stainless steel rat: Hi-Lo vs Zen

[stainless steel rat]

> I just did a quick tally of the "Top
> 25" posters and the number of their
> posts to this entire site. I don't know how
> far back it goes, but my totals are 3,272!!

> So, when you write that you are frustrated
> to receive a "this is wrong" from
> me, without a detailed, lengthy explanation
> as to why, my answer to you is that,
> occasionally, I don't have the time, in my
> 3,272 posts, to write as much as the
> questioner would like me to.

That's fine, Don. But my point was this: If you take the time to post "you are wrong" then why not take the time to post where. Or else just wait until you either have the time, or run the test, and can answer in the "you are wrong and here is why" methodology.

Clearly in the Zen vs HiLo there was nothing "wrong". There was an anomaly in that Zen came out worse. Because the original Zen as entered in CVCX didn't have late surrender indices. My first question would be as someone that simply answered a question that got somewhat hijacked (is PE a good measure as opposed to score?) should I (or anyone) know everything there is to know about Zen in order to run a sim on it? I suspect the answer is no as the thread went on for a good while before Parker found the answer (original zen has no LS indices).

> Often, I write things like, "See BJA3,
> p. xxx." This may frustrate some people
> who don't have the book. My response is:
> "I've written it up at length once --
> and rather painstakingly, at that -- I'm not
> required to write it ad infinitum and ad
> nauseam, because the same question has been
> asked for the tenth time.

Again, that is OK. However, there is another approach. I answer the same questions all the time, but I realize that for every time I answer them, one more person (at least) has learned the answer. I _could_ refer posters to my Ph.D. thesis when they ask about a particular parallel algorithm I developed. I _could_ refer posters to a paper I had published in a journal. But I rarely do. I do have several of these papers on my web site in HTML format and don't feel bad about giving someone a link and say "read these and if you have questions, feel free to ask after you finish."

I don't particularly mind the book references. I will add that I have BJA2. Based on page #'s you have posted, often with little "context" finding what you are talking about could be difficult (or even impossible). At least one AP has helped me a couple of times by saying "Don's BJA3 page number is XXX in your BJA2 version." And you can feel free to continue doing that as there's not a thing wrong with referring to what anybody might call one of "the" BJ references. I do that often by referring someone to Knuth's series of books, by saying "See Knuth VOL 2, page xxx.

Of course if they don't have it, we go full circle and I end up answering directly.

> You could have taken the "something is
> wrong" as a challenge -- to find the
> error yourself. Parker did.

Now you are using a term I will challenge. "error". Please cite the "error" in my data. Note that "error" does not mean "something against your intuition." "Error" is defined as "mistake". "something done incorrectly".

Omitting indices was hardly an error on my part. I simply gave a specific set of game rules, quoted Norm's sims as distributed, after the "you are wrong" I ran some 1B round sims myself to make _certain_ all the rules and settings were consistent between the two sims, and the answers came out as they did. I don't consider it an error to use standard, as distributed indices for a counting system.

This relates to a particular NSF proposal I once wrote, and left off one person on the reference list that a reviewer thought I should have included. Turns out the reference I had omitted had not yet published any results on this topic, the reviewer just happened to know about the research by personal contact. How exactly would I have been able to cite a reference that had not been published? NO idea. How would I have known to edit and add non-standard indices to a Zen system I know nothing about? So while I won't disagree that Zen (or any good L2 count) is better than Hi-Lo in PE or ultimately in SCORE, in general, the "zen" you tested is not the "accepted zen"...

Were you wrong? I wouldn't say so, no. You knew of some better indices. Was I wrong for not knowing about them? I stand firmly on "no". Was Norm wrong for distributing the standard zen indices? I don't see why.

> Instead, you
> decided, as above, yet again, to take the
> long-winded, defensive route. Everyone here
> is telling you how counter-productive that
> is, but you refuse to listen.

