[Tutor] Help needed
Danny Yoo
dyoo at hashcollision.org
Sat Apr 26 20:41:56 CEST 2014
>>> I want to create two arrays using the above file (Links array and Prob
>>> array) that should give following output:
>>>
>>> *Links *= { [3,5] [5,4] [5,8] [7,8] [14,10] [14,13] [17,13] [14,18]
>>> [10,13] [14,13] [17,13] [12,13] [11,6] [11,9][11,12] [11,19] [19,20]
>>> [15,20] [21,20] [20,21] [21,16] [21,22] }
>>>
>>> *Prob *= {[0.28] [0.15] [0.08] [0.04] [0.04] [0.04] [0.08] [0.04] [0.08]
>>> [0.08] [0.08] [0.08] [0.24] [0.24] [0.34] [0.27] [0.27]}
>>
>>
>> I don't understand how you develop this? The first list has 22 items the
>> second 17. I would have expected them to be the same?
>
>
> In the "Prob" array the elements are less because if you read the note
> below: I said the links that are repeating for example [3,5] their
> probabilities get added and stored as a single value in the "Prob" array.
But what will you plan to do with these values afterwards? I think
Alan's point here is that if there's no direct relationship between
the elements in Links and the elements in Probs, those values aren't
going to be very useful to solve the rest of the problem.
One way to look at this problem is to simplify or normalize the input;
the original structure in the file is slightly weird to process, since
a single line of the input represents several link/failure pairs.
One concrete example is:
4,10,13,14,13,17,13,12,13,0.04
where all these numbers are uninterpreted.
You can imagine something that takes the line above, and breaks it
down into a series of LinkFailure items.
######
class LinkFailure(object):
"""Represents a link and the probability of failure."""
def __init__(self, start, end, failure):
self.start = start
self.end = end
self.failure = failure
######
which represent a link and failure structure. If we have a structure
like this, then it explicitly represents a relationship between a link
and its failure, and the string line:
4,10,13,14,13,17,13,12,13,0.04
can be distilled and represented as a collection of LinkFailure instances:
[LinkFailure(10, 13, 0.04), LinkFailure(14, 13, 0.04),
LinkFailure(17, 13, 0.04), LinkFailure(12, 13, 0.04)]
Then the relationship is explicit.
>> Also why do you want these two lists? What do you plan on doing with them?
>> I would have thought a mapping of link to probability would be much more
>> useful? (mapping => dictionary)
>
>
> I want these two lists because using the links and probs array, I will check
> which link has the lowest probability. Here lowest probability means the
> risk of failure for that link. So based on which link has least probability
> of failure, I will use it to setup a connection (I have a source and
> destination and the links mentioned above are the paths between them) I want
> to select the path which has least failure probability.
>
> Did it make sense?
Unfortunately, I'm still confused. If you just have the Links and the
Probs lists of unequal length, unless there's some additional
information that you're represented, then I don't see the necessary
connection between the two lists that lets you go any further in the
problem. There's no one-to-one-ness: given a link in Links, which
Probs do you want to look at? If you don't represent that linkage in
some way, I don't understand yet where you go next.
>>> So the first element in Links array is [3,5] and its probability of
>>> failure is the first element in Prob array i.e. 0.28
But if the two lists are different lengths, what probability of
failure is associates with the last element in the Prob array?
The representation of data is important: if you choose an awkward
representation, it makes solving this problem more difficult than it
needs be. That is, if you're already compressing multiple elements in
Prob that correspond to the same link, you also need some way to
figure out what link that a compressed probability refer to.
Otherwise, you don't have enough information to solve the problem
anymore.
More information about the Tutor
mailing list