Last edited by blahblahblah; 11-10-2010 at 10:53 PM.
First, where is the hunter? Either at the North pole, or at any line of latitude such that, if you go 4000 miles south, you reach a line of latitude whose length times some natural number is 4000 miles. And when I talk about the length of the line of latitude, I mean the circumference of the cross-section of the Earth at that latitude.
If the hunter's at the north pole, the bear is probably white-looking (a polar bear). If that's not where he is, then the bear might be brownish, blackish, or greyish; those are the only other colours I can think that bears are found in, and I think there are enough places in the world that the hunter could be so that the bear might be any one of those colours.
If you travel 4000 miles south you are down around the 60 degree north line of latitude. The length of this line is about 12000 miles. The line of latitude that is 4000 miles is about 80 degrees north. Since this is well within 4000 miles of the north pole, you can't go 4000 miles south to get to it. This means that the cabin is at the north pole.
This makes the bear orange. Since no polar bears live that far north, and, as I recall, a wealthy prankster who hated this joke mounted an expedition to the pole to place the orange-dyed, frozen carcass of a bear at the exact pole.
but u have taken it as sinx
anyone here who has been in australia and been to terence tao`s classes or lectures anytime?
For the case that there is only one affair, it has to be known that there has been an affair by all members of the community otherwise the individual has no means of determining that a lack of knowledge of any other affairs results in him being the one.
Let k=2 for the number of affairs: then by the second day, it is common knowledge by both men that: initially they both knew of at least 1 other affair, as a result of the other not killing themselves, each of them must have known of another affair, by the second day, they can reason that it was them and the other they know of.
k=3. Let a,b and c be the cheated on men. At day 2, following from k=2, a assumes that 'had he not been cheated on', then b and c would have by the case k=2, killed themselves on the 2nd day. Thus a was also cheated on.
This follows in our case (for k=9) that by day 8, each of the nine men know of 8 other affairs and by induction know that; had they not also been a victim, the other 8 men would have known of only 7 victims, and by induction all hung themselves on the 8th day. Finally on the 9th day it's common knowledge than 9 men have been cheated on so they all hang themselves.
Last edited by Redbacks; 07-11-2010 at 12:41 AM. Reason: poor wording.... :(
cool stuff redbacks..
u got any problems u willing to share?
u r welcome =)
Tangentially maths-related but anyone got any experience in using Bayesian stats and SEM in their research? Been mulling over introducing it to existing and new projects here but I work mostly with psychologists so if doesn't resemble SPSS output, they freak out. Running my models in R so the fact you actually have to type commands is too much already.
Looking for experiences in terms of whether they were more likely to get their papers up in a good journal, whether you got results which made more sense, clearer conclusions, etc. It's yet to really catch on here but I know both R and Bayes are the new black over in the US right now.
Bayesian stats is an odd one. I know quite a few economists that think it's "overrated," but I think that comment is more often that not specific to the paper's idea as a whole, not the method. It's seductive to think one has stumbled across something significant because one can cite a bunch of statistics, but in the end what matters is what the data brings to an interesting conjecture.... for example, lots of papers are written about mass valuation tools for housing prices, using NN, vector support machines, etc, but they're generally useful only for a practitioner planning to go IPO, not for publication in a serious journal. They tend to rehash established models and are useless from the point of view of furthering knowledge - who the heck cares if prices are not actually linear in number of bathrooms in some suburb of Atlanta. Typically, people have some access to an oddball data set and try their mightiest to publish by spinning it through the washer w/o bothering to think too hard about what might be going on. Editors tend to give such papers to junior faculty like me to reject; it's a pain in the *** but comes with the territory.
BTW, I'd change collaborators if they're that stuck on SPSS! Just my 2cents.
Last edited by Quaggas; 15-11-2010 at 09:48 PM.
I see your point but I do think those who characterise themselves as either Bayesian or frequentist do over-rate Bayes a bit. It's a tool, not a religion and there's room for either/both/all depending on the context.
Just seems like something I should get some work here with, I guess. I saw a stats job for Google, both Bayes and R were mentioned as essential and it's popping up more in research spots here. That said, I do think they're looking more for IT types who can sort-of understand stats rather than a statistician who programs, if you know what I mean. Stats is sort of looked at as a formality; as long as your QQ's look right and p < .05, all good. If they don't, transpose the data until they do; don't, for the love of god, ask questions about your sampling methodology or whether your data might be ****.
Can I just say how much I love that this thread exists?
You guys are the best.
'It seems that perfection is attained, not when there is nothing more to add, but when there is nothing more to take away.'
There are currently 1 users browsing this thread. (0 members and 1 guests)