Measuring complexity for powering a number - complexity-theory

I implemented a program for powering a number (a^n) using the divide and conquer technique. i implemented two versions of the same problem:
Version 1:
def input_params():
a=input('Input \'a\' & \'n\' for a^n:')
n=input('')
result=power(a,n)
print (result)
def power(a,n):
if n<=1:
return a
elif n%2==0:
return pow(power(a,n/2),2)
else:
return pow(power(a,(n-1)/2),2)*a
if __name__ == "__main__":
input_params()
Version 2:
def input_params():
a=input('Input \'a\' & \'n\' for a^n:')
n=input('')
result=power(a,n)
print (result)
def power(a,n):
if n<=1:
return a
elif n%2==0:
return power(a,n/2)*power(a,n/2)
else:
return power(a,(n-1)/2)*power(a,(n-1)/2)*a
if __name__ == "__main__":
input_params()
Version 3:
def input_params():
a=input('Input \'a\' & \'n\' for a^n:')
n=input('')
result=power(a,n)
print (result)
def power(a,n):
if n<=1:
return a
elif n%2==0:
return square(power(a,n/2))
else:
return square(power(a,(n-1)/2))*a
def square(num):
return num*num
if __name__ == "__main__":
input_params()
Q1: Which one of the above versions have a complexity of θ(lg n)?
Q2: If version 2 has the complexity of θ(lg n), why? Because although the problem size in Version 2 is being divided by two, the number of sub problems are also two, so I feel Version 2 will grow in the order of θ(nlg n) .
I am not sure about my conclusion.
Thanks

In version 1, there's no answer as you use a function called pow which you never define, so one can't really say what the complexity is.
In version 2, there are redundant calculations, so the answer depends on whether you consider those redundant calculations as being part of your complexity or not (as they can easily be left out).
Try writing version 3 in terms of a function called square and include a definition for square
In all cases you need some assumptions about the cost of the basic operations (*, /, and +) you use. You probably want to assume they all cost O(1), but you should make that clear in your analysis.

Related

How to fix the recursive codeso that it has the same asymptotic running time as the iterative version (but is still recursive)

I was struggling to figure out what changes need to be made to the recursive version of my code below so that the asymptotic run time just about matches the iterative version. Of course, the recursive version of the code still needs to be recursive, but I was stuck on how I should approach cutting down the run time on the recursion.
def binarySearch(alist, item): //ITERATIVE VERSION
first = 0
last = len(alist)-1
found = False
while first<=last and not found:
midpoint = (first + last)/2
if alist[midpoint] == item:
found = True
else:
if item < alist[midpoint]:
last = midpoint-1
else:
first = midpoint+1
return found
def binarySearch(alist, item): //RECURSIVE VERSION
if len(alist) == 0:
return False
else:
midpoint = len(alist)/2
if alist[midpoint]==item:
return True
else:
if item<alist[midpoint]:
return binarySearch(alist[:midpoint],item)
else:
return binarySearch(alist[midpoint+1:],item)
Tried to replicate my function recursively, but the asymptotic running time was much slower than the iterative version.
Your recursive version creates new lists -- which is detrimental to both time and space complexity.
So instead of slicing alist, use first and last like you did in the iterative version:
def binarySearch(alist, item, first=0, last=None):
if last is None:
last = len(alist) - 1
if last < first:
return False
else:
midpoint = (start + end) // 2
if alist[midpoint] == item:
return True
elif item < alist[midpoint]:
return binarySearch(alist, item, first, midpoint - 1)
else:
return binarySearch(alist, item, midpoint + 1, last)

Ruby - Is there a recursive solution to find the divisors of a number?

