Se vivo con Pepe

Say something nice

2008-10-30T18:44:37Z

Doesn't that feel better?

Mikki Huxtable, 46, McCain voter from Liberty, Missouri– say something good about Obama.

“He‘s a really good family man,” she said. “He’s a momma’s boy, and I say that in a good way– he cares about the moms in this country, because he was raised by his mother and his grandmother. He stopped his campaign to go visit his grandmother because she’s sick. I have a son who I wish was as much as a momma’s boy as Obama is. I wish he would depend on me more."

Lifted from Aaronson's Blog

2008-10-27T17:21:57Z

Comment by ScentOfViolets via shtetl:

While you make valid points, Scott, it is my considered opinion that you’re fighting the good fight just by doing what you love to do and are paid to do. And making more significant contributions. Myself, I see this as a fight between enlightenment-style thinking, with it’s emphasis on empiricism, hypothesis testing, willingness to admit error, and acknowledgment of lack of surety, versus the sort of ‘Dominionist’ thinking that goes with received wisdom, unquestioned dogma, and unwillingness to admit error. Which is almost always in the service of the established order. That’s what this election is all about. A McCain victory would just us up for another round of narrative-style thinking: Government is the problem, not the solution. Our enemies hate us for our freedoms. Antichristianists are corrupting our societal morals and our precious bodily fluids. This whole financial mess could be solved with a tax cut for the rich. There is no crises in medical care, just a crises in government intervention. And so on and so forth.

It won’t last of course, and the present round Dominionism is just about played out.

So just pause, take a deep breath, and turn your thoughts to your garden, as Candide would say.

And as an aside

2008-10-17T19:17:19Z

I think I'm amazed that anyone can write software for a living. I've written about 400 lines in the last two days [nothing terribly challenging; mostly proof of concept for what will become a C++ implementation]. I have spent the last hour or so with IDLE finding all of the naming things that I have hosed up.

EDIT:

File "C:\Documents and Settings\dscoughl\Desktop\SBGEM\py\bullet.py", line 18, in tick

v_new = self.velocity + drag*dt + self._gravity*dt

TypeError: can't multiply sequence by non-int of type 'float'

Who knew?

While I grind away at this ...

2008-10-17T19:03:58Z

do all functions in a python class get 'self' as an implicit argument?

Simple geometry question ...

2008-10-16T13:33:25Z

that I think I understand, but want some back up.

Say I have vector in the x-y plane, and want to rotate it towards the z-axis. I need to use Euler's rotations. Is the scheme then to rotate the vector to an axis, rotate around the other axis in the plane, towards the z-axis, then do the inverse of the first rotation? That should get me back to where is was I think.

freelikebeer @ 2008-10-10T15:26:00

2008-10-10T19:35:54Z

The final auction results are up.

Looks like JPM set the bottom of the market with a big buy at 8.625.
Merrill had the biggest fill [670] at 10.25, and Goldman filled 625 @ 9.75.

Barclays had 775 lots filled at 9.75, over 5 orders [200, 50, 25, 200, 300], and very little anywhere else. I guess they had a CLEAR valuation in their minds.

Citi filled 500 at 9.125.

Funny to see some bottom feeders taking shots at four figure lots in the 2% range.

I count 13 firms acquiring the Lehman's debt, most of which is concentrated in Lehman's direct competitors. Makes me wonder how this plays into the CDS game. Is it cheaper to acquire the debt and pay yourself in the swap, for a lower net outlay?

freelikebeer @ 2008-10-10T14:01:00

2008-10-10T18:03:08Z

I'm pretty stunned by the Lehman auction. Someone is going to make a killing on that debt. Probably in line with what debt collection agencies pay for bad consumer debtors, though.

freelikebeer @ 2008-10-03T10:37:00

2008-10-03T14:38:21Z

Gelman is a statistician at Columbia. Mostly he does social science research. He has a fun historical graphic up on the website for his book.

Buffett Interview

2008-10-03T13:54:03Z

It's an hour long, but he says some interesting things. I've lifted the synopsis from Steve Hsu:

Thursday, October 02, 2008
Buffet on the credit crisis

Charlie Rose interview

Skeptics will claim he is talking his own book, with the recent GS and GE investments. But I think he knows what he is talking about. Buffet says that if he could take a 1 percent stake in the bailout (investing $7 billion), he would. He thinks the bailout will make money investing in distressed mortgage securities at current market prices.

