Re: Literate Programming, Axiom, and Physically Based Rendering
Despite the outline of my attempts at funding I haven't spent a lot of time on it. Money has never been a motivation. I did get research funding at City College for "indefinite integers" but "the team" simply dissolved, likely due to my lack of leadership skills. I have two areas of focus for Axiom, community and research. One focus is to "make Axiom live", hence the focus on literate programming. I contacted the Physically Based Rendering authors and they gave me a copy of their tools for creating their literate document. Unfortunately, Axiom is an order of magnitude bigger than PBR so it makes no sense to print it out. Instead I've concentrated on things like collecting published papers, adding bibliographic references to the algorithms, improving the documentation, etc. in an attempt to gather and explain the details, internals, and design decisions. I don't have any leadership skills. The community went elsewhere. The other focus is "make Axiom correct", hence the focus on merging proof tech. "Correct" is a specification issue. I now prefer the term "sane", which has synonyms "rational", "coherent", "judicious", and "sound". This particular journey tries to merge the computer algebra and proof technology into "computational mathematics". This is also showing my lack of leadership. The proof community, according to private emails I've received, is actively hostile to the idea of proving programs correct (although they did prove Buchberger's algorithm, for example). The computer algebra community has shown no interest in proving algorithms correct. Nevertheless, my effort at merging the two is making progress. I'm doing research and I'm having fun. Life is good. Tim On Mon, Apr 3, 2023 at 6:32 AM Johann 'Myrkraverk' Oskarsson < joh...@myrkraverk.com> wrote: > Dear Tim, > > On Sun, Apr 2, 2023 at 5:35 PM Tim Daly wrote: > > > > > Moving forward with Axiom is going to be hard, and will probably have > > > to require some new people to get involved. That is best served with > > > financial or other incentives, but I have no suggestions on how to > > > achieve that goal. > > > > I have contacted about a dozen companies for grant money. > > > > I have contacted several grant-generating organizations. > > > > I set up a "contribute to Axiom" account of which I was the only > contributor. > > > > I have contacted NSF and DARPA. > > > > The only support Axiom had was from Gilbert Baumslag at City College > > of New York. I worked with Gilbert in his Infinite Theory Group at CAISS. > > He supported the initial effort to make Axiom run standalone (the NAG > > version required a running Axiom to create a new Axiom). > > > > One of the key issues is managing the funding. That requires someone > > to hold the funds, manage the receipts, and pay the taxes. That is, it > > requires an independent accountant. NSF and DARPA usually give funds > > to a University office to manage. No accountants, no grant money. > > Yeah. I have thought about this for other similar preservation > efforts, and hence I will not personally do much for Axiom, except > perhaps buy an updated tutorial. > > There are options now, like Patreon, SubscribeStar, and things I > haven't even heard of. Afaict, the successful efforts on those lines > are the ones that provide entertainment, though possibly in the form > of "training material." I believe that's how people start on YouTube, > but I'll admit some ignorance about how well that works. > > And once you go that route, your main "job" is to get funding, and to > get that funding you end up "entertaining" your audience. > > Which means you aren't spending your time writing about Axiom internals. > > Now, to refer to something /I/ like in a similar fashion, is the > LibTomMath internals book, available here: > > https://github.com/libtom/libtommath/releases/tag/v1.1.0 > > The file tommath-1.1.0.pdf. > > Note this is the last release of that book, since the book has been > removed from the repository. > > > https://github.com/libtom/libtommath/commit/17ca193fa70890033b35625ca44041c51171b50a > > Still, now that I know about it, I /might/ resurrect it; but I have no > time frame for it. > > The reason I like this book, is that as far as I've read it, it's all > about teaching the reader how to make a multi-precision library. > Although I would probably not make the same technical decisions Tom > did. I could use it as a base for my own work; we'll see. > > > I contacted several companies asking them to set up a small 1 or 2 person > > accounting shop for open source projects to accept and manage grants. > > > > So over the last 20 years no attempt at funding succeeded. > > The open source world has changed /slightly/ to the better in the past > few years, but there's still a lot to do, to get funding for our > efforts. > > I have no concrete suggestions for you, nor do I think it's my place > to do so, even if I had. > > > Over those 20 years I averaged about $3000 per year of
Re: Literate Programming, Axiom, and Physically Based Rendering