I am doing a ruby problem that wants a method to find all divisors of a number except itself with the output being a sorted array. If the number is prime, list that it is prime.
I am currently trying to teach myself recursion. Simple recursive problems like finding the factorial of a number is pretty basic to understand but I wanted to know if this particular problem could be done recursively. It seems it fits the criteria of one that could but I could not figure it out.
Example n = 15, divisors besides itself are [3,5].
My code that solved the problem.
require 'prime'
def divisors(n)
return "#{n} is prime" if Prime.prime?(n)
x = n/2
arr = []
until x == 1
arr << x if n % x == 0
x -= 1
end
arr.sort
end
Any help doing this recursively would be great or just letting me know it's not a problem that can be done this way would be helpful too.
def divisors(n, x=nil)
return "#{n} is prime" if Prime.prime?(n)
x ||= n/2
arr = []
return arr if x == 1
if n % x == 0
arr << x
end
(arr.concat divisors(n, x - 1)).sort
end
The function is refactored to handle three things:
the initial call (x ||= /2)
base cases (early returns)
iteration logic done through recursion.
An important thing is that the variable which changes during the iteration (x) is placed as a parameter for the method (with a default value, so it can essentially be used as a private parameter)
By the way, I personally found learning Elixir very helpful in understanding recursion. With pattern matching and multiple functional clauses, the initial call, base case, and iteration can be split into their own methods.

algorithm - Searching for an input for which a function returns a specific value

I have the following function:
F(0) = 0
F(1) = 1
F(2) = 2
F(2*n) = F(n) + F(n+1) + n , n > 1
F(2*n+1) = F(n-1) + F(n) + 1, n >= 1
I am given a number n < 10^25 and I have to show that exists a value a such as F(a)=n. Because of how the function is defined, there might exist a n such as F(a)=F(b)=n where a < b and in this situation I must return b and not a
What I have so far is:
We can split this function into two strict monotone series, one for F(2*n) and one for F(2*n+1) and can find the specified value in logarithmic time, so the finding is more or less done.
I've also found that F(2*n) >= F(2*n+1) for any n, so I first search for it in F(2*n) and if I don't find it there, I search in F(2*n+1)
The problem is calculating the function value. Even with some crazy memoization up to 10^7 and then falling back to recursion, it still couldn't calculate values above 10^12 in a reasonable time.
I think I have the algorithm for actually finding what I need all figured out, but I can't calculate F(n) fast enough.
Simply use memoisation all the way up to the target value, e.g. in Python:
class Memoize:
def __init__(self, fn):
self.fn = fn
self.memo = {}
def __call__(self, *args):
if not self.memo.has_key(args):
self.memo[args] = self.fn(*args)
return self.memo[args]
#Memoize
def R(n):
if n<=1: return 1
if n==2: return 2
n,rem = divmod(n,2)
if rem:
return R(n)+R(n-1)+1
return R(n)+R(n+1)+n
This computes the answer for 10**25 instantly.
The reason this works is because the nature of the recursion means that for a binary number abcdef it will only need to at most use the values:
abcdef
abcde-1,abcde,abcde+1
abcd-2,abcd-1,abcd,abcd+1,abcd+2
abc-2,abc-1,abc,abc+1,abc+2
ab-2,ab-1,ab,ab+1,ab+2
a-2,a-1,a,a+1,a+2
At each step you can move up or down 1, but you also divide the number by 2 so the most you can move away from the original number is limited.
Therefore the memoised code will only use at most 5*log_2(n) evaluations.