"I love to buy distressed assets... I just don't have $700B to do it with." (At about 13-14 minutes into the hour long interview.)

Bubble logic: "Innovators, Imitators and then the Idiots" (20 minutes)

"Confidence in markets and institutions is like oxygen... when you have it you don't think about it... but you can't go 5 minutes without it." (24 minutes)

"Beware of geeks bearing formulas!" (takes a shot at quants at 27 minutes)

Upper income people should pay more taxes (basically endorses Obama's tax plan) (42 minutes)

"It is terrible that income from investments (capital gains) should be taxed less than income from labor." (against regressive taxes) (44 minutes)

"If AIG had to unwind their derivatives book, it would have hit every institution in the world." (50 minutes)

"The Fed structured the AIG deal very well. They are very likely to get their money back or more." (51 minutes)

Number 4

2008-10-02T13:39:23Z

Number 4 on the list of links for my name in the Google 2001 search for my name had my friends in Tucson all freaked out. They thought I had been killed.

Any ideas?

2008-09-30T18:03:18Z

Anybody have any idea how the specialists are doing today? The last couple days?

Recommendations?

2008-09-24T19:24:43Z

Anyone have recommendations for technical books on fractal geometry?

Eppstein

2008-09-15T18:17:39Z

via 11011110, P=NP

It has been previously demonstrated by the present author that the process of converting a decimal number into a binary number can be represented as a form of the 0-1-Knapsack problem, and therefore is NP-Complete.

Congress, industry and naval analysts remain confused as to why the Navy now says the DDG 1000 cannot use the Standard missile. "Our [combat system] design has the SM-2 using the same link as used in all the other ships," said Raytheon's Smith. ... "I can't answer the question as to why the Navy is now asserting that, after years of funding and years of documentation, that Zumwalt is not equipped with an SM-2 capability," Smith said. -- Chris Cavas, Defensenews.com

Bless his heart.

Abnormal Returns

2008-09-12T18:45:51Z

Just reading the headlines at Abnormal Returns.

If you were going to put money into play, would you be long or short the prevailing wisdom for the next 30 days, 90 days, year?

Boost

2008-09-10T17:54:58Z

Is there a reason not to use boost/thread ?

Indeed

2008-09-05T18:29:37Z

Deputies found food containers ripped apart and strewn everywhere, cans with bear teeth marks, claw marks and bear prints across the Garfield County camp on Tuesday.

Perkins said the operation on Boulder Mountain included 4,000 "starter" sacks of pot and 888 young plants.

"This particular bear apparently was not going to give up and basically chased these marijuana farmers away," Perkins said. "Our county is so tough on drugs that even the wildlife are getting in on the action."

Yes, this bear was doing his civic duty. I'm sure.

GRE Subject Tests

2008-09-02T19:13:40Z

The raw scores for equivalent scaled scores between the math and computer science subject gre's is wacky. A scaled 600 on the CS exam is 10 right answers fewer than on the math exam. It must be incredibly intimidating to prepare for the CS exam and consistently struggle to get even half the questions correct.

Reading and math rambling

2008-08-22T18:21:31Z

Poker got me reading a while ago. I'll get back to the poker in a little while, I think, but today, I was reading about exterior algebras and I finally got it. *DING* On went the light. Algebra and geometry usually bind very tightly in my head, but I couldn't make the jump to understanding it. Maybe I needed a good example to wrap my head around [poker not it this time, though it is a great general purpose problem for a lot of thinking].

This was all because I am trying to avoid building a big stonkin' matrix. Maybe I'm being parochial because of my upbringing, but I need a formalism. Read lots of papers lately that imply, "If you bear with me and make sense of the notation that I use, then this will make sense."

I can't remember if I already wrote about this and I'm too lazy to look it up, so here goes.

markgritter did some work with glpk and sparse matrices [of the variety that I'm trying to avoid!]. We had a little chat about [over?] specifying your constraints for the linear program. That's not what this is about.

