[1]: https://github.com/jostylr/Reals-as-Oracles/blob/main/articl...
816 karma · joined February 15, 2012
[1]: https://github.com/jostylr/Reals-as-Oracles/blob/main/articl...
Similarly with decimals and Cauchy sequences, what is lurking around to make those useful is an interval. If I tell you the sequence consists of a trillion approximations to pi, to within 10^-20 precision, but I do not tell you anything about the tail of the sequence, then one has no information. The next term could easily be -10000. It is having that criterion about all the rest of the terms being within epsilon that matters and that, fundamentally, is an interval notion.
The older and longer paper of Defining Real Numbers as Oracles contains some exploration of these ideas in terms of continued fractions. In section 6, I explore the use of mediants to compute continued fractions, as inspired by the old paper Continued Fractions without Tears ( https://www.jstor.org/stable/2689627 ). I also explore a bit of Bill Gosper's arithmetic in Section 7.9.2. In there, I square the square root of 2 and the procedure, as far as I can tell, never settles down to give a result as you seem to indicate in another comment.
For fun, I am hoping to implement a version of some of these ideas in Julia at some point. I am glad to see a version in Python and I will no doubt draw inspiration from it and look forward to using it as a check on my work.
Many worlds gets around this since experiments do not have definite results in that theory. Instead, the experimenter splits into multiple copies, each of which thinks there is a result of the experiment, but that is an illusion.
There are about 4 million people in an age cohort below, say, 80 (80 * 4 million = 320 million). Let's say a generation is 20 years. So for Gen X to retire all at this comfortable level, that would be 4 million * 2 million * 20 = 160 trillion dollars in assets. This is staggered, but still.
The total US stock market value was recently about 55 trillion [1] and 110 trillion for global stock market [2]
These numbers seem a bit off from each other. Looking at the chart for the US growth, the value doubled from 1998 to 2014, then doubled again from 2014 to now. The later bit is presumably due to the massive injection of money (inflation) from the past 5 years which all seems supported by looking at the massive jump in value in 2020 in the chart.
And here is some analysis about GDP to total market value [3]. Also not a promising conclusion. Has market value at 60trillion at the moment and GDP at 30 trillion.
1: https://siblisresearch.com/data/us-stock-market-value/ 2: https://www.visualcapitalist.com/the-109-trillion-global-sto... 3: https://www.gurufocus.com/stock-market-valuations.php
In particular, it is quite possible to prove a theorem that a sequence is Cauchy, but that there is no way to explicitly figure out N for a given epsilon. The sequence is effectively useless. This presumably is possible, and common, with using the Axiom of Choice. One can even imagine an algorithm for such a sequence that produces numbers and eventually converges, but the convergence is not knowable. Again, if this is just approximating something, then we can simply say it is a useless approximation scheme. But defining real numbers as the equivalence class of Cauchy sequences suggests taking such a sequence seriously in some sense and is the answer.
In contrast, consider integer and rational number versions, it is quite immediate how to reduce them to their canonical form, assuming unlimited finite arithmetic ability. For example, 200/300 ~ 2/3 and one recognizes that 200/300 and 2/3 are different forms of what we take to be the same object for most of our purposes. There is no canonical Cauchy sequence to reduce to and concluding two sequences are equivalent could take a potentially infinite number of computations /comparisons. While that is somewhat inherent to the complexity of real numbers, it feels particularly acute when it is something that must be done in defining the object.
Dedekind cuts have the opposite problem. There is only one of them for an irrational number, but it is not entirely clear what we would be computing out as an approximation, particularly if the lower cut viewpoint is adopted.
Intervals, on the other hand, inherently contain the approximation information. By dividing them and picking out the next subinterval, one also has a method to computing out a sequence of ever better approximations. I suppose one could prove the existence of the family of containment intervals without explicitly being able to produce them, but at least the emptiness of the statement would be quite clear (nothing is produced) in contrast to the sequences that could produce essentially meaningless numbers for an arbitrarily large number of terms.