Dear Tim, On Sun, Apr 2, 2023 at 5:35 PM Tim Daly wrote: > > > Moving forward with Axiom is going to be hard, and will probably have > > to require some new people to get involved. That is best served with > > financial or other incentives, but I have no suggestions on how to > > achieve that goal. > > I have contacted about a dozen companies for grant money. > > I have contacted several grant-generating organizations. > > I set up a "contribute to Axiom" account of which I was the only contributor. > > I have contacted NSF and DARPA. > > The only support Axiom had was from Gilbert Baumslag at City College > of New York. I worked with Gilbert in his Infinite Theory Group at CAISS. > He supported the initial effort to make Axiom run standalone (the NAG > version required a running Axiom to create a new Axiom). > > One of the key issues is managing the funding. That requires someone > to hold the funds, manage the receipts, and pay the taxes. That is, it > requires an independent accountant. NSF and DARPA usually give funds > to a University office to manage. No accountants, no grant money. Yeah. I have thought about this for other similar preservation efforts, and hence I will not personally do much for Axiom, except perhaps buy an updated tutorial. There are options now, like Patreon, SubscribeStar, and things I haven't even heard of. Afaict, the successful efforts on those lines are the ones that provide entertainment, though possibly in the form of "training material." I believe that's how people start on YouTube, but I'll admit some ignorance about how well that works. And once you go that route, your main "job" is to get funding, and to get that funding you end up "entertaining" your audience. Which means you aren't spending your time writing about Axiom internals. Now, to refer to something /I/ like in a similar fashion, is the LibTomMath internals book, available here: https://github.com/libtom/libtommath/releases/tag/v1.1.0 The file tommath-1.1.0.pdf. Note this is the last release of that book, since the book has been removed from the repository. https://github.com/libtom/libtommath/commit/17ca193fa70890033b35625ca44041c51171b50a Still, now that I know about it, I /might/ resurrect it; but I have no time frame for it. The reason I like this book, is that as far as I've read it, it's all about teaching the reader how to make a multi-precision library. Although I would probably not make the same technical decisions Tom did. I could use it as a base for my own work; we'll see. > I contacted several companies asking them to set up a small 1 or 2 person > accounting shop for open source projects to accept and manage grants. > > So over the last 20 years no attempt at funding succeeded. The open source world has changed /slightly/ to the better in the past few years, but there's still a lot to do, to get funding for our efforts. I have no concrete suggestions for you, nor do I think it's my place to do so, even if I had. > Over those 20 years I averaged about $3000 per year of Axiom related > costs such as equipment, online hosting, travel for presentations, and > Axiom-related swag (e.g. coffee mugs), all out of personal funds. That > figure does not include my time. Yeah. And as I alluded to above, once you start on the funding route, you're basically committed to that instead of writing. That is the reality of it. -- Johann I'm not from the internet, I just work there.
Re: Literate Programming, Axiom, and Physically Based Rendering
> Moving forward with Axiom is going to be hard, and will probably have > to require some new people to get involved. That is best served with > financial or other incentives, but I have no suggestions on how to > achieve that goal. I have contacted about a dozen companies for grant money. I have contacted several grant-generating organizations. I set up a "contribute to Axiom" account of which I was the only contributor. I have contacted NSF and DARPA. The only support Axiom had was from Gilbert Baumslag at City College of New York. I worked with Gilbert in his Infinite Theory Group at CAISS. He supported the initial effort to make Axiom run standalone (the NAG version required a running Axiom to create a new Axiom). One of the key issues is managing the funding. That requires someone to hold the funds, manage the receipts, and pay the taxes. That is, it requires an independent accountant. NSF and DARPA usually give funds to a University office to manage. No accountants, no grant money. I contacted several companies asking them to set up a small 1 or 2 person accounting shop for open source projects to accept and manage grants. So over the last 20 years no attempt at funding succeeded. Over those 20 years I averaged about $3000 per year of Axiom related costs such as equipment, online hosting, travel for presentations, and Axiom-related swag (e.g. coffee mugs), all out of personal funds. That figure does not include my time. Tim On Sun, Apr 2, 2023 at 2:49 AM Johann 'Myrkraverk' Oskarsson < joh...