What I refuse to listen to is someone saying that I made an error when it is intuitively obvious to the casual observer that at least in this _one_ case, absolutely no error was made whatsoever. Until we get past that, yes, I will refuse to listen. I can compare numbers. If someone asks me which is better for a particular type of software engineering project, C++ or Java, I can answer that. But I will be talking about C++ or Java without any non-standard features added by someone that found a deficiency. That is the purpose of "standards" in the first place, so we can all talk the same language. When I say Hi-Lo with full indices, I mean as defined by Wong. Not as defined by Einstein or Elvis. I'd refer to those (if they existed) as Hi-Lo (Elvis version) or something to avoid mangling the standard that everyone knows.

So error is wrong.

Period.

> Viktor and Parker have actually asked me to
> post less frequently here. They think I am
> TOO generous with my time on the Free pages,
> and they think that I should spend the large
> percentage of my time on Don's Domain only.
> I have resisted that over the years, but,
> truth be told, discussions like these, and
> your attitude, in particular, are beginning
> to make me see their point of view! :-)

Perhaps you might re-think that "and your attitude." I'm generally a pretty mellow person. But I do expect the common courtesy of not only telling me "you are wrong" but pointing out exactly where and why. I don't see why that is too much to ask.

> You very badly need to lighten up.

_I_ need to lighten up?

I just tried to answer a question. And it turned into some sort of Zen war...

> Don

[Random Poster]

Sorry, I could resist... ;-) You can find the Zen count (along with the LS indices) in Blackbelt in Blackjack 3rd Ed., which of course, you can order here, at the advantageplayer.com catalog. ;-)

> I would certainly hope so!

[Random Poster]

Wow... 35 years.. that's longer than I've been alive!! and unless I'm mistaken, longer than MOST colleges even had a CS department. I know when my father (a programmer) was going for his undergrad and even as high as his post-doc in the early to mid 80's, he had to learn programming through Electrical Engineering classes... You see, this sounds a little suspect to me, considering you would have started teaching in the late 60's, when computers were only available at places like the JPL and, at most, 2 colleges in the world. There wasn't a need for a CS class.

I agree with bfbagain... your credibility is in doubt with me. It didn't help matters when you posted, in such great detail, about your MGM DD LS game and later, said that your memory was a little fuzzy.

"June of this year I played one at the MGM grand... Have not been back. It was in a BJ pit that was on one side of the Dodge Viper that was on display with slots around it luring people to try to win the thing.

I was playing and saw this "Surrender is available" on what I suppose you would call the "podium" in the pit. I asked the dealer and he said "yep". Never gave it any more thought, just surrendered when it seemed to be correct based on the count..."

Here's something else I question... you mentioned on this same thread that you're 57 years old. Given that you've been teaching for the last 35 years, that would make you, what? 22? when you received your professorship and started teaching COMPUTER SCIENCE to most likely grad students (CS wasn't offered to undergrads at that time)? Your style of writing doesn't reflect that of a 57 year old genius professor who taught computer programming (possibly in machine language?) at age 22, _IE_ your use of words such as moron, among other idiosyncrasies you have. Professors are usually more diplomatic than that.

> We need context here. Here is a bit. I'm not
> a hayseed that just fell off a turnip truck.
> I've been a computer science professor for
> 35 years now. That to say that I have done
> my share of research, I have done my share
> of scientific experiments, I have done my
> share of writing reports on those results,
> and so forth.

[mdlbj]

I am a novice and have just gotten BS down. Now I would like to figure out what would be the best system to use. I am not quick with numbers so the simplest counting system would be the way for me to go. Any suggestions?

