Sales of the book don't matter to Wolfam -- Mathematica is MUCH more lucrative. The book is just an advert for his software.
Believe it or not there are open source computer algebra systems! Two that I'm familiar with are Maxima and PARI-GP (sorry, don't have a link at the moment). Maxima is general purpose, while PARI-GP is mostly about number theory.
Maxima's history is interesting. It is based on the source code (Lisp!) of the Macsyma system developed at MIT circa 1970-1980. Mathematica is essentially a rewrite of Macsyma with very slightly different syntax. You know what they say about imitation.
If Pinochet was such a great guy, let him win the next election.
If it was really such a mess, the voters would
be only too happy to give Allende the boot.
Or was the CIA just speeding up the process?
Over 15 years, 3000 people were killed, but this was remarkably humane compared to the communist revolutions at the time.
(1) Remarkably inhumane, compared to the
means by which Allende got himself into the
presidency.
(2) Military governments have a such a great
track record in economic reform; 3000 souls is
really not much to pay for such expertise.
(3) In really civilized countries, they don't need
to kill anyone to implement economic reforms.
More precisely, there were many new fields within mathematics to explore. However, there was already quite a large body of existing knowledge. It's just that it was about as much as a sophomore engineering student knows(give or take).
No way, dude. The original poster who said "A century ago, mathematics was primarily a new field" was way off base, and the follow up isn't any closer. Sophmore engineering students are pretty amazing, I know -- check out those concrete canoes! -- but their math curriculum encompasses about one percent of the math available a century ago.
The last person who might possibly have
mastered the whole of mathematics as it existed
in his era was Henri Poincare'. Incidentally, he
did much of his most memorable work just about 100 years ago. The suggestion that today's undergrads might have a comprehension comparable to his, is just silly.
No way would Voltaire have ever said anything about self-sacrifice. In his writing, he is much more detached and urbane, and not nearly so self-righteously overbearing.
According to this website this bit about dying is a paraphrase of "Think for yourselves and let others enjoy the privilege to do so too" which Voltaire did write. I dunno; Voltaire didn't say anything about himself or anybody else dying in the name of free speech.
Aside from Voltaire's lack of interest in self-sacrifice, the misquote is just too clumsy for him. He was consistently witty and well-polished, although I admit the above quote ("Think for yourselves...") doesn't really show his writing skills at their best.
In short, the misquote is certainly NOT the sort of thing Voltaire would write. Thomas Paine, perhaps.
For statistical analysis, my advice is to go for R (GPL implementation of S, see www.r-project.org), or S+ (the commercial implementation of S). R has functionality equivalent to S+ and it's GPL. There are a lot of S language scripts to be found on the web, e.g., StatLib. Both R & S+ have very active user communities.
Especially if you're doing policy analysis (i.e., you're not data mining a 10 million record database) R is a great environment to work in. Hmm, it might be possible to process 10 million record databases with R, I just haven't tried it.
R has more or less equivalent functionality to Octave, but the programming language (S) is more interesting, and plotting functions are more sophisticated. For matrix operations, both R and Octave use BLAS.
My advice is to stay away from SAS and SPSS. These two were invented back when "everybody knows using a computer is hard". Both have GUI interfaces now, but that's just an attempt to hide the essential ugliness of their command languages. If you must use a commercial product then S+ is my recommendation.
There are, believe it or not, beautiful pieces of Fortran IV out there --
You better believe it. If we're looking for long-lived code, we should start looking at Netlib. Netlib is full of beautiful code -- beautiful in part because it is beautifully written, but also because it expresses beautiful ideas. What makes the bazillions of lines of C++ and Javascript disposable is that it's mostly about ideas that might just as well be forgotten.
My nomination for the Beautiful Fortran Award would have to be ARPACK. It is clear, concise, elegant, efficient, and TOTALLY ROCKS. Eigenvalue problems are not easy, which only increases the beauty of any solution.
...like this is just another stab at A.I.? It's hardly a robot of any type according to most standards, but rather a program that has
some limited self-awareness?
Well, let's not use the term AI -- it has too much baggage and nobody knows what it means anyway.
There are three aspects to making workable self-fixable photocopiers and water plants: (1) figure out how such a system should figure out what's happening and how it should figure out what to do about it, (2) implement a system that carries out such figuring, and (3) encode knowledge about a specific domain so that the system can crunch on it.
Problem (1) is solved, AFAICT. The right thing to do is to use probability to represent beliefs about the world, and to use utility theory to express how happy you are with the state of the world. The combination of those two concepts is conventionally called decision theory.