His results turned up an unusual fold-the-best hand result [which, IIRC, he attributed to the fact that the nemesis would have already folded, so no loss]. It was interesting, and in passing, that he noted that there would probably be degenerate solutions.

Degenerate solutions to a linear program mean interesting things. If there are two solutions that are degenerate, and your underlying objective function is truly linear, then you should have an infinite number of solutions, namely:
αx_1 + (1-α)x_2 ; α ∈ [0,1]
If that is not the case, then you have formulated your objective function in a way that hides your non-linearities. [There's other interesting conclusions about that that I'm going to ignore for now]

Why am I thinking about this? The Nash equilibrium is formulated so that it is relies on local information. If you are playing a Nash equilibrium strategy, then your [difference-ial|differential] costs decrease your performance against your objective [I say difference-ial because your space might just be a discrete set of choices; the nonlinearity of the objective becomes apparent in the strategic mixing].

This is important to me to understand because, ideally, we are playing against people who are worse than us. They are probably not playing an equilibrium strategy. So, you can't just pick an equilibrium and go with it. The best response to one equilibrium strategy will not be the nemesis strategy of a different equilibrium point.

So, defining the best response dynamic requires more than picking an equilibrium strategy, it requires that you bound your opponents strategy well enough that you can start pick an opposing strategy that is between your opponent's strategy and an equilibrium with no minima in between [and maybe some other stuff depending on how your maxes shake out; maxes in a constraint situation aren't necessarily continuously differentiable].

Then, you figure out if it is worthwhile to move slowly or quickly towards the equilibrium [convolution of future returns to figure out the sweet spot, yada yada yada; infopractical probably has a much better line on that than I do].

I keep saying equilibrium, but I shouldn't. Nash equilibria are extrema of some sort. Must remember that.

Sweet Tooth

2008-08-18T14:42:10Z

I have one. In trying to lose weight, I've completely cut out sugary soda. Now, I'm not so sure that that is a good life strategy. Must... find... balance.

Yahtzee

2008-08-13T18:38:37Z

Say you are playing Yahtzee and it is the first couple of turns of the game. Your first roll leaves you with 1,2,4,5,6, what do you do?

I finally got a hold of the recruiter over at the power company. He was much more forthcoming since he saw that my resume had made it through the first round of HR downselects, and that I am being considered for a couple of positions. For example, he's been reluctant to give me his email address, but now that I am 'in', he sent me an email [the online application system wouldn't eat my cover letter]. I told my stepmother that it would take me getting another position at CurrentCo to get the cosmic forces aligned at the power company. It will be a moderately tough decision if I have to choose between the lab and the power company.

Google meanness?

2008-08-12T19:29:32Z

The rhythmic gymnast on google is a cow.

Physics Ruminations

2008-08-06T19:43:39Z

Also, there has been some discussion between agthorr and prock about physics simulation of big statistical things. I was digging through my brain and remembering. What makes the physics sim a problem, statistically, is that particles go *poof* and appear out of the vacuum. They blip back out, too, but if a particle appears near something that it can interact with, it will, and then you have a branching mess. If you replicate the entire sim state when you get one of these interaction events, you are going to be in a world of computing hurt.

So, the idea is that you order the columns in your state matrix using the characteristic interaction scales as a sorting heuristic [important, for example, in weak-force interactions, which have really short ranges, so don't have to simulate its effects on things 'too' far away]. [Sometimes you read that the Standard Model of particle physics is a local theory; this is part of the rationale for that statement; fundamental theoretical values have been computed using assumptions like this; the other interpretation is that state is described by differential equations, rather than global parameterizations]. So, the effect of the spontaneous production of, say, a gluon, here, doesn't need to be calculated on the proton way the hell over there.

Probably the closest computing analogy is a neural network.

Lab

2008-08-05T19:55:30Z

The manager tells me, "You are telling me a lot of interesting stuff about yourself that is not communicated to me in your resume."

I was pretty flat the whole time. Probably not good, but he said that he is going to get the manager knowledge transfer from my manager. We'll see.

The power company

2008-08-05T17:29:02Z

They are teasing me.

VMWare

2008-08-02T01:46:29Z

Do you virtualize your linux distro and run on Windows, or vice versa?