[sam]

SSR,

I think you set off bullsh*t detectors for many folks from your start here. Your "golly gee, I never knew such places as this existed" initially didn't match with your later declaration of your history in computer science and a four year counting career. It just didn't figure that you're a counter for four years and a computer pro and you didn't know there were websites like this one. Do you see how that might raise questions about who and what you are? You had no concern for cover on these pages. For example, you posted with your university email address which included your name and school and department--seemingly oblivious to the fact that your photo and bio were available as a result. When I brought this to your attention, your reaction was that nobody could track you down. Do you see how that might raise questions about your awareness? Many of your anecdotes contain inconsistencies and/or contradictory details. Most AP's are very conscious of the kind of real life details that you neglect because those details affect the dynamic of the game. Do you see how this causes folks to question you? You seem to have no idea of the way you're perceived by others which causes me to question what you might be like in a casino setting. I wonder if you could play as much as you say and as successfully as you say with such absence of street smarts. Do you see how you could possibly come off as something less than an authority in the seedier but necessary side of this game? I asked you shortly after you began posting if you actually played or if your interest was purely academic because your anecdotes seemed more fiction than fact. Do you see how folks might ask that? My guess is that the conflicts you've had on this site are part of a lifelong pattern of similar conflicts: It's your personality, Man. Wise up and people will be more likely to give you the kind of attention and respect you crave.

Sam

> Wow... 35 years.. that's longer than I've
> been alive!! and unless I'm mistaken, longer
> than MOST colleges even had a CS department.
> I know when my father (a programmer) was
> going for his undergrad and even as high as
> his post-doc in the early to mid 80's, he
> had to learn programming through Electrical
> Engineering classes... You see, this sounds
> a little suspect to me, considering you
> would have started teaching in the late
> 60's, when computers were only available at
> places like the JPL and, at most, 2 colleges
> in the world. There wasn't a need for a CS
> class.

> I agree with bfbagain... your credibility is
> in doubt with me. It didn't help matters
> when you posted, in such great detail, about
> your MGM DD LS game and later, said that
> your memory was a little fuzzy.

> "June of this year I played one at the
> MGM grand... Have not been back. It was in a
> BJ pit that was on one side of the Dodge
> Viper that was on display with slots around
> it luring people to try to win the thing.

> I was playing and saw this "Surrender
> is available" on what I suppose you
> would call the "podium" in the
> pit. I asked the dealer and he said
> "yep". Never gave it any more
> thought, just surrendered when it seemed to
> be correct based on the count..."

> Here's something else I question... you
> mentioned on this same thread that you're 57
> years old. Given that you've been teaching
> for the last 35 years, that would make you,
> what? 22? when you received your
> professorship and started teaching COMPUTER
> SCIENCE to most likely grad students (CS
> wasn't offered to undergrads at that time)?
> Your style of writing doesn't reflect that
> of a 57 year old genius professor who taught
> computer programming (possibly in machine
> language?) at age 22, _IE_ your use of words
> such as moron, among other idiosyncrasies
> you have. Professors are usually more
> diplomatic than that.

[Koolipto]

1. This forum is free.
2. Any amount of energy Don, Parker, Norm, Cac or even the experience players put into answering your questions should be appreciated as a gift.
3. They are helping the development of a large number of players - many of whom have been contributing to this forum for years.
4. The volume of your posts in number and length eclipses that of any one any else - probably in multiples.
5. What interests you or me is probably of more limited interest to the broader communtity.
6. There are a large numbers of readers who try to soak up all they can - even if that means reading every post on these forums.
7. Just reading your posts - let alone thinking and responding to them is a very large and mostly unprofitable undertaking for many of us.

So, I am going to make some suggestions and requests:

1. Limit the number of your posts (actually decide on a number per per day or week and stick to it).
2. Limit the length of your posts.
3. Limit the number of times you will respond in a string.
4. Create an intellectual "shelf" in your mind and use it actively for unresolved questions. After one or two responses from Don or other experts that don't square with your thinking. put the issue on the shelf for future exploration.
5. Limit situational background and self-exposure in your posts to only the minimum necessary.
6. Always think very humbly about whether your contribution or question is of broad interest and will add to the value of this forum.
7. Admit without hedging, obfuscation or justification when you make a mistake.

It is still possible to have fun and discuss issues with this forum. I for one want you to stick around. But these are the Parker Pages and I don't want them renamed.