Problem (2) is very difficult. Having decided you want to build a general decision theoretic system, you immediately get into serious computational difficulties. Here's the tip of the iceberg: if there is more than one chain of events that lead from causes to effects, it's a hard problem. Time-dependent problems (i.e., most real problems) have this characteristic left and right.
Problem (3) is difficult in a practical sense as opposed to a theoretical sense. This is basic engineering -- translating a real problem into a mathematical representation suitable for solution.
The payoff of this approach is that you can put together the model of the system from the bottom up -- each little component can be modeled separately, and then the rules of decision theory tell you how to combine all the pieces. Essentially, you construct a system that is an "automatic heuristic generator". Lights out? Check the battery? Battery is good? Check the wires... etc. Such a system can generate an appropriate response in many more scenarios than human heuristic-generators.
auai.org is a great resource for information on automated reasoning systems.
I attempted some research in this direction -- see riso.sourceforge.net; all the results are in my dissertation.
I don't know if this particular product is really going to be the long-awaited breakthrough in multi-modal input, but: MS is headed in that direction and they're putting up the $$ to make it happen.
As you know, MS has deep pockets, and they've used some of the money extorted from W95 lusers to fund serious research in machine learning. In my own research field (belief networks,
shameless plug) they have hired several of the big wheels in the field. Patrice Simard used to work in neural networks; don't know what he's been up to lately -- you can probably find papers by him at Research Index.
Recognition of handwritten characters is a hard, interesting problem; pattern recognition on bitmap images can only take you so far, and to go beyond that you need to start incorporating additional information such as stroke order and letter and word context. Voice recognition is even more difficult.
The bottom line is that since there exist computers (namely our brains) that can reliably carry out text and voice recognition, we know it's possible. Getting there will require solving several substantial engineering problems, and it's possible a university department somewhere will carry through the solution, or perhaps a corporation that can afford to have a group dedicated to technology that won't pay off in the next three months... such as MS.
Here are some additional references, on-line & off, about Bayesian probability.
On the web, see: Assoc. for Uncertainty in Artificial Intelligence -- this is the primary conference devoted to belief networks, which are a class of graphical (in the circles and arrows sense) Bayesian probability models. There are tutorials and other papers on the main AUAI web page, and links to the last several years of conference proceedings. By the way, Heckerman and Horvitz, now doing belief networkish work at MS Research, are in the AUAI crowd.
In print, my favorite reference is E.T. Jaynes, "Probability Theory: The Logic of Science", which is due out soon. See this web site devoted to Jaynes' work for the status. I am also fond of Castillo, Gutierrez, & Hadi, "Expert Systems and Probabilistic Network Models".
There are a vast (well, maybe just large) number of alternative models to classify things; a good introduction is Hastie, Tibshirani, & Friedman, "Elements of Statistical Learning". Incidentally, they use spam classification to illustrate several kinds of models.
Finally, if you're wondering what the heck is the difference between Bayesian probability and any other kind -- just google the posts in sci.stat.math; there is a Bayesian vs frequentist flame war about once a year.:^)
I wonder how cool ideas at Google fare against the need to get into some serious code optimization, scaling problems, etc. By "cool idea" I mean something like the page rank algorithm. Does Google spend a lot of time coming up with more cool ideas, or do you already have plenty of ideas, and you need to invest more in developing the existing ideas? How many developers are there for every ideas person -- 1 to 1, 10 to 1, 100 to 1 ??
I do quite a bit of number crunching. Here are
some of the resources I use:
Netlib (www.netlib.org) -- Yes, it's mostly Fortran, but that's a good thing! Just use f2c (easy to find) and translate to C if that's what you want. Don't underestimate the power of decades-old programs -- old == widely used and well-tested.
StatLib (lib.stat.cmu.edu) -- Collection of statistical software, in various languages, including C, Fortran, and S.
SAL, Scientific Applications on Linux (sal.kachinatech.com) -- a very large collection of links.
Freshmeat (www.freshmeat.net) -- Not scientifically oriented, but there is much scientific stuff there, along with all kinds of miscellany.
Octave (www.octave.org) -- A package for matrix manipulations, similar to Matlab, but free. Useful for all kinds of problems.
R (www.r-project.org) -- An implementation of the S language for statistics, but also useful for general problems, similar to Octave. S+ is a commercial implementation of S.
Well, that ought to be enough to get started. To echo something other posters have mentioned -- don't even bother with Windows software. If your budget is tight, save your money for hardware, don't waste it on the MS tax.