1681: eight years later was glorious revolution with a bill of rights, marking individual freedoms
1764: ten years later, beginning of American Revolution and being free of monarchs
1849: ten years-ish later, start of US civil war; was the time of an attempt by the British to end slavery around the world
1936: ten years later, colonial empires were being dismantled, UN established to attempt global cooperation, US in the ascendancy with a seed of ties being established more by economics than military force, great economic upswing lifting people out of poverty (60% in poverty then, 10% now) while the global population blossoms
2035: Majority of the global population in middle class or better, triumph of individuals over technocrats, bureaucrats, and corporatists :)
More or less, one can think of a Cauchy sequence of generating intervals that contain the real number, but it can be arbitrarily long before the sequence gets to "small" intervals. So comparing two Cauchy sequences could be quite difficult. Contrast that with the rational numbers where a/b ~ c/d if and only if ad = bc. This is a relatively simple thing to check if a, b, c, and d are comfortably within the realm of human computation.
Dedekind cuts avoid this as there is just one object and it is assumed to be completed. This is unrealistic in general though the n-roots are wonderful examples to think it is all okay and explicit. But if one considers e, it becomes clear that one has to do an approximation to get bounds on what is in the lower cut. The (lower) Dedekind cut can be thought of as being the set of lower endpoints of intervals that contain the real number.
My preference is to define real numbers as the set of inclusive rational intervals that contain the real number. That is a bit circular, of course, so one has to come up with properties that say when a set of intervals satisfies being a real number. The key property is based on the idea behind the intermediate value theorem, namely, given an interval containing the real number, any number in the interval divides the interval in two pieces, one which is in the set and the other is not (if the number chosen "is" the real number, then both pieces are in the set).
There is a version of this idea which is theoretically complete and uses Dedekind cuts to establish its correctness[1] and there is a version of this idea which uses what I call oracles that gets into the practical messiness of not being able to fully present a real number in practice[2].
1: https://github.com/jostylr/Reals-as-Oracles/blob/main/articl... 2: https://github.com/jostylr/Reals-as-Oracles/blob/main/articl...
The $10 a burger price increase would be a stupid prediction unless the worker just sells one burger an hour and nothing else. If an employee's share of the selling amounted to, say, 10 burgers an hour on average, then a $1 price increase on burgers would counter the wage raise.
US government had about a trillion in deficits before the pandemic then spiked up to about 3 trillion in each of 2020 and 2021 [1]. That is a bump of 4 trillion dollars suddenly over those two years. The past two years seems to be about 1.5 trillion each and the inflation rate has decreased.
The theory is that printing money to spend leads to raising prices. Somebody gets that excess money so they can buy more of whatever they need, maybe start new projects or whatever and this bids up prices. Since this is not based on removing the ability for others to buy what they want (no tax increase) then the overall demand goes up. If you also have supply restrictions while spiking demand, it is natural for prices to rise dramatically.
This creates wage rising demands which, after much agony, more or less gets everything back at the same actual price levels though usually with a range of incompleted projects and a decent chunk of people impoverished while some got very rich. The particulars depend on whatever sector was inappropriately stimulated in the start of the process.
The Austrian school of economics is a good place to learn more.
Taxes are an explicit mechanism to say who the government is taking money from in order to give money to whomever they are giving it to. Deficit spending is an implicit way of taking money as determined by the market which means those with the least power are likely to lose the money.
The pandemic had about a trillion going directly to people. The rest went elsewhere.
There is also the issue of this is US spending versus global spending. Not sure what other countries did and also unsure how much the US dollar being the main global reserve currency factors into this.
1: https://fiscaldata.treasury.gov/americas-finance-guide/natio...
In ideal capitalism, the basic bargain is increased wealth for all at the cost of there being losers and extreme winners along with failures of endeavors being an essential ingredient to overall societal success. It is based on choices being made by individuals to maximize what they value given the options at hand.
What is the bargain in ideal socialism? Historically, it looks like equality for the non-elites being pushed over wealth creation. Is there a different trade-off being proposed. There are, after all, no solutions, just trade-offs.
In the context of this thread, what would the options in the computer market be in an ideal socialist society? Would there be many choices? Who would make those decisions? Who would make the products and why them? How would one determine how to distribute the product? Why would innovation happen, particularly how would one idea triumph over a differing one, particularly over a status quo idea that pushes out entrenched interests?
Would dissent be allowed? How would it be addressed?