> That's enough of the soap box. Hopefully you
> understand my perspective in this. I play BJ
> for fun. I have taught people to count. I've
> taught 'em to have fun. I'm beginning to
> realize that it is more fun to hold
> discussions in the traditional computing
> venues where common courtesy abounds in
> general. A little too much acrimony here, is
> the sense I am getting. Unfortunately. I had
> thought that discussing BJ with others would
> be a lot of fun after sort of "living
> alone in the casino" for 4 years. I'm
> beginning to believe that a "that is
> wrong" would definitely be appropriate
> here. I can figure out why.

[stainless steel rat]

> Wow... 35 years.. that's longer than I've
> been alive!! and unless I'm mistaken, longer
> than MOST colleges even had a CS department.

Yep. I started at the University of Southern Mississippi, and took my first computer science course in 1968. I believe the department was formed in 1966, which makes it one of the oldest in the country.

> I know when my father (a programmer) was
> going for his undergrad and even as high as
> his post-doc in the early to mid 80's, he
> had to learn programming through Electrical
> Engineering classes... You see, this sounds
> a little suspect to me, considering you
> would have started teaching in the late
> 60's, when computers were only available at
> places like the JPL and, at most, 2 colleges
> in the world. There wasn't a need for a CS
> class.

Sorry, but the details are correct. I finished my BS in 1970. I was hired to stay at USM which I did until 1985. So fishy or not, it is true. If you subtract 1970 from 2005, you get approximately 35 years? If you are in the Birmingham area, I'd be happy to have you drop by my office, all three of my diplomas (BS MS and PhD) are hanging on my office wall. Dates are 70, 83, and 88 respectively.

> I agree with bfbagain... your credibility is
> in doubt with me. It didn't help matters
> when you posted, in such great detail, about
> your MGM DD LS game and later, said that
> your memory was a little fuzzy.

I still have not resolved that. If you asked me to testify in court, I would _still_ say that to the best of my recollection, that was where I played the game. If you were to ask me "are you 100% sure?" I would have to say "no."

If you want to email me offline, I'll be happy to provide some information that you can verify as to who I am and what I say I am. It is quite easy to verify by visiting a CS department web site I can point you to.

You are free to believe what you want.

> "June of this year I played one at the
> MGM grand... Have not been back. It was in a
> BJ pit that was on one side of the Dodge
> Viper that was on display with slots around
> it luring people to try to win the thing.

> I was playing and saw this "Surrender
> is available" on what I suppose you
> would call the "podium" in the
> pit. I asked the dealer and he said
> "yep". Never gave it any more
> thought, just surrendered when it seemed to
> be correct based on the count..."

> Here's something else I question... you
> mentioned on this same thread that you're 57
> years old. Given that you've been teaching
> for the last 35 years, that would make you,
> what? 22? when you received your
> professorship and started teaching COMPUTER
> SCIENCE to most likely grad students (CS
> wasn't offered to undergrads at that time)?

Correct. Born in 1948, finished BS in CS at USM in 1970. Hired there and worked there until 1985. Can it get any simpler than that. There was _no_ grad program in CS at USM until about 1980 or so. We started with a BS only, and in 1980 added a MS degree. After I left they added a Ph.D. program.

The university I am at now started with a Ph.D. program in 1970 or so, and much later added a BS/MS program. Programs grow up both ways.

You can easily do a little research with the data I gave above to verify it.

> Your style of writing doesn't reflect that
> of a 57 year old genius professor who taught
> computer programming (possibly in machine
> language?) at age 22, _IE_ your use of words
> such as moron, among other idiosyncrasies
> you have. Professors are usually more
> diplomatic than that.

You have a _way_ distorted view of "professors", sorry. But again, believe what you want, or do a little research with the department I gave above. It is _trivial_ to determine whether my statements are true or not.

[stainless steel rat]

> SSR,

> I think you set off bullsh*t detectors for
> many folks from your start here. Your
> "golly gee, I never knew such places as
> this existed" initially didn't match
> with your later declaration of your history
> in computer science and a four year counting
> career.