@myrkraverk.com> wrote: > Dear Tim, > > On Sat, Apr 1, 2023 at 10:22 AM Tim Daly wrote: > [...] > > I had hoped that Axiom could have been reborn as a literate program. > > The idea was to "make it live beyond the authors". Each algorithm > > would provide an explanation, various literature references, and some > > examples as well as limitations. That way a person with no background > > knowledge in computer algebra could learn and contribute. Such a > > literate program would have set the standard for mathematical software. > > > > Sadly that's not going to happen. > > My personal involvement with Axiom is probably limited to buying a > manual ages ago. > > I recently learned that the LibTomMath book had been removed from the > source repository on GitHub. [1] > > /Nothing lasts forever/. > > Like Axiom, LibTomMath was a project aimed at teaching people how to > implement multiprecision arithmetic. > > Now, that aspect of it has been removed, apparently since Tom himself > is no longer involved. > > Moving forward with Axiom is going to be hard, and will probably have > to require some new people to get involved. That is best served with > financial or other incentives, but I have no suggestions on how to > achieve that goal. > > Selling physical copies of [incomplete] Axiom internals might help, > and there are other options such as Patreon. > > Good luck, and please don't give up -- even if you take a year or two > off for other things. > > [1] The PDF can still be found in the older releases on the same > GitHub, but no new editions will be made there. > > > -- > Johann > > I'm not from the internet, I just work there. >
Re: Literate Programming, Axiom, and Physically Based Rendering
> So, why exactly it is not going to happen ? Axiom was a research platform for new ideas and new algorithms, originally developed at IBM Research as "Scratchpad". The funding dried up and we had to shut it down. The idea was proposed of selling it but, for various reasons, IBM was not going to sell it as an IBM product. Axiom was removed from IBM Research and sold to the Numerical Algorithms Group (NAG). NAG sold it for a few years and withdrew it from the market. They generously gave me the source code. It took a while to restructure Axiom for open source. As one of the original IBM team I had a lot of knowledge about the program no one else working on the open source had. It was clear that when I stopped working on it the knowledge would be gone. This was already a problem as some of the algebra authors were dead. I thought about ways we could make it possible to capture, structure, and pass on the knowledge. Eventually I settled on Knuth's Literate Programming idea. Textbooks, such as Lisp In Small Pieces, were what attracted me to the idea. Like the Physically Based Rendering book, it should be possible to find any algorithm. Each algorithm would have an explanation (note: NOT documentation), literature references (since many algorithms are PhD research results), proofs, cross references to supporting algorithms, limitations, examples, tests, etc. I gave a talk on this at a conference: https://www.youtube.com/watch?v=Av0PQDVTP4A_channel=NextDayVideo When I introduced the Axiom project I also announced project goals. One of those goals was to restructure Axiom into a literate program. Apparently, like other project goals, this one was ignored. So far I have done the primitive restructuring. See: https://en.wikipedia.org/wiki/Axiom_(computer_algebra_system) These books contain the actual source code which, during the Axiom build process, is "tangled" and compiled. So most of the system is "literate" but only in skeleton form. I've been concentrating on collecting and entering historical documents that only I own as well as getting permission from Scratchpad-related paper authors for re-creating and re-publishing as part of the books. Along with other project goals (e.g. the Computer Algebra Test Suite) http://axiom-developer.org/axiom-website/CATS/index.html no-one seems to want to contribute. At the moment, and for the last few years, I've been working on another project goal. I'm trying to prove Axiom algorithms correct. I spent 6 years attached to CMU as a visiting scholar working on merging proof technology with computer algebra. Like other work I have it partially implemented. The problems are hard and subtle. I gave a talk on this effort at ICMS 2018 http://icms-conference.org/2018/sessions/session14/ This is taking up all of my Axiom-related time and effort so although I really want to expand and improve the literate programs I can't seem to find the time to do it. The potential for proven, explained, and tested computational mathematics that anyone can "pick up, read, and contribute" exists but won't happen. "Literate Axiom" is not going to win a Fields Medal. Tim On Sun, Apr 2, 2023 at 1:30 AM Svjatoslav Agejenko wrote: > Hello Tim ! > > > Idea is wonderful! > > It can also serve as a living math book. Book that captures human > understanding about math. It starts with basic concepts and next > chapters can build upon this. Book can contain executable code > alongisde explanation. So someone learning particular subject can > also run the related code in the book to see it in action and solve > equations symbolically or numerically. > > Math is a timeless thing and is worth preserving in some canonical self > contained way. > > Another interesting finding: This Chat-GPT is very good at teaching and > explanation. Yes, sometimes it lies convincingly, so everything it says > have to be verified. Despite that, most of the time it is correct. It > can also write code in different languages and problem domains. Maybe > it can be good tool to facilitate this effort. > > By the way, there is The Long Now Foundation. They encourage long term > thinking and preservation work. Maybe they can finance this project ? > https://longnow.org/contact/ > > > > > > > So, why exactly it is not going to happen ? > > > Best regards, > Svjatoslav > > > > On Fri, 2023-03-31 at 22:21 -0400, Tim Daly wrote: > > I just received the 4th edition copy of Physically Based > > Rendering[0]. > > Donald Knuth wrote "This book has deservedly won an Academy Award. > > I believe it should also be nominated for a Pulitzer Prize." > > > > It is a 1200 page literate program. It not only contains the actual > > source code, it contains complete explanations of the physical theory > > and the implementation ideas. This textbook is used in a various > > upper-class and graduate programs. I can highly recommend it. > > > > I had hoped that Axiom could have been reborn as a literate program. > > The idea was to "make it