Research Index (previously called Citeseer) is truly amazing. The RI collection is built by a web crawler, which finds papers published on the web, and -- get this -- automatically extracts bibliographic info such as the title, author, abstract, and citations. For each paper in the collection, RI also finds citations of that paper, so the index is cross-referenced. I am quite amazed that they have automated this whole process. There are research reports available at the RI web site which describe how they went about the task.
Reseach Index is the wave of the future, I believe. It exploits key properties of the web -- global, open access of documents in standard formats (mostly PostScript and PDF), and therefore susceptible to automatic analysis.
Incidentally, there are papers I (Robert Dodier) have written which are indexed by Research Index. I didn't do anything other than put my papers on the web (or contribute to a conference which put its proceedings on the web). I notice that I've been cited exactly once -- O happy day!
Slashdot Mirror
Sales of the book don't matter to Wolfam -- Mathematica is MUCH more lucrative. The book is just an advert for his software.
Believe it or not there are open source computer algebra systems! Two that I'm familiar with are Maxima and PARI-GP (sorry, don't have a link at the moment). Maxima is general purpose, while PARI-GP is mostly about number theory.
Maxima's history is interesting. It is based on the source code (Lisp!) of the Macsyma system developed at MIT circa 1970-1980. Mathematica is essentially a rewrite of Macsyma with very slightly different syntax. You know what they say about imitation.
You moron. People like you are so very dangerous.
If Pinochet was such a great guy, let him win the next election. If it was really such a mess, the voters would be only too happy to give Allende the boot.
Or was the CIA just speeding up the process?
(1) Remarkably inhumane, compared to the means by which Allende got himself into the presidency.
(2) Military governments have a such a great track record in economic reform; 3000 souls is really not much to pay for such expertise.
(3) In really civilized countries, they don't need to kill anyone to implement economic reforms.
No way, dude. The original poster who said "A century ago, mathematics was primarily a new field" was way off base, and the follow up isn't any closer. Sophmore engineering students are pretty amazing, I know -- check out those concrete canoes! -- but their math curriculum encompasses about one percent of the math available a century ago.
The last person who might possibly have mastered the whole of mathematics as it existed in his era was Henri Poincare'. Incidentally, he did much of his most memorable work just about 100 years ago. The suggestion that today's undergrads might have a comprehension comparable to his, is just silly.
According to this website this bit about dying is a paraphrase of "Think for yourselves and let others enjoy the privilege to do so too" which Voltaire did write. I dunno; Voltaire didn't say anything about himself or anybody else dying in the name of free speech.
Aside from Voltaire's lack of interest in self-sacrifice, the misquote is just too clumsy for him. He was consistently witty and well-polished, although I admit the above quote ("Think for yourselves...") doesn't really show his writing skills at their best.
In short, the misquote is certainly NOT the sort of thing Voltaire would write. Thomas Paine, perhaps.
Especially if you're doing policy analysis (i.e., you're not data mining a 10 million record database) R is a great environment to work in. Hmm, it might be possible to process 10 million record databases with R, I just haven't tried it.
R has more or less equivalent functionality to Octave, but the programming language (S) is more interesting, and plotting functions are more sophisticated. For matrix operations, both R and Octave use BLAS.
My advice is to stay away from SAS and SPSS. These two were invented back when "everybody knows using a computer is hard". Both have GUI interfaces now, but that's just an attempt to hide the essential ugliness of their command languages. If you must use a commercial product then S+ is my recommendation.
Good luck & have fun!
My nomination for the Beautiful Fortran Award would have to be ARPACK. It is clear, concise, elegant, efficient, and TOTALLY ROCKS. Eigenvalue problems are not easy, which only increases the beauty of any solution.
Well, let's not use the term AI -- it has too much baggage and nobody knows what it means anyway.
There are three aspects to making workable self-fixable photocopiers and water plants: (1) figure out how such a system should figure out what's happening and how it should figure out what to do about it, (2) implement a system that carries out such figuring, and (3) encode knowledge about a specific domain so that the system can crunch on it.
Problem (1) is solved, AFAICT. The right thing to do is to use probability to represent beliefs about the world, and to use utility theory to express how happy you are with the state of the world. The combination of those two concepts is conventionally called decision theory.
Problem (2) is very difficult. Having decided you want to build a general decision theoretic system, you immediately get into serious computational difficulties. Here's the tip of the iceberg: if there is more than one chain of events that lead from causes to effects, it's a hard problem. Time-dependent problems (i.e., most real problems) have this characteristic left and right.
Problem (3) is difficult in a practical sense as opposed to a theoretical sense. This is basic engineering -- translating a real problem into a mathematical representation suitable for solution.
