fivemack: (Default)
[personal profile] fivemack
that it is not the beginning of the work, but the continuation thereof until it be thoroughly finished, that yieldeth the true glory

[which I remember as a commonly-used prayer at school, but cannot find through Google; on the other hand, of all the organisations of the Earth, my secondary school is among those whose possession of a totally independent prayer-book would be considered least surprising]

drwxr-xr-x     4 tom  tom    136 Feb  4  2005 mandy
drwxrwxrwx    11 tom  tom    374 Feb  6  2005 Spheres in a cube
drwxrwxrwx   286 tom  tom   9724 Feb  6  2005 FRACTREE
drwxr-xr-x    12 tom  tom    408 Feb 11  2005 fixedprec
drwxr-xr-x    16 tom  tom    544 Mar 15  2005 random_graph
drwxr-xr-x     6 tom  tom    204 Mar 17  2005 enceladus
drwxrwxrwx    31 tom  tom   1054 Mar 17  2005 orbital
drwxr-xr-x    68 tom  tom   2312 Mar 25  2005 srtm
drwxr-xr-x    31 tom  tom   1054 Mar 31  2005 pentagon
drwxr-xr-x    15 tom  tom    510 Apr 19 19:55 crt-fft
drwxr-xr-x    18 tom  tom    612 Apr 19 21:17 primestrings
drwxr-xr-x    15 tom  tom    510 Apr 25 21:42 irreducible
drwxr-xr-x     7 tom  tom    238 May  2 22:36 streetmap
drwxrwxrwx    24 tom  tom    816 Jun  3 21:08 repel
drwxr-xr-x    64 tom  tom   2176 Jun 13 19:09 optbin
drwxr-xr-x     7 tom  tom    238 Jun 19 19:19 mmxsort
drwxr-xr-x    12 tom  tom    408 Jun 23 21:08 eurcit
drwxr-xr-x    13 tom  tom    442 Jun 23 23:09 demographic
drwxr-xr-x    10 tom  tom    340 Jun 29 10:10 smallres
drwxr-xr-x     7 tom  tom    238 Jun 29 11:34 squareful
drwxr-xr-x     9 tom  tom    306 Aug  1 19:40 connected
-rw-r--r--     1 tom  tom   1276 Aug 14 11:12 ling.cpp
-rwxr-xr-x     1 tom  tom  29160 Aug 14 14:33 a.out
-rw-r--r--     1 tom  tom   1769 Aug 14 14:34 ling2.cpp
drwxr-xr-x    41 tom  tom   1394 Aug 22 23:18 haskell_stuff
drwxr-xr-x     8 tom  tom    272 Sep 15 19:59 primefilt
drwxr-xr-x    21 tom  tom    714 Sep 28 23:25 kylix
drwxr-xr-x     8 tom  tom    272 Oct  8 22:22 dudcc
drwxr-xr-x    28 tom  tom    952 Oct 12 00:42 diffeq


Not one of them even close to finished, though I can remember nearly all of their goals; and there are more on the other computer. Some I've worked at intensely for one evening, some for a week of evenings; I don't know how many of them would take much more than a month of evenings to complete, or be worth the effort once completed.

Date: 2005-10-12 08:48 am (UTC)
From: [identity profile] scat0324.livejournal.com
which I remember as a commonly-used prayer at school

Attributed to Sir Francis Drake:


When thou givest to thy servants
to endeavour any great matter
Grant us also to know that it is not the beginning
but the continuing of the same unto the end
until it be thoroughly finished
which yieldeth the true glory.

Date: 2005-10-12 10:40 am (UTC)
From: [identity profile] kaet.livejournal.com
Hey, I worked on the a.out project, too. Perhaps we should share development work? :).

Date: 2005-10-12 02:24 pm (UTC)
From: [identity profile] jojomojo.livejournal.com
What's dud_cc? :)

Date: 2005-10-12 04:40 pm (UTC)
From: [identity profile] fivemack.livejournal.com
Error-correcting codes are what you get if your model of failures is 'I transmit down this wire, and sometimes what comes out isn't what's sent, but I can't tell where the errors are'.

There's also a failure model 'I have N hard discs; sometimes, when I turn the computer on, some of these hard discs will have decided to depart and contemplate their inner doorstop; you can recognise which ones have done this by the grinding noise'. I was trying to figure out useful encodings for that case (something like RAID5, but able to handle two or three disc failures); DudCorrectingCode rather than ErrorCorrectingCode.

Cycling to work this morning, I realised that correcting N errors was equivalent to surviving 2N disc failures [for N errors, you want the encodings of different messages to be at least 2N+1 bits apart; for M failures, you need only M+1 differences], so I'd just be re-inventing Hamming codes.

The only moderately interesting observation was that, if you just start at 00000 and count up, writing down the first thing you come to that differs in the right number of places from everything else you've written down, the bits of the result seem always to be linear functions of the bits you're trying to encode.

Date: 2005-10-12 04:46 pm (UTC)
From: [identity profile] jojomojo.livejournal.com
Ahhh - I was thinking you were writing your own C compiler :)

SURPRISINGNESS THEORY 1)

Date: 2005-10-15 04:45 am (UTC)
From: (Anonymous)
SURPRINGNESS THEORY:
-----------------------
Written on 4am, Saturday 15th October 2005 after waking up with the germ of the theory buzzing in my

mind. It had to be written down.



This note describes the motivation and basis for making mathematical our heuristic notion of how

surprising (or credible) a day, month or year in the life of someone is. This will give us an

mathematical tool to give a credibility score to stories told us by people we know. Is someone taking

too many sick days? Do too many flukish events happen to a given person to be credible? Is someone

pulling the wool over our eyes? We want to know this. Governments and industry want to know this. Our

bosses want to know this. If this notion of the credibility score turns out to have statistical

validity then its applications are endless. No more will people be able to lie and tell tall stories

and get away with it.