Checking if two strings are permutations of each other in Python

I'm checking if two strings a and b are permutations of each other, and I'm wondering what the ideal way to do this is in Python. From the Zen of Python, "There should be one -- and preferably only one -- obvious way to do it," but I see there are at least two ways:
sorted(a) == sorted(b)
and
all(a.count(char) == b.count(char) for char in a)
but the first one is slower when (for example) the first char of a is nowhere in b, and the second is slower when they are actually permutations.
Is there any better (either in the sense of more Pythonic, or in the sense of faster on average) way to do it? Or should I just choose from these two depending on which situation I expect to be most common?
Here is a way which is O(n), asymptotically better than the two ways you suggest.
import collections
def same_permutation(a, b):
d = collections.defaultdict(int)
for x in a:
d[x] += 1
for x in b:
d[x] -= 1
return not any(d.itervalues())
## same_permutation([1,2,3],[2,3,1])
#. True
## same_permutation([1,2,3],[2,3,1,1])
#. False
"but the first one is slower when (for example) the first char of a is nowhere in b".
This kind of degenerate-case performance analysis is not a good idea. It's a rat-hole of lost time thinking up all kinds of obscure special cases.
Only do the O-style "overall" analysis.
Overall, the sorts are O( n log( n ) ).
The a.count(char) for char in a solution is O( n 2 ). Each count pass is a full examination of the string.
If some obscure special case happens to be faster -- or slower, that's possibly interesting. But it only matters when you know the frequency of your obscure special cases. When analyzing sort algorithms, it's important to note that a fair number of sorts involve data that's already in the proper order (either by luck or by a clever design), so sort performance on pre-sorted data matters.
In your obscure special case ("the first char of a is nowhere in b") is this frequent enough to matter? If it's just a special case you thought of, set it aside. If it's a fact about your data, then consider it.
heuristically you're probably better to split them off based on string size.
Pseudocode:
returnvalue = false
if len(a) == len(b)
if len(a) < threshold
returnvalue = (sorted(a) == sorted(b))
else
returnvalue = naminsmethod(a, b)
return returnvalue
If performance is critical, and string size can be large or small then this is what I'd do.
It's pretty common to split things like this based on input size or type. Algorithms have different strengths or weaknesses and it would be foolish to use one where another would be better... In this case Namin's method is O(n), but has a larger constant factor than the O(n log n) sorted method.
I think the first one is the "obvious" way. It is shorter, clearer, and likely to be faster in many cases because Python's built-in sort is highly optimized.
Your second example won't actually work:
all(a.count(char) == b.count(char) for char in a)
will only work if b does not contain extra characters not in a. It also does duplicate work if the characters in string a repeat.
If you want to know whether two strings are permutations of the same unique characters, just do:
set(a) == set(b)
To correct your second example:
all(str1.count(char) == str2.count(char) for char in set(a) | set(b))
set() objects overload the bitwise OR operator so that it will evaluate to the union of both sets. This will make sure that you will loop over all the characters of both strings once for each character only.
That said, the sorted() method is much simpler and more intuitive, and would be what I would use.
Here are some timed executions on very small strings, using two different methods:
1. sorting
2. counting (specifically the original method by #namin).
a, b, c = 'confused', 'unfocused', 'foncused'
sort_method = lambda x,y: sorted(x) == sorted(y)
def count_method(a, b):
d = {}
for x in a:
d[x] = d.get(x, 0) + 1
for x in b:
d[x] = d.get(x, 0) - 1
for v in d.itervalues():
if v != 0:
return False
return True
Average run times of the 2 methods over 100,000 loops are:
non-match (string a and b)
$ python -m timeit -s 'import temp' 'temp.sort_method(temp.a, temp.b)'
100000 loops, best of 3: 9.72 usec per loop
$ python -m timeit -s 'import temp' 'temp.count_method(temp.a, temp.b)'
10000 loops, best of 3: 28.1 usec per loop
match (string a and c)
$ python -m timeit -s 'import temp' 'temp.sort_method(temp.a, temp.c)'
100000 loops, best of 3: 9.47 usec per loop
$ python -m timeit -s 'import temp' 'temp.count_method(temp.a, temp.c)'
100000 loops, best of 3: 24.6 usec per loop
