Universal Rule

Book Reviews

Monday, May 24, 2004

Solving the Einstein's Riddle in Python -- Comment()
Using the skills so far learned I managed to implement a Python code for solving the so-called Einstein's Riddle. I solved the problem in two ways, first the straightforward way by using lists and list comprehensions, etc. Then I refactored to code using the Python object model, by deriving new datatypes from the Python list object. This object-oriented version (updated the link on May 29, 2007) is about 50% slower, but easier to understand. Here follows a short extract from the object-oriented version:

def alltrue(l, op=operator.and_):
    return reduce(op, l)

def anytrue(l, op=operator.or_):
    return reduce(op, l)

class Pattern(list):
    def __eq__(self, other):
        return alltrue([u==v for (u,v) in zip(self,other) if u!='_'])
    def match(self, street):
        return anytrue([self==h for h in street])
    def __call__(self, street):
        return [h for h in street if self.match(h)]
    def nr(self, street):
        for i in xrange(len(street)):
            if self==street[i]: return i

class House(list): pass

class Street(list):
    def __init__(self,*l):
        try:
            self[:] = [House()+h for h in l[0]]
        except IndexError:
            self[:] = [House() for i in xrange(5)]
    def new(self,items):
        st = Street(self[:])
        [h.append(d) for (h,d) in zip(st,items)]
        return st
    def __str__(self):
        t = unzip(self)
        s = ''
        for u in t:
            s = s + ''.join([("%-10s" % str(v)) for v in u]) + 'n'
        return s

My input data was in Finnish, so I'm afraid the following result may be difficult to read:

1         2         3         4         5         
kelt      sini      pun       vihr      valk      
norja     tanska    suomi     saksa     ruotsi    
vesi      tee       maito     kahvi     olut      
blues     ooppera   rock      iskelma   hevi      
kissa     hevonen   lintu     kalat     koira

(Also, I don't like smoking, so I changed that part of the data to different types of music.)

But what is the final solution? Well, the German ('saksa') owns the fish ('kalat') in house 4.

Here is a follow-up question. What if we interpret the rule

4. The Green house is on the left of the White house.

so that there may be houses between the Green and White house. How many different solutions do you get?

List processing in Python -- Comment()
Python is an elegant language, but the lack of some features bothers me. On the other hand, most of the features are easy to implement as Python functions.

One of the missing features was a function for "flattening" a nested list structure. Here is some code to provide this functionality:

def flatten(L):
    if not isinstance(L,list): return [L]
    if L == []: return L
    return flatten(L[0]) + flatten(L[1:])

l1 = ['a', 'b']
l2 = [1, 2, 3]
l3 = [1.0J]
L = [-1, l1, [l2, l3], 3]
print L
print flatten(L)

The output looks like this:

[-1, ['a', 'b'], [[1, 2, 3], [1j]], 3]
[-1, 'a', 'b', 1, 2, 3, 1j, 3]

The tricky part was how to check whether the argument L is not a list. At first I tried just indexing L, and caught the exception in a try: ... except IndexError: structure. But this resulted in an infinitely recursive loop when a list items was a string. The solution above works rather nicely, also for classes which are derived from the list class.

Are Mac users the muslims of PC-dominated world? -- Comment()
What do Macs and Islam have in common? "The choice to use a Mac or to be Muslim in today's world is not easily understood. In a world where the vast majority of computers are PC, where Islam is perceived as the enemy, why would someone choose to be a Mac user or a Muslim? Bias against both is rampant."

-- Comment()
Camilo commented The Mathematics of Marriage: "I read James Murray's Mathematical Biology, and it is awesome; he is one of the authors of the book, and a superb researcher and expert in NLD. I do intend to get my hands on this book!"