Re: Literate Programming, Axiom, and Physically Based Rendering
Dear Tim, On Sat, Apr 1, 2023 at 10:22 AM Tim Daly wrote: [...] > I had hoped that Axiom could have been reborn as a literate program. > The idea was to "make it live beyond the authors". Each algorithm > would provide an explanation, various literature references, and some > examples as well as limitations. That way a person with no background > knowledge in computer algebra could learn and contribute. Such a > literate program would have set the standard for mathematical software. > > Sadly that's not going to happen. My personal involvement with Axiom is probably limited to buying a manual ages ago. I recently learned that the LibTomMath book had been removed from the source repository on GitHub. [1] /Nothing lasts forever/. Like Axiom, LibTomMath was a project aimed at teaching people how to implement multiprecision arithmetic. Now, that aspect of it has been removed, apparently since Tom himself is no longer involved. Moving forward with Axiom is going to be hard, and will probably have to require some new people to get involved. That is best served with financial or other incentives, but I have no suggestions on how to achieve that goal. Selling physical copies of [incomplete] Axiom internals might help, and there are other options such as Patreon. Good luck, and please don't give up -- even if you take a year or two off for other things. [1] The PDF can still be found in the older releases on the same GitHub, but no new editions will be made there. -- Johann I'm not from the internet, I just work there.
Re: Literate Programming, Axiom, and Physically Based Rendering
Hello Tim ! Idea is wonderful! It can also serve as a living math book. Book that captures human understanding about math. It starts with basic concepts and next chapters can build upon this. Book can contain executable code alongisde explanation. So someone learning particular subject can also run the related code in the book to see it in action and solve equations symbolically or numerically. Math is a timeless thing and is worth preserving in some canonical self contained way. Another interesting finding: This Chat-GPT is very good at teaching and explanation. Yes, sometimes it lies convincingly, so everything it says have to be verified. Despite that, most of the time it is correct. It can also write code in different languages and problem domains. Maybe it can be good tool to facilitate this effort. By the way, there is The Long Now Foundation. They encourage long term thinking and preservation work. Maybe they can finance this project ? https://longnow.org/contact/ So, why exactly it is not going to happen ? Best regards, Svjatoslav On Fri, 2023-03-31 at 22:21 -0400, Tim Daly wrote: > I just received the 4th edition copy of Physically Based > Rendering[0]. > Donald Knuth wrote "This book has deservedly won an Academy Award. > I believe it should also be nominated for a Pulitzer Prize." > > It is a 1200 page literate program. It not only contains the actual > source code, it contains complete explanations of the physical theory > and the implementation ideas. This textbook is used in a various > upper-class and graduate programs. I can highly recommend it. > > I had hoped that Axiom could have been reborn as a literate program. > The idea was to "make it live beyond the authors". Each algorithm > would provide an explanation, various literature references, and some > examples as well as limitations. That way a person with no background > knowledge in computer algebra could learn and contribute. Such a > literate program would have set the standard for mathematical > software. > > Sadly that's not going to happen. > > Tim > > [0] Matt Pharr, Wenzel Jakob, and Greg Humphreys "Physically Based > Rendering" > https://www.pbr-book.org/ > -- Svjatoslav Agejenko WWW: http://svjatoslav.eu
Literate Programming, Axiom, and Physically Based Rendering
I just received the 4th edition copy of Physically Based Rendering[0]. Donald Knuth wrote "This book has deservedly won an Academy Award. I believe it should also be nominated for a Pulitzer Prize." It is a 1200 page literate program. It not only contains the actual source code, it contains complete explanations of the physical theory and the implementation ideas. This textbook is used in a various upper-class and graduate programs. I can highly recommend it. I had hoped that Axiom could have been reborn as a literate program. The idea was to "make it live beyond the authors". Each algorithm would provide an explanation, various literature references, and some examples as well as limitations. That way a person with no background knowledge in computer algebra could learn and contribute. Such a literate program would have set the standard for mathematical software. Sadly that's not going to happen. Tim [0] Matt Pharr, Wenzel Jakob, and Greg Humphreys "Physically Based Rendering" https://www.pbr-book.org/