Keep in mind that the strings used are very small. The time complexity of the methods are different, so you'll see different results with very large strings. Choose according to your data, you may even use a combination of the two.
Sorry that my code is not in Python, I have never used it, but I am sure this can be easily translated into python. I believe this is faster than all the other examples already posted. It is also O(n), but stops as soon as possible:
public boolean isPermutation(String a, String b) {
if (a.length() != b.length()) {
return false;
}
int[] charCount = new int[256];
for (int i = 0; i < a.length(); ++i) {
++charCount[a.charAt(i)];
}
for (int i = 0; i < b.length(); ++i) {
if (--charCount[b.charAt(i)] < 0) {
return false;
}
}
return true;
}
First I don't use a dictionary but an array of size 256 for all the characters. Accessing the index should be much faster. Then when the second string is iterated, I immediately return false when the count gets below 0. When the second loop has finished, you can be sure that the strings are a permutation, because the strings have equal length and no character was used more often in b compared to a.
Here's martinus code in python. It only works for ascii strings:
def is_permutation(a, b):
if len(a) != len(b):
return False
char_count = [0] * 256
for c in a:
char_count[ord(c)] += 1
for c in b:
char_count[ord(c)] -= 1
if char_count[ord(c)] < 0:
return False
return True
I did a pretty thorough comparison in Java with all words in a book I had. The counting method beats the sorting method in every way. The results:
Testing against 9227 words.
Permutation testing by sorting ... done. 18.582 s
Permutation testing by counting ... done. 14.949 s
If anyone wants the algorithm and test data set, comment away.
First, for solving such problems, e.g. whether String 1 and String 2 are exactly the same or not, easily, you can use an "if" since it is O(1).
Second, it is important to consider that whether they are only numerical values or they can be also words in the string. If the latter one is true (words and numerical values are in the string at the same time), your first solution will not work. You can enhance it by using "ord()" function to make it ASCII numerical value. However, in the end, you are using sort; therefore, in the worst case your time complexity will be O(NlogN). This time complexity is not bad. But, you can do better. You can make it O(N).
My "suggestion" is using Array(list) and set at the same time. Note that finding a value in Array needs iteration so it's time complexity is O(N), but searching a value in set (which I guess it is implemented with HashTable in Python, I'm not sure) has O(1) time complexity:
def Permutation2(Str1, Str2):
ArrStr1 = list(Str1) #convert Str1 to array
SetStr2 = set(Str2) #convert Str2 to set
ArrExtra = []
if len(Str1) != len(Str2): #check their length
return False
elif Str1 == Str2: #check their values
return True
for x in xrange(len(ArrStr1)):
ArrExtra.append(ArrStr1[x])
for x in xrange(len(ArrExtra)): #of course len(ArrExtra) == len(ArrStr1) ==len(ArrStr2)
if ArrExtra[x] in SetStr2: #checking in set is O(1)
continue
else:
return False
return True
Go with the first one - it's much more straightforward and easier to understand. If you're actually dealing with incredibly large strings and performance is a real issue, then don't use Python, use something like C.
As far as the Zen of Python is concerned, that there should only be one obvious way to do things refers to small, simple things. Obviously for any sufficiently complicated task, there will always be zillions of small variations on ways to do it.
In Python 3.1/2.7 you can just use collections.Counter(a) == collections.Counter(b).
But sorted(a) == sorted(b) is still the most obvious IMHO. You are talking about permutations - changing order - so sorting is the obvious operation to erase that difference.
This is derived from #patros' answer.
from collections import Counter
def is_anagram(a, b, threshold=1000000):
"""Returns true if one sequence is a permutation of the other.
Ignores whitespace and character case.
Compares sorted sequences if the length is below the threshold,
otherwise compares dictionaries that contain the frequency of the
elements.
"""
a, b = a.strip().lower(), b.strip().lower()
length_a, length_b = len(a), len(b)
if length_a != length_b:
return False
if length_a < threshold:
return sorted(a) == sorted(b)
return Counter(a) == Counter(b) # Or use #namin's method if you don't want to create two dictionaries and don't mind the extra typing.
This is an O(n) solution in Python using hashing with dictionaries. Notice that I don't use default dictionaries because the code can stop this way if we determine the two strings are not permutations after checking the second letter for instance.
