Question

pset 8 problem 4 ??

it took me a long time to do this one. I used a test dictionary with only 4 items and it worked. I'd like to compare results using the 320 item dictionary from the problem set . tia

These are my results, anyone get the same? different?:

with work constraint = 5
result:
Total Value: 41
Total Work: 5

with work constraint = 10
result:
Total Value: 70
Total Work: 10

with work constraint = 20
result:
Total Value: 114
Total Work: 20

with work constraint = 40
result:
Total Value: 177
Total Work: 40


thnxpset 8 problem 4 ??

it took me a long time to do this one. I used

Answer 1

i didn't understand ur question

Answer 2

I wrote a program for pset8 problem four that i think works but I haven't taken the time to figure out the optimal results by hand so I'd like to compare my results with others so that I will have confidence in my solution - so I guess it wasn't actually a question. Those results are posted above. Anyone want to share?

Also (more importantly i guess)-
just because my results may be right doesn't mean that I wrote an optimal solution so i would like to also compare the number of calls for the recursive part - if you are game. Here's what I got:

work = 5 numcalls: 2112
work = 10 numcalls: 4272
work = 15 numcalls: 6852
work = 20 numcalls: 9803
work = 30 numcalls: 16080
work = 40 numcalls: 22308

for the number of calls, i'm not looking for precise comparisons but I'm curious how the 'magnitude' compares and how it changes with work.

thnx

Answer 3

man, im stuck on this one....gonna sleep on it

Answer 4

Got the same results than you! Actually that makes me feel better about mine as well :)

Answer 5

thnx

Answer 6

yeay!!....
numcalls too??

Answer 7

yours is faster...

Answer 8

hmm - a lot different or about the same magnitudes?? did you get in the 'thousands' for work = 5 and 10?

And did it seem to increase by a little more than 2 each time you doubled the work?

Answer 9

The rate of growth was about the same, but was almost twice your numbers

Answer 10

If you don't mind posting your functions, I'd like to figure out why there is a difference between ours. I think it will help me understand it better, Even though mine works I'm not sure I completely understand it. here's mine...... hope its not too unreadable, i left a bunch of comments and prit statements in it.

Answer 11

Mine is also not very readable, let's see if we can figur out the differences

Answer 12

I did a quick comparison and they both look pretty much the same except -

At line 300/304 yours has an if/else statement:
you compare the with and without results inside of this else clause -and memoize and return the greater of the two.

Mine has the same if/else statement at lines 157/160
except I do the take and dont_take comparison (starting at line 168) outside of the else clause (also memoizing and returning the greater of the two)

I haven't had time to play with yours so I don't know if this difference is making the performance different.

It is interesting though that this is the part of my code that I don't understand - originally i thought my lines 168 - 190 would only be executed once in the topmost recursion (did that make sense?) but when I put a print statement at line 188 I found that it gets executed multiple times - so i was confused again. I haven't revisited that til now - guess i'm going to have to figure it out.

Answer 13

Because the relationship between mine and yours is that mine is about twice as big as your but the rate of change doesn't really change, maybe it has to do with the fact that I store in my memo the index of the solution plus the associated total value, while you handle the total value in dpadvisor (instead of dpadvisorHelper) can this be the difference?

Answer 14

i'll have to think about that. I'm tempted to say that storing the value as well as the index shouldn't cause the number of calls to be greater. The idea of the memo is to reduce the number of calls - and that should rely on the dictionary keys not the dictionary values - we both use the same thing for a key

I actually thought yours handled that better - grabbing the value right then when it was available instead of mine that has to recreate it again.

The only structural difference i see is at the end there where you test inside the else clause and i test outside it - maybe mine gets to the test more often and can memoize more 'whole branches'. maybe I'll put a print statement in at the test and see there is difference on how often that happens.

Answer 15

You might want to look at my heavily annotated version: http://codepad.org/Krm2BRz6

Answer 16

yep i've seen that one before - way more efficient. unfortunately i didn't spend enuff time understanding that 20bits article and went off trying to mimic the decision tree process using a dictionary for the memo