Can't help that. If you visit a usenet newsgroup like rec.games.chess.computers, or comp.arch, or whatever, you can find posts dating back to whenever DejaNews started archiving newsgroups. I suspect you will recognize my postings.

> It just didn't figure that you're a
> counter for four years and a computer pro
> and you didn't know there were websites like
> this one. Do you see how that might raise
> questions about who and what you are?

Hardly. I recently had a conversation with someone I knew back in the early 1970's where he was active in computer chess. I mentioned the computer chess usenet newsgroup and a message board for computer chess (www.talkchess.com) and he was amazed.

As far as my counting goes, I have been very up-front with what I have been doing. As I have previously said, I've never played with significant frequency. My first two-three years I might have played 1-2 times a year when we would go back to MS, or I would attend a conference in Vegas or Reno (if you look back you might even figure out which conferences I attended in those locations.)

I still don't play that often most of the time, although for the last 6 months I have had more than the usual number of opportunities due to my traveling back to south mississippi for several reasons, from outside program review at USM to visiting family in various places down there where I grew up...

I'm not a pro. I _really_ was doing this only because my wife got interested in slots and BJ on a visit to Biloxi while my son and I were at a Summer week-long boy scout camp.

> You
> had no concern for cover on these pages. For
> example, you posted with your university
> email address which included your name and
> school and department--seemingly oblivious
> to the fact that your photo and bio were
> available as a result. When I brought this
> to your attention, your reaction was that
> nobody could track you down.

Quite the contrary, my reaction was "anybody can track me or you down." Not "nobody can do so." I think you can find examples of that. I've done it more than once where I am now, tracking down people that are breaking in to our systems and causing massive headaches for everyone. I did listen, and I did change my email as you notice, but not because I believe it makes me completely anonymous, only because it made sense in listening to others...

> Do you see how
> that might raise questions about your
> awareness? Many of your anecdotes contain
> inconsistencies and/or contradictory
> details. Most AP's are very conscious of the
> kind of real life details that you neglect
> because those details affect the dynamic of
> the game. Do you see how this causes folks
> to question you?

I've said this multiple times. I really do not consider myself an AP in the classic sense. I don't play frequently, I rarely play for big stakes (I have played a $100 min table twice in my life, I believe, I have played $25 min tables relatively frequently, but usually play lower. I do this because (a) my wife enjoys going; (b) I enjoy winning; (c) I originally wanted to see if I could actually do it. I have many more things that I do seriously. I lump BJ in with my other hobbies such as hunting and fishing. I can tell you my boat will run 80 MPH. I can't tell you exactly where I caught my last 8+ pound largemouth or my last 25+ pound freshwater striper. Things that I do for fun, I probably am a bit loose with the details. I can tell you when my computer chess program played its first game. I can tell you the years I won the world computer chess championship using that program. I can tell you the years that I gave demos in places like Paris, London, and the like. Because that was all "carreer-oriented". But at 57 years old, I've slowly learned what is important to remember and what are just mind-clogging details that are unimportant...

I'll give you the same advice I give people on the internet _all_ the time. If there is a poster you don't like, or one you don't trust, or one you don't want to read, then just don't read. This is not a forced readership enterprise, at least as far as I have determined...

So whether you believe I have been counting sporadically for five years or not, whether you believe I have been a faculty member for 35 years or not, whether you believe I have even been in a casino or not, doesn't matter to me. I simply chose to participate because I realized long back that there was probably a lot about counting I didn't know. However, it is becoming a bit apparent that maybe this is not the best place for me to "hang out" which is also ok...

> You seem to have no idea of
> the way you're perceived by others which
> causes me to question what you might be like
> in a casino setting. I wonder if you could
> play as much as you say and as successfully
> as you say with such absence of street
> smarts.