def if_two_words_are_permutations(s1, s2):
if len(s1) != len(s2):
return False
dic = {}
for ch in s1:
if ch in dic.keys():
dic[ch] += 1
else:
dic[ch] = 1
for ch in s2:
if not ch in dic.keys():
return False
elif dic[ch] == 0:
return False
else:
dic[ch] -= 1
return True
This is a PHP function I wrote about a week ago which checks if two words are anagrams. How would this compare (if implemented the same in python) to the other methods suggested? Comments?
public function is_anagram($word1, $word2) {
$letters1 = str_split($word1);
$letters2 = str_split($word2);
if (count($letters1) == count($letters2)) {
foreach ($letters1 as $letter) {
$index = array_search($letter, $letters2);
if ($index !== false) {
unset($letters2[$index]);
}
else { return false; }
}
return true;
}
return false;
}
Here's a literal translation to Python of the PHP version (by JFS):
def is_anagram(word1, word2):
letters2 = list(word2)
if len(word1) == len(word2):
for letter in word1:
try:
del letters2[letters2.index(letter)]
except ValueError:
return False
return True
return False
Comments:
1. The algorithm is O(N**2). Compare it to #namin's version (it is O(N)).
2. The multiple returns in the function look horrible.
This version is faster than any examples presented so far except it is 20% slower than sorted(x) == sorted(y) for short strings. It depends on use cases but generally 20% performance gain is insufficient to justify a complication of the code by using different version for short and long strings (as in #patros's answer).
It doesn't use len so it accepts any iterable therefore it works even for data that do not fit in memory e.g., given two big text files with many repeated lines it answers whether the files have the same lines (lines can be in any order).
def isanagram(iterable1, iterable2):
d = {}
get = d.get
for c in iterable1:
d[c] = get(c, 0) + 1
try:
for c in iterable2:
d[c] -= 1
return not any(d.itervalues())
except KeyError:
return False
It is unclear why this version is faster then defaultdict (#namin's) one for large iterable1 (tested on 25MB thesaurus).
If we replace get in the loop by try: ... except KeyError then it performs 2 times slower for short strings i.e. when there are few duplicates.
In Swift (or another languages implementation), you could look at the encoded values ( in this case Unicode) and see if they match.
Something like:
let string1EncodedValues = "Hello".unicodeScalars.map() {
//each encoded value
$0
//Now add the values
}.reduce(0){ total, value in
total + value.value
}
let string2EncodedValues = "oellH".unicodeScalars.map() {
$0
}.reduce(0) { total, value in
total + value.value
}
let equalStrings = string1EncodedValues == string2EncodedValues ? true : false
You will need to handle spaces and cases as needed.
def matchPermutation(s1, s2):
a = []
b = []
if len(s1) != len(s2):
print 'length should be the same'
return
for i in range(len(s1)):
a.append(s1[i])
for i in range(len(s2)):
b.append(s2[i])
if set(a) == set(b):
print 'Permutation of each other'
else:
print 'Not a permutation of each other'
return
#matchPermutaion('rav', 'var') #returns True
matchPermutaion('rav', 'abc') #returns False
Checking if two strings are permutations of each other in Python
# First method
def permutation(s1,s2):
if len(s1) != len(s2):return False;
return ' '.join(sorted(s1)) == ' '.join(sorted(s2))
# second method
def permutation1(s1,s2):
if len(s1) != len(s2):return False;
array = [0]*128;
for c in s1:
array[ord(c)] +=1
for c in s2:
array[ord(c)] -=1
if (array[ord(c)]) < 0:
return False
return True
How about something like this. Pretty straight-forward and readable. This is for strings since the as per the OP.
Given that the complexity of sorted() is O(n log n).
def checkPermutation(a,b):
# input: strings a and b
# return: boolean true if a is Permutation of b
if len(a) != len(b):
return False
else:
s_a = ''.join(sorted(a))
s_b = ''.join(sorted(b))
if s_a == s_b:
return True
else:
return False
# test inputs
a = 'sRF7w0qbGp4fdgEyNlscUFyouETaPHAiQ2WIxzohiafEGJLw03N8ALvqMw6reLN1kHRjDeDausQBEuIWkIBfqUtsaZcPGoqAIkLlugTxjxLhkRvq5d6i55l4oBH1QoaMXHIZC5nA0K5KPBD9uIwa789sP0ZKV4X6'
b = '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'
print(checkPermutation(a, b)) #optional
def permute(str1,str2):
if sorted(str1) == sorted(str2):
return True
else:
return False
str1="hello"
str2='olehl'
a=permute(str1,str2)
print(a
from collections import defaultdict
def permutation(s1,s2):
h = defaultdict(int)
for ch in s1:
h[ch]+=1
for ch in s2:
h[ch]-=1
for key in h.keys():
if h[key]!=0 or len(s1)!= len(s2):
return False
return True
print(permutation("tictac","tactic"))

