ideas from second part of this answer may be used here.
Multiply together these numbers to get a set of large numbers. Either trivial (quadratic time) algorithm or Toom–Cook/Karatsuba (sub-quadratic time) methods may be used.
Multiply together large numbers. Either Fürer's or Schönhage–Strassen algorithm may be used. Which gives O(N polylog N) time complexity for the whole process.
Binary exponentiation may give some (not very significant) speed improvement, because most (if not every) of complex multiplication algorithms mentioned here do squaring faster than multiplication of two unequal numbers. Also we could factorize every "small" number and use binary exponentiation only for prime factors. For evenly distributed "small" numbers this will decrease number of exponentiations by factor log(number_of_values) and slightly improve balance of squarings/multiplications.
Divide and conquer is OK when numbers are evenly distributed. Otherwise (for example when input array is sorted or when binary exponentiation is used) we could do better by placing all multiplicands into priority queue, ordered (may be approximately ordered) by number length. Then we could multiply two shortest numbers and place the result back to the queue (this process is very similar to Huffman encoding). There is no need to use this queue for squaring. Also we should not use it while numbers are not long enough.
More information on this could be found in the answer by A. Webb.
If overflows may be ignored we could multiply the numbers with linear-time or better algorithms.
Sub-linear time algorithm is possible if input array is sorted or input data is presented as set of tuples {value, number of occurrences}. In latter case we could perform binary exponentiation of each value and multiply results together. Time complexity is O(C log(N/C)), where C is number of different values in the array. (See also this answer).
If input array is sorted, we could use binary search to find positions where value changes. This allows to determine how many times each value occurs in the array. Then we could perform binary exponentiation of each value and multiply results together. Time complexity is O(C log N). We could do better by using one-sided binary search here. In this case time complexity is O(C log(N/C)).
But if input array is not sorted, we have to inspect each element, so O(N) time complexity is the best we can do. Still we could use parallelism (multithreading, SIMD, word-level parallelism) to obtain some speed improvement. Here I compare several such approaches.
To compare these approaches I've chosen very small (3-bit) values, which are pretty tightly packed (one value per 8-bit integer). And implemented them in low-level language (C++11) to get easier access to bit manipulation, specific CPU instructions, and SIMD.
Here are all the algorithms:
accumulate from standard library.
Trivial implementation with 4 accumulators.
Word-level parallelism for multiplication, as described in the answer by templatetypedef. With 64-bit word size this approach allows up to 10-bit values (with only 3 multiplications instead of each 4) or it may be applied twice (and I did it in the tests) with up to 5-bit values (requiring only 5 multiplications out of each 8).
Table lookup. In the tests 7 multiplications out of each 8 are substituted by single table lookup. With values larger than in these tests, number of substituted multiplications decreases, slowing down the algorithm. Values larger than 11-12 bits make this approach useless.
Binary exponentiation (see more details below). Values larger than 4 bits make this approach useless.
SIMD (AVX2). This implementation can use up to 8-bit values.
Here are sources for all tests on Ideone. Note that SIMD test requires AVX2 instruction set from Intel. Table lookup test requires BMI2 instruction set. Other tests do not depend on any particular hardware (I hope). I run these tests on 64-bit Linux, compiled with gcc 4.8.1, optimization level -O2.
Here are some more details for binary exponentiation test:
for (size_t i = 0; i < size / 8; i += 2)
{
auto compr = (qwords[i] << 4) | qwords[i + 1];
constexpr uint64_t lsb = 0x1111111111111111;
if ((compr & lsb) != lsb) // if there is at least one even value
{
auto b = reinterpret_cast<uint8_t*>(qwords + i);
acc1 *= accumulate(b, b + 16, acc1, multiplies<unsigned>{});
if (!acc1)
break;
}
else
{
const auto b2 = compr & 0x2222222222222222;
const auto b4 = compr & 0x4444444444444444;
const auto b24 = b4 & (b2 * 2);
const unsigned c7 = __builtin_popcountll(b24);
acc3 += __builtin_popcountll(b2) - c7;
acc5 += __builtin_popcountll(b4) - c7;
acc7 += c7;
}
}
const auto prod4 = acc1 * bexp<3>(acc3) * bexp<5>(acc5) * bexp<7>(acc7);
This code packs values even more densely than in the input array: two values per byte. Low-order bit of each value is handled differently: since we could stop after 32 zero bits is found here (with result "zero"), this case cannot alter performance very much, so it is handled by simplest (library) algorithm.
Out of 4 remaining values, "1" is not interesting, so we need to count only occurrences of "3", "5" , and "7" with bitwise manipulations and "population count" intrinsic.
Here are the results:
source size: 4 Mb: 400 Mb:
1. accumulate: 0.835392 ns 0.849199 ns
2. accum * 4: 0.290373 ns 0.286915 ns
3. 2 mul in 1: 0.178556 ns 0.182606 ns
4. mult table: 0.130707 ns 0.176102 ns
5. binary exp: 0.128484 ns 0.119241 ns
6. AVX2: 0.0607049 ns 0.0683234 ns
Here we can see that accumulate library algorithm is pretty slow: for some reason gcc could not unroll the loop and use 4 independent accumulators.
It is not too difficult to do this optimization "by hand". The result is not particularly fast. But if we allocate 4 threads for this task, CPU would approximately match memory bandwidth (2 channels, DDR3-1600).
Word-level parallelism for multiplications is almost twice as fast. (We'll need only 3 threads to match memory bandwidth).
Table lookup is even faster. But its performance degrades when input array cannot fit in L3 cache. (We'll need 3 threads to match memory bandwidth).
Binary exponentiation has approximately the same speed. But with larger inputs this performance does not degrade, it even slightly improves because exponentiation itself uses less time compared to value counting. (We'll need only 2 threads to match memory bandwidth).
As expected, SIMD is the fastest. Its performance slightly degrades when input array cannot fit in L3 cache. Which means we are close to memory bandwidth with single thread.
I have one solution. Let us discuss it with other solutions.
The key part of question is how to reduce times of multiply. And integers are small but set is big.
My solution:
use an small array to record how many times each number appears.
Remove number 1 from array. You don't need to count it.
Find the number which appears least times n. Then multiply all numbers and get result K. Then count K^n.
Remove this number (For instance, you can switch it with the last number of array and reduce size of array for 1). So next time you won't consider this number any more. At same time, the appearance times of other numbers need to be reduced with the times of removed number.
Once again get the number which appears least times. Do same thing as step 2.
Repeatedly do step 2-4 and complete counting.
Let me use an example to show how many multiply we need to do: Assume
we have 5 numbers [1, 2, 3, 4, 5]. Number 1 appears 100 times, number
2 appears 150 times, number 3 appears 200 times, number 4 appears 300
times, and number 5 appears 400 times.
method 1: multiply it directly or use divide and conquer
we need 100+150+200+300+400-1 = 1149 multiply to get result.
method 2: we do (1^100)(2^150)(3^200)(4^300)(5^400)
(100-1)+(150-1)+(200-1)+(300-1)+(400-1)+4 = 1149.[same as method 1]
Cause n^m will do m-1 multiply in deed. Plus you need time to go through all numbers, though this time is short.
method in this post:
First, you need time to go through all numbers. It can be discarded compare to time of multiply.
The real counting you are making is:
((2*3*4*5)^150)*((3*4*5)^50)*((4*5)^100)*(5^100)
Then you need to do multiply 3+149+2+49+1+99+99+3 = 405 times