"As much as I claim"? Never realized I claimed to play very often. I believe I have been quite clear that if I am lucky I will get to vegas once or twice a year. I have been three times in the last two years, roughly. Twice for conferences, once for a week+ vacation with my wife, brother and his wife. I get to play more frequently on the MS coast, about 5 hours away from here, because it is near family where I grew up in South Mississippi. Over the past 5 years I'd bet I have not averaged 4 trips per year, although for the past 6 months I've been a bit luckier in getting down there as family down there often say "lets drive down to the boats tonight" which suits me...

Do you see how you could possibly
> come off as something less than an authority
> in the seedier but necessary side of this
> game?

I definitely _am_ "something less than authority." I can tell you what I have done. I can tell you some mistakes I have made. I can tell you some things I have learned. Playing as infrequently as I play, I don't see how I could be any kind of "authority" at all, other than in the mechanics of counting which I can do pretty well. At least CVBJ seems to think so.

> I asked you shortly after you began
> posting if you actually played or if your
> interest was purely academic because your
> anecdotes seemed more fiction than fact. Do
> you see how folks might ask that? My guess
> is that the conflicts you've had on this
> site are part of a lifelong pattern of
> similar conflicts

Don't really have any "life-long pattern of conflicts" sorry. So that speculation you will have to answer yourself.

>: It's your personality,
> Man. Wise up and people will be more likely
> to give you the kind of attention and
> respect you crave.

> Sam

I don't really "crave" anything. I am one of "those" that actually has students call me by first name if they will. I'm not hung up on the "doctor" stuff or the "professor" stuff.

You seem to be pretty quick to make the "classic internet mistake" of making character assumptions based on internet posts only, no body language, no facial expressions, no speech inflections, or anything. A person's internet postings offers a very poor one-dimensional view of what they are like. Sometimes that view can be accurate, sometimes not. I've learned several things you might try for yourself. (1) don't form personal opinions from text posted on a public message forum; (2) remember that most any written sentence can be interpreted in multiple ways, if you want to view the "dark side" you will find one, the inverse is true as well. (3) you can almost always find mistakes in what people write. If you let the mistakes get in the way of the main point they are writing about, you fall into the "can't see the forest for all those damned trees in the way" trap. It happens to everyone. And some finally figure it out and stop tripping over it repeatedly. (4) things posted on the internet should not be taken personally, almost anything can be, although almost everything is not intended that way (perhaps Snyder's "fight club" could be considered an exception of course.)

If you remember those points, you'll end up a lot happier.

whether you like me or not doesn't matter at all to me. We'll probably never meet. If we were to, we could either become best friends or worst enemies, neither having anything to do with what is written here. And no, that is _not_ a personal remark. Simply an observation.

[Norm Wattenberger]

> In
> simulations, N0 is meaningless to determine
> how many sim rounds to run, because the
> properties of the random number generator
> have a lot to do with this.

Not if you use a decent RNG. A decent RNG is for all practical purpose random in a BJ sim. All you need to worry about is standard error.

> for
> example you can run too many rounds trying
> to get the result to stabilize and run afoul
> of IEEE floating point rounding/truncation
> errors and actually start to get less
> accurate answers.

Not if the sim is written correctly. Counters in CVData are integers anyhow. Floating point should never be used for counters.

[stainless steel rat]

> 1. This forum is free.

Of course. There are many free forums on the internet.

> 2. Any amount of energy Don, Parker, Norm,
> Cac or even the experience players put into
> answering your questions should be
> appreciated as a gift.

It certainly is appreciated.

> 3. They are helping the development of a
> large number of players - many of whom have
> been contributing to this forum for years.
> 4. The volume of your posts in number and
> length eclipses that of any one any else -
> probably in multiples.

what can I say? We "academicians" get paid to write. It becomes a way of life.

> 5. What interests you or me is probably of
> more limited interest to the broader
> communtity.
> 6. There are a large numbers of readers who
> try to soak up all they can - even if that
> means reading every post on these forums.
> 7. Just reading your posts - let alone
> thinking and responding to them is a very
> large and mostly unprofitable undertaking
> for many of us.

> So, I am going to make some suggestions and
> requests:

> 1. Limit the number of your posts (actually
> decide on a number per per day or week and
> stick to it).