Can I reduce the computational complexity of this?

Well, I have this bit of code that is slowing down the program hugely because it is linear complexity but called a lot of times making the program quadratic complexity. If possible I would like to reduce its computational complexity but otherwise I'll just optimize it where I can. So far I have reduced down to:
def table(n):
a = 1
while 2*a <= n:
if (-a*a)%n == 1: return a
a += 1
Anyone see anything I've missed? Thanks!
EDIT: I forgot to mention: n is always a prime number.
EDIT 2: Here is my new improved program (thank's for all the contributions!):
def table(n):
if n == 2: return 1
if n%4 != 1: return
a1 = n-1
for a in range(1, n//2+1):
if (a*a)%n == a1: return a
EDIT 3: And testing it out in its real context it is much faster! Well this question appears solved but there are many useful answers. I should also say that as well as those above optimizations, I have memoized the function using Python dictionaries...
Ignoring the algorithm for a moment (yes, I know, bad idea), the running time of this can be decreased hugely just by switching from while to for.
for a in range(1, n / 2 + 1)
(Hope this doesn't have an off-by-one error. I'm prone to make these.)
Another thing that I would try is to look if the step width can be incremented.
Take a look at http://modular.fas.harvard.edu/ent/ent_py .
The function sqrtmod does the job if you set a = -1 and p = n.
You missed a small point because the running time of your improved algorithm is still in the order of the square root of n. As long you have only small primes n (let's say less than 2^64), that's ok, and you should probably prefer your implementation to a more complex one.
If the prime n becomes bigger, you might have to switch to an algorithm using a little bit of number theory. To my knowledge, your problem can be solved only with a probabilistic algorithm in time log(n)^3. If I remember correctly, assuming the Riemann hypothesis holds (which most people do), one can show that the running time of the following algorithm (in ruby - sorry, I don't know python) is log(log(n))*log(n)^3:
class Integer
# calculate b to the power of e modulo self
def power(b, e)
raise 'power only defined for integer base' unless b.is_a? Integer
raise 'power only defined for integer exponent' unless e.is_a? Integer
raise 'power is implemented only for positive exponent' if e < 0
return 1 if e.zero?
x = power(b, e>>1)
x *= x
(e & 1).zero? ? x % self : (x*b) % self
end
# Fermat test (probabilistic prime number test)
def prime?(b = 2)
raise "base must be at least 2 in prime?" if b < 2
raise "base must be an integer in prime?" unless b.is_a? Integer
power(b, self >> 1) == 1
end
# find square root of -1 modulo prime
def sqrt_of_minus_one
return 1 if self == 2
return false if (self & 3) != 1
raise 'sqrt_of_minus_one works only for primes' unless prime?
# now just try all numbers (each succeeds with probability 1/2)
2.upto(self) do |b|
e = self >> 1
e >>= 1 while (e & 1).zero?
x = power(b, e)
next if [1, self-1].include? x
loop do
y = (x*x) % self
return x if y == self-1
raise 'sqrt_of_minus_one works only for primes' if y == 1
x = y
end
end
end
end
# find a prime
p = loop do
x = rand(1<<512)
next if (x & 3) != 1
break x if x.prime?
end
puts "%x" % p
puts "%x" % p.sqrt_of_minus_one
The slow part is now finding the prime (which takes approx. log(n)^4 integer operation); finding the square root of -1 takes for 512-bit primes still less than a second.
Consider pre-computing the results and storing them in a file. Nowadays many platforms have a huge disk capacity. Then, obtaining the result will be an O(1) operation.
(Building on Adam's answer.)
Look at the Wikipedia page on quadratic reciprocity:
x^2 ≡ −1 (mod p) is solvable if and only if p ≡ 1 (mod 4).