The payoff of this approach is that you can put together the model of the system from the bottom up -- each little component can be modeled separately, and then the rules of decision theory tell you how to combine all the pieces. Essentially, you construct a system that is an "automatic heuristic generator". Lights out? Check the battery? Battery is good? Check the wires... etc. Such a system can generate an appropriate response in many more scenarios than human heuristic-generators.
auai.org is a great resource for information on automated reasoning systems. I attempted some research in this direction -- see riso.sourceforge.net; all the results are in my dissertation.
As you know, MS has deep pockets, and they've used some of the money extorted from W95 lusers to fund serious research in machine learning. In my own research field (belief networks, shameless plug) they have hired several of the big wheels in the field. Patrice Simard used to work in neural networks; don't know what he's been up to lately -- you can probably find papers by him at Research Index.
Recognition of handwritten characters is a hard, interesting problem; pattern recognition on bitmap images can only take you so far, and to go beyond that you need to start incorporating additional information such as stroke order and letter and word context. Voice recognition is even more difficult.
The bottom line is that since there exist computers (namely our brains) that can reliably carry out text and voice recognition, we know it's possible. Getting there will require solving several substantial engineering problems, and it's possible a university department somewhere will carry through the solution, or perhaps a corporation that can afford to have a group dedicated to technology that won't pay off in the next three months... such as MS.
On the web, see: Assoc. for Uncertainty in Artificial Intelligence -- this is the primary conference devoted to belief networks, which are a class of graphical (in the circles and arrows sense) Bayesian probability models. There are tutorials and other papers on the main AUAI web page, and links to the last several years of conference proceedings. By the way, Heckerman and Horvitz, now doing belief networkish work at MS Research, are in the AUAI crowd.
In print, my favorite reference is E.T. Jaynes, "Probability Theory: The Logic of Science", which is due out soon. See this web site devoted to Jaynes' work for the status. I am also fond of Castillo, Gutierrez, & Hadi, "Expert Systems and Probabilistic Network Models".
There are a vast (well, maybe just large) number of alternative models to classify things; a good introduction is Hastie, Tibshirani, & Friedman, "Elements of Statistical Learning". Incidentally, they use spam classification to illustrate several kinds of models.
Finally, if you're wondering what the heck is the difference between Bayesian probability and any other kind -- just google the posts in sci.stat.math; there is a Bayesian vs frequentist flame war about once a year. :^)
I wonder how cool ideas at Google fare against the need to get into some serious code optimization, scaling problems, etc. By "cool idea" I mean something like the page rank algorithm. Does Google spend a lot of time coming up with more cool ideas, or do you already have plenty of ideas, and you need to invest more in developing the existing ideas? How many developers are there for every ideas person -- 1 to 1, 10 to 1, 100 to 1 ??
I do quite a bit of number crunching. Here are
some of the resources I use:
Netlib (www.netlib.org) -- Yes, it's mostly Fortran, but that's a good thing! Just use f2c (easy to find) and translate to C if that's what you want. Don't underestimate the power of decades-old programs -- old == widely used and well-tested.
StatLib (lib.stat.cmu.edu) -- Collection of statistical software, in various languages, including C, Fortran, and S.
SAL, Scientific Applications on Linux (sal.kachinatech.com) -- a very large collection of links.
Freshmeat (www.freshmeat.net) -- Not scientifically oriented, but there is much scientific stuff there, along with all kinds of miscellany.
Octave (www.octave.org) -- A package for matrix manipulations, similar to Matlab, but free. Useful for all kinds of problems.
R (www.r-project.org) -- An implementation of the S language for statistics, but also useful for general problems, similar to Octave. S+ is a commercial implementation of S.
Well, that ought to be enough to get started. To echo something other posters have mentioned -- don't even bother with Windows software. If your budget is tight, save your money for hardware, don't waste it on the MS tax.
Research Index (previously called Citeseer) is truly amazing. The RI collection is built by a web crawler, which finds papers published on the web, and -- get this -- automatically extracts bibliographic info such as the title, author, abstract, and citations. For each paper in the collection, RI also finds citations of that paper, so the index is cross-referenced. I am quite amazed that they have automated this whole process. There are research reports available at the RI web site which describe how they went about the task.
Reseach Index is the wave of the future, I believe. It exploits key properties of the web -- global, open access of documents in standard formats (mostly PostScript and PDF), and therefore susceptible to automatic analysis.
Incidentally, there are papers I (Robert Dodier) have written which are indexed by Research Index. I didn't do anything other than put my papers on the web (or contribute to a conference which put its proceedings on the web). I notice that I've been cited exactly once -- O happy day!