I'm really thinking the right number here is "zero" which is my plan after responding this morning. BJ is (to me) more about fun than money. Yes winning makes it more fun. But for the forseeable future, it is a fun thing to do on sporadic occasions.

Here, there is far more "nit-picking" (IMHO) that is needed. Which means that I probably don't belong as a "fun part-timer" if the majority are _really_ wanting to become "pros" (which makes little sense but that is another topic).

> 2. Limit the length of your posts.
> 3. Limit the number of times you will
> respond in a string.
> 4. Create an intellectual "shelf"
> in your mind and use it actively for
> unresolved questions. After one or two
> responses from Don or other experts that
> don't square with your thinking. put the
> issue on the shelf for future exploration.
> 5. Limit situational background and
> self-exposure in your posts to only the
> minimum necessary.
> 6. Always think very humbly about whether
> your contribution or question is of broad
> interest and will add to the value of this
> forum.
> 7. Admit without hedging, obfuscation or
> justification when you make a mistake.

> It is still possible to have fun and discuss
> issues with this forum. I for one want you
> to stick around. But these are the Parker
> Pages and I don't want them renamed.

I agree. I think my comments are best left in the casinos where I play or at family gatherings. I've seen this same sort of dichotomy in computer chess. 99% of the people involved are in it for "fun" (me included). 1% (or less) have been involved in it as a commercial career, selling computer chess programs is a small niche market that can make money. And in general, the "fun people" and the "commercial people" don't get along. The "fun people" want to share ideas and push computer chess forward. The "commercial people" won't reveal their secret tricks of the year, but they lurk on the various CC message boards grabing new ideas posted by others, but never giving anything back to the general fun-lovers.

Sound familiar? I thought it would...

[stainless steel rat]

> Not if you use a decent RNG. A decent RNG is
> for all practical purpose random in a BJ
> sim. All you need to worry about is standard
> error.

> Not if the sim is written correctly.
> Counters in CVData are integers anyhow.
> Floating point should never be used for
> counters.

There is another issue. "period". A 32 bit RNG can't have a period longer the 4 billion for obvious reasons. Most do far worse. I personally use a "mersenne twister" RNG from the book "Numerical Recipes". But students often run afoul of the 4 billion random number limit. Run a billion round sim and I'll guarantee you that if you use a 32 bit RNG, that you fall into a cycle and get the same pattern of numbers over and over and over. The cycle can be long, but not long enough. For example, I've had students that ran sims (my programming assignment based on blackjack BS verification for distributed computing) for 100 billion cycles. Or more. They are using 64 bit hardware (AMD opterons or sun ultrasparcs) which push the cycle length to something obscene. But when they try to run on 32 bit machines at home, and let them run for a long time, sometimes they don't get the right answer because once you get wrapped into a cycle, no further accuracy will be produced.

Also, in my chess program, integers are no good. I search (on my dual xeon 2.8ghz in my office) about 2.5 million chess positions per second. An unsigned integer counter can count to about 4 billion (signed int is 1/2 that). I had to go to 64 bit counters because a 20 minute search will overflow a 32 bit counter. I ran on a quad 2.4ghz opteron for a chess tournament last year, and there was searching up to 16 million nodes per second. A 32 bit counter would be worthless there. (most modern C compilers support either "long long" or "__int64" data types to get 64 bit counters on a 32 bit architecture, although there is a performance penalty.)

People here tend to think in terms of 2 billion round sims as huge. I've seen (non-BJ) simulations run for _weeks_ on a Cray. On a project where we used 100 billion rounds to produce a first approximation and get some good starting points for the real runs...

Might well be overkill in BJ, but it won't be long before a 2B round BJ sim will not take long. Once you get your threaded version up and drop it into a 4-way opteron with dual core CPUs, I'll bet that 2B round run of complete sims will be measured in minutes rather than in hours.