I would also be interested in an honest analysis of environmental stewardship questions of socialist versus capitalist. From what I have read, environmental destruction has been way worse under government management (Soviets were notorious, but also the US military and its private contractors) than private management unless that private management is somehow being directed and/or protected by the government. Is there a reason to believe socialists would be good stewards of the environment? What is the incentive structure of the people in the system to make that happen? Under a true capitalist system, it is not getting sued for destroying other people's properties as well as the ability to profit off of well-managed lands.
Also, under anarcho-capitalism, voluntary associations that can be categorized as socialism/communism can easily arise and would be perfectly acceptable, but the reverse is not true. Why would socialist voluntary associations not be sufficient, i.e., why would enforced collectivization be necessary and morally correct? In a certain sense, anarcho-capitalism is what emerges when the right of secession is taken to the individual level while global socialism is explicitly disallowing any secession. Why is uniform collectivization of all humans under a single controlling entity a good thing?
Again, I am hoping for a reference accepted by pro-socialists that thoughtfully addresses these questions.
It would be interesting if those studying biological evolution could see how much of their techniques, theories, and predictive abilities could apply in this realm.
Government in the US pays 60% of the medical costs in the US. State licensing applies to all levels of healthcare, not just doctors. Doing minor in home care requires some kind of nursing credential with a supervising doctor-type.
The common experience of patients is that they are interacting with private health systems and insurance companies. These systems are grotesquely obscure with prices and it feels like one is being fed to the wolves of capitalism unlike in almost any other market experience in the US. This makes it feel like it is unregulated. But, of course, this can only happen because the government has, by force of law and gun, created a medical cartel, limiting options, prices, and divergent practices.
As for why the month comes first, I get the feeling we Americans care about precision up to a month and less so about the numerical day such as "When does school start?" with an answer of "School starts in September". Obviously people need the precise date if you actually have a student going, but in a lot of conversation, people are just looking for the month. I can imagine that being even more true in earlier times.
We also peg things to months such as our elections being the second Tuesday of November or Labor Day being the first Monday of September. Note that in those phrases we are putting the day kind of first, but not in a particular numerical way. The month is what sticks out as a rough guesstimate of the time period.
One old post from him is a detailed plan for paying for it: https://www.scottsantens.com/how-to-reform-welfare-and-taxes... It's not a plan I support, but it is not an unreasonable set of notions.
For me, I would view UBI as a way to transition off of bureaucratic control of people's lives. The simplest in my mind would be a flat 10% income tax for it which would then be divided up equally across to all citizens, leading to about $6,000 (20 trillion economy, 10% is 2 trillion, divided across 330 million) a year and a break even at about $60,000 in income. While $6,000 may not sound like a lot, if one were to share living costs with three other people, then that is $24,000 in cash and there are plenty of places in this country (I live in a metro that is such a place) where one could live on that amount, potentially revitalizing less crowded places. When you factor in the fact that one can work without losing it, unlike welfare, then it becomes plausible to see it as a stepping stone to being more productive without being too much of a hindrance. If there is to be a social safety net, this seems the least damaging.
The concept of UBI is not a clear proposal without the funding mechanism. Many proponents unfortunately talk about the paying out without figuring out how to pay in. The recent attempt with the child tax credit increase, for example, was part of a time period of massive deficit spending and led to inflation, undercutting much of the benefit. MMT + UBI would be a feel good path to economic devastation.
A related idea would be to also eliminate public schools and convert that already property taxed money to direct payments to families of children. In my city, we spend about 16k per student. A quarter of that is for special needs so setting that aside, we would have 12k for the typical child. Funding children instead of bureaucrats. Grouping 20 children would lead to 240k which could easily fund 4 full-time teachers / staff for 40k each with 80k leftover for space costs. Compare this to 30 students per classroom in a decrepit buildings as is the case in my school district with documented, horrible failed learning outcomes.
The ultimate goal for me is the elimination of the government, but getting from here to there in a stable fashion is difficult. My thought is that simple, direct cash payments is a way to eliminate the bureaucrats, try to heal the harms caused by a century of economic government meddling, and eventually lead to a populace that doesn't think it needs a nanny state.