This note was inspired by a female friend I shall call Pamela. We were friends for six months from

Jan to June 2005 and she told me many unlikely stories about her life, each one surprising, but in

totality they seemed quite incredible. I pondered how I could use my mathematical skills to draw a

line in the sand and say, "any more tall stories Pamela and I do not believe you any more - I am 95%

confident that you are lying to me." I had to discount the possibility that she is merely a unlucky/

amazing / unusual person. This note is an attempt to make that vision reality.

EXAMPLE
---------

I knew Pamela for six months. During that time she told me:
S1) she was a psychiatrist
S2) she was Indian in origin
S3) she had parents divorced
E4) she got chicken pox in the second week i knew her at age of 26.
E5) her builder fell off a ladder just before doing her patio
S6) she had a fear of acquaintances falling off ladders
S7)her brother studied maths at cambridge like I had and studied to be an actuarial trainee like me

and was very IT literate like I am supposed to be.
E8)she had had a car accident last year which left her with back trouble
S10) she was only the second girl corresponded with on match.com and she immediately asked that I

ring her.
S11) she was a house owner
S12) both her parents were doctors
E13) she drove her car into her builder's van last year and he was unhappy about that.

SURPRISINGNESS THEORY PART 2) - THE MATHS

Date: 2005-10-15 04:48 am (UTC)
From: (Anonymous)
SURPRISINGNESS THEORY - THE MATHS

HEURISTIC THINKING
--------------------

Events Ei and States Si have each occured to only Ni different people on this planet Earth. An event is something that occured at a particular instant and we would use the perfect tense to describe. A state is something that has happened at no particular time and we use the imperfect tense to describe. Thus we can say that S2 [Indian origin]occurs to approximately 1 billion people on this planet as there are a billion indians. E4 occurs to less people but is hard to pin down how many. The basic idea is to work out how many people the combination of Ei and Si occurs to. One can carefully use assumptions of independence of events to assist with this. All of Ei and Si have the property N that they are noteworthy events and states, in that ' they stuck in my mind'. This is where it gets heuristic because any given person P will have e noteworthy events E1...Ee, and s noteworthy states S1...Ss. We seek to calculate the duplication factor of a person D(P,K) which is simply the number of people we estimate there are on the planet Earth who have undergone noteworthy events E1...Ee, and states S1..Ss and K is a measure of how well we know them and so how many noteworthy states and events we know about [clearly I know lots of noteworthy events and states about myself and not many about a stranger in Minneapolis]. If D(P,K) is below some threshold value t(K)then we would say that it is likely we are being lied to, or that we are in the company of a particularly extraordinary person.

This is very heuristic and I don't know much stats. I think there might be some mileage in this. The concept of the number of people D(P,K)=D(P,E1,...,Ee,S1,...Ss) is a solid concept BUT all we end up with is a series of spot values D(P,K) for each person on Earth. However we could get everyone on earth to compute their own local D(P,K) for every other person on earth and then plot D(P,K) against K for all K. This gives a graph G(P) for each person which is their SURPRISINGNESS graph. The difficulty is taking our heuristic notion of 'noteworthiness' N and how well we each know each other K and deducing the axis of a graph from it.

I guess we could measure how well we know someone in other ways. How long have we read about them or talked to them? I know about Tony Blair but have never talked to him. I know about Mum and Dad but never read about them.

METHOD
--------

For me to calculate my D(P,K) for Pamela is easy. I just make crude estimates of how many people there are on planet earth with events Ei having happened to them and States Si being undergone by them.

I then need to calculate K which is how well I know Pamela. I have scores for time spent talking, reading about her and gossiping about her. I can combine this to get a number k which is how well I know her compared to everyone else in the world. So k=1 for myself k=2 for mum k=3 for dad (or vice versa) etc. Possibly Tony Blair would have a low k in this. Then for Pamela (P) I get everyone in the world to plot their own personal D(P,K) on the graph and arrive at a G(P) which is the SURPRISINGNESS GRAPH.

CONCLUSION
------------

Here we have a tool which, [if statistically rigorous], can be used to solve a wide variety of world ills. We shall discover whether it is indeed BLAIR or BLIAR. We shall never pull another sickie. Leaves on the line will be a distant memory. More importantly people like myself who are troubled by psychotic symptoms can calculate the surprisingness of what they are told, get the graph G(P) and know whether to trust people without resorting to psychotic speculation. We will know from G(P) whether someone is likely to be trustworthy to immediate family or to strangers.

SURPRISINGNESS THEORY PART 3) - FURTHER WORK

Date: 2005-10-15 04:49 am (UTC)
From: (Anonymous)
FURTHER IMPROVEMENTS
------------------------

I am a bit worried that some people e.g. Tony Blair have far more noteworthy events than others and each one e.g.Starting Iraq war is not done by many on Earth. However we can accommodate this because Tony Blair is a very well known person. Also, the graph G(Blair) just shows really how many people on earth are like Tony Blair in their experience from the viewpoint of all people in the world. Clearly under any criteria, Tony Blair is a very unusual person and so he is genuinely unusual rather than being inferred to be unusual due to providing us with fallacious events and states e.g. I am having a sickie from food poisoning every day this month.

The stats needs tightening on this. What do I mean tightening? There is no stats yet. It's all heuristic arm waving as yet!! Ho hum.

March 2024

S M T W T F S
     12
3456789
10111213141516
17181920212223
24 252627282930
31      

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Sep. 18th, 2025 08:37 pm
Powered by Dreamwidth Studios