[Norm Wattenberger]

> Wow... 35 years.. that's longer than I've
> been alive!! and unless I'm mistaken, longer
> than MOST colleges even had a CS department.
> I know when my father (a programmer) was
> going for his undergrad and even as high as
> his post-doc in the early to mid 80's, he
> had to learn programming through Electrical
> Engineering classes... You see, this sounds
> a little suspect to me, considering you
> would have started teaching in the late
> 60's, when computers were only available at
> places like the JPL and, at most, 2 colleges
> in the world. There wasn't a need for a CS
> class.

Sorry but your history is off. First computer I used was 39 years ago at tiny Haverford College. I wrote an heuristic checker program 38 years ago on an IBM 7040 at U of P. 36 years ago I was hired by a guy with a PHD in CS. Obviously there were classes before that 36 years ago I had already used the following computers: IBM 7040, IBM 7044, IBM 1130, IBM 1800, IBM 360/40, IBM 360/20, Monroe Mark IV, LGP-30 (General Precision), RPC-4000 (Royal Precision), SDS 940 (Scientific data systems), and DEC PDP-8. I was also familiar with the GE 635, RCA Spectra, Xerox Sigma, Univac 1106 & CDC 6600. This stuff goes back farther than most people realize.

[Norm Wattenberger]

On RNGs, I use MarZam 2. I forget the period but I once figured it could sim all hands that could have been played since the big bang. But more importantly, it passes all the Marsaglia Diehard tests. It's a combination of two 32bit RNGs and is also parallel capable.

On counters, CVData is a two-stage sim. There are about 100,000 integer counters. The major sim cycle is 1,000,000 rounds. (25,000,000 rounds for CVCX.) Every million rounds, all the counters are added to 100,000 floating point accumulators and reset. This adds to the significance of the floating point values and removes the rounding problems of billions of increments. It also means that integer counters can be used without runnning into the 4 billion limit.

On sim length and speed, All sims in BJA 3e are 20 billion rounds. So I certainly agree about sim length.

> There is another issue. "period".
> A 32 bit RNG can't have a period longer the
> 4 billion for obvious reasons. Most do far
> worse. I personally use a "mersenne
> twister" RNG from the book
> "Numerical Recipes". But students
> often run afoul of the 4 billion random
> number limit. Run a billion round sim and
> I'll guarantee you that if you use a 32 bit
> RNG, that you fall into a cycle and get the
> same pattern of numbers over and over and
> over. The cycle can be long, but not long
> enough. For example, I've had students that
> ran sims (my programming assignment based on
> blackjack BS verification for distributed
> computing) for 100 billion cycles. Or more.
> They are using 64 bit hardware (AMD opterons
> or sun ultrasparcs) which push the cycle
> length to something obscene. But when they
> try to run on 32 bit machines at home, and
> let them run for a long time, sometimes they
> don't get the right answer because once you
> get wrapped into a cycle, no further
> accuracy will be produced.

> Also, in my chess program, integers are no
> good. I search (on my dual xeon 2.8ghz in my
> office) about 2.5 million chess positions
> per second. An unsigned integer counter can
> count to about 4 billion (signed int is 1/2
> that). I had to go to 64 bit counters
> because a 20 minute search will overflow a
> 32 bit counter. I ran on a quad 2.4ghz
> opteron for a chess tournament last year,
> and there was searching up to 16 million
> nodes per second. A 32 bit counter would be
> worthless there. (most modern C compilers
> support either "long long" or
> "__int64" data types to get 64 bit
> counters on a 32 bit architecture, although
> there is a performance penalty.)

> People here tend to think in terms of 2
> billion round sims as huge. I've seen
> (non-BJ) simulations run for _weeks_ on a
> Cray. On a project where we used 100 billion
> rounds to produce a first approximation and
> get some good starting points for the real
> runs...

> Might well be overkill in BJ, but it won't
> be long before a 2B round BJ sim will not
> take long. Once you get your threaded
> version up and drop it into a 4-way opteron
> with dual core CPUs, I'll bet that 2B round
> run of complete sims will be measured in
> minutes rather than in hours.