Then you can avoid the search of a root precisely for those odd prime n's that are not congruent with 1 modulo 4:
def table(n):
if n == 2: return 1
if n%4 != 1: return None # or raise exception
...
Based off OP's second edit:
def table(n):
if n == 2: return 1
if n%4 != 1: return
mod = 0
a1 = n - 1
for a in xrange(1, a1, 2):
mod += a
while mod >= n: mod -= n
if mod == a1: return a//2 + 1
It looks like you're trying to find the square root of -1 modulo n. Unfortunately, this is not an easy problem, depending on what values of n are input into your function. Depending on n, there might not even be a solution. See Wikipedia for more information on this problem.
Edit 2: Surprisingly, strength-reducing the squaring reduces the time a lot, at least on my Python2.5 installation. (I'm surprised because I thought interpreter overhead was taking most of the time, and this doesn't reduce the count of operations in the inner loop.) Reduces the time from 0.572s to 0.146s for table(1234577).
def table(n):
n1 = n - 1
square = 0
for delta in xrange(1, n, 2):
square += delta
if n <= square: square -= n
if square == n1: return delta // 2 + 1
strager posted the same idea but I think less tightly coded. Again, jug's answer is best.
Original answer: Another trivial coding tweak on top of Konrad Rudolph's:
def table(n):
n1 = n - 1
for a in xrange(1, n // 2 + 1):
if (a*a) % n == n1: return a
Speeds it up measurably on my laptop. (About 25% for table(1234577).)
Edit: I didn't notice the python3.0 tag; but the main change was hoisting part of the calculation out of the loop, not the use of xrange. (Academic since there's a better algorithm.)
Is it possible for you to cache the results?
When you calculate a large n you are given the results for the lower n's almost for free.
One thing that you are doing is repeating the calculation -a*a over and over again.
Create a table of the values once and then do look up in the main loop.
Also although this probably doesn't apply to you because your function name is table but if you call a function that takes time to calculate you should cache the result in a table and just do a table look up if you call it again with the same value. This save you the time of calculating all of the values when you first run but you don't waste time repeating the calculation more than once.
I went through and fixed the Harvard version to make it work with python 3.
http://modular.fas.harvard.edu/ent/ent_py
I made some slight changes to make the results exactly the same as the OP's function. There are two possible answers and I forced it to return the smaller answer.
import timeit
def table(n):
if n == 2: return 1
if n%4 != 1: return
a1=n-1
def inversemod(a, p):
x, y = xgcd(a, p)
return x%p
def xgcd(a, b):
x_sign = 1
if a < 0: a = -a; x_sign = -1
x = 1; y = 0; r = 0; s = 1
while b != 0:
(c, q) = (a%b, a//b)
(a, b, r, s, x, y) = (b, c, x-q*r, y-q*s, r, s)
return (x*x_sign, y)
def mul(x, y):
return ((x[0]*y[0]+a1*y[1]*x[1])%n,(x[0]*y[1]+x[1]*y[0])%n)
def pow(x, nn):
ans = (1,0)
xpow = x
while nn != 0:
if nn%2 != 0:
ans = mul(ans, xpow)
xpow = mul(xpow, xpow)
nn >>= 1
return ans
for z in range(2,n) :
u, v = pow((1,z), a1//2)
if v != 0:
vinv = inversemod(v, n)
if (vinv*vinv)%n == a1:
vinv %= n
if vinv <= n//2:
return vinv
else:
return n-vinv
tt=0
pri = [ 5,13,17,29,37,41,53,61,73,89,97,1234577,5915587277,3267000013,3628273133,2860486313,5463458053,3367900313 ]
for x in pri:
t=timeit.Timer('q=table('+str(x)+')','from __main__ import table')
tt +=t.timeit(number=100)
print("table(",x,")=",table(x))
print('total time=',tt/100)
This version takes about 3ms to run through the test cases above.
For comparison using the prime number 1234577
OP Edit2 745ms
The accepted answer 522ms
The above function 0.2ms

Resources