Fast division algorithm for binary numbers

I'm currently building a 16 bit ALU using Logisim (ie logic gates only), and am stuck on a division process. I am currently just using the simple standard "division algorithm loop" (as shown below):
Read input values;
Compare input values. Wait until comparison process has finished;
If A<B go to step 10. If A≥B, go to next step;
Subtract B from A;
Wait until subtraction process has finished;
Add one to count;
Wait until counting process has finished;
Write value from subtraction process to input;
Go to step 1;
Answer is count remainder A
This, however, takes a very long time for processes with large answers (repeating a 300 tick cycle 65,000 times isn't fun).
I'm just wondering if there are similar algorithms which are quicker (that exclusively use addition and/or subtraction and/or multiplication and any boolean logic) that could be implemented using logic gates.
Any help or ideas would be greatly appreciated!
Fraser
Use long-division. In binary, there is no multiplication, since the quotient at each bit position can only be 1 or 0. So it can be implemented as a conditional subtract (subtract if result non-negative) and shift.
That's just a crude outline, of course.
A typical approach for a 32/16:16+16 division would be to have the dividend stored in a pair of 16-bit registers (which get updated during operation) and the divisor in its own register (which doesn't). Sixteen times, subtract the upper 17 bits of the dividend from the divisor; if a borrow results, discard the result and shift the divisor left one place, putting a 0 into the lsb. If no borrow results, store the result into the divisor while shifting it left, but put a 1 into the lsb. After sixteen such steps, the lower 16 bits of the dividend register will hold the quotient, and the upper 16 bits will hold the remainder. Note that this operation will only work if the quotient is representable in 16 bits. Note also that on a processor which implements 32/16:16+16 division in this fashion, one may conveniently divide arbitrarily-large numbers by a 16-bit quantity, since the upper 16 bits of dividend for each step should be the remainder from the previous step.

Distinct digit count

Is is possible to count the distinct digits in a number in constant time O(1)?
Suppose n=1519 output should be 3 as there are 3 distinct digits(1,5,9).
I have done it in O(N) time but anyone knows how to find it in O(1) time?
I assume N is the number of digits of n. If the size of n is unlimited, it can't be done in general in O(1) time.
Consider the number n=11111...111, with 2 trillion digits. If I switch one of the digits from a 1 to a 2, there is no way to discover this without in some way looking at every single digit. Thus processing a number with 2 trillion digits must take (of the order of) 2 trillion operations at least, and in general, a number with N digits must take (of the order of) N operations at least.
However, for almost all numbers, the simple O(N) algorithm finishes very quickly because you can just stop as soon as you get to 10 distinct digits. Almost all numbers of sufficient length will have all 10 digits: e.g. the probability of not terminating with the answer '10' after looking at the first 100 digits is about 0.00027, and after the first 1000 digits it's about 1.7e-45. But unfortunately, there are some oddities which make the worst case O(N).
After seeing that someone really posted a serious answer to this question, I'd rather repeat my own cheat here, which is a special case of the answer described by #SimonNickerson:
O(1) is not possible, unless you are on radix 2, because that way, every number other than 0 has both 1 and 0, and thus my "solution" works not only for integers...
EDIT
How about 2^k - 1? Isn't that all 1s?
Drat! True... I should have known that when something seems so easy, it is flawed somehow... If I got the all 0 case covered, I should have covered the all 1 case too.
Luckily this case can be tested quite quickly (if addition and bitwise AND are considered an O(1) operation): if x is the number to be tested, compute y this way: y=(x+1) AND x. If y=0, then x=2^k - 1. because this is the only case when all the bits needed to be flipped by the addition. Of course, this is quite a bit flawed, as with bit lengths exceeding the bus width, the bitwise operators are not O(1) anymore, but rather O(N).
At the same time, I think it can be brought down to O(logN), by breaking the number into bus width size chunks, and AND-ing together the neighboring ones, repeating until only one is left: if there were no 0s in the number tested, the last one will be full 1s too...
EDIT2: I was wrong... This is still O(N).

Resources