The first books, of any kind, my daughter read were Harry Potter books. She went from not reading at age 8, to reading chats with her friends at age 9 (pandemic time), to having read the entire 7 book series by the end of age 10. No one taught her how to read. Writing, using devices, arithmetic, she largely acquired them all on her own as she matured.
While I am biased, she seems well-adjusted and quite bright. I have no doubt that she is capable of mastering whatever she would like to master.
Note that the article was looking at a cohort study. It was not based on randomly assigning "pleasure reading in early childhood" which means cause and effect cannot be ascertained. My guess is that if they were to measure lifelong Sudbury students they would find well-developed cortical areas regardless of early childhood reading or not. The development probably relates to constructing complex understanding of the world (or fictional worlds) which independent play as well as reading can do, but conventional teaching cannot.
For those who are curious, a simple algorithm for generating these good approximations and continued fractions is to take the mediant of a Farey pair that enclose the given number, such as [3/1, 4/1], here 7/2, and then choose the sub-interval that contains the desired number. Repeat as desired. This first step has [3/1, 7/2] as the next interval to look at as 3/1 < 3.213432 < 7/2. When doing this to generate the continued fraction representation, start with the nominal interval [0/1, 1/0]. A few steps: 0/1 : 1/1 : 1/0 R 1/1 : 2/1 : 1/0 R 2/1 : 3/1 : 1/0 R 3/1 : 4/1 : 1/0 L 3/1 : 7/2 : 4/1 L 3/1 : 10/3 : 7/2 L 3/1 : 13/4 : 10/3 L 3/1 : 16/5 : 13/4 R 16/5 : 29/9 : 13/4 L 16/5 : 45/14 :29/9 L 16/5 : 61/19 : 45/14 R 61/19 : 106/33 : 45/14 R ... The number 106/33 is 3.2121... and is generated by easy computation of the mediants.
Each time the direction of selection is changed, this results in having an ultra-best approximation. If we count the number of steps in between the switches, we get the continued fraction representation: [3; 4, 1, 2, ...].
This is also related to the Stern-Brocot tree, for those interested.
Playing around with these ideas led me to the idea of defining a real number as a number that, given a rational interval, says Yes or No depending on whether it ought to be in that interval. To define the space of such objects, I used properties that the Yes/No answers have to satisfy. The key property is that if you split a Yes interval into two pieces, one of the pieces should be Yes and the other No (unless the splitting point is the real number).
If curious for details, my paper repository on this matter is at https://github.com/jostylr/Reals-as-Oracles where I not only prove that it all works out as it should, but also give an example of doing continued fraction arithmetic in section 7.9 of the main paper for those who just love continued fractions.
There are a number of real numbers that one can define which can depend on whether we can prove something or not. It may turn out that we can never prove it and the number is never resolved.
My attitude, which may not be satisfactory, is that we do what we can and we should have a language/framework for facilitating that. I think the oracle approach highlights what we know and marks what we can't compute clearly. I call it the resolution of the oracle. I don't want known precision to be lacking just because of a poor definition of what a real number is.
As for the example being algebraic, it is a particularly nice example and it is the same example in every example of a Dedekind cut. Another example of such a rule, one which is not algebraic, would be whether pi is in an interval or not. Given a<b, one rule could be that it is No if b < 3 or a > 4 or sin(a)sin(b) > 0. Alternatively, it would say Yes if a >= 3 and b<= 4 and sin(a)sin(b) <= 0. To compute this, one needs to compute sine of a and b sufficiently precisely to determine their sign.
The flavor I am trying to convey is having a definition convey a useful goal. The interval approach says that we are trying to generate precision about our inaccuracy. I think this is something which would greatly benefit those learning about real numbers. Most of the time, the error is presented as secondary and an annoyance, the concerns of the error propagation in further computations is pushed to the side, and it is all relegated to experts or computers. The expansionary nature of arithmetic in error propagation is pushed to a usually unsatisfactory discussion about significant digits.
My goal here is to change the mental framework so that these concerns come to the forefront. Ideally, they also come with useful tools to handle the uncertainties such as ways to compute how narrow the input intervals need to be when doing a computation. And maybe, just maybe, students could become more comfortable with fractions.
If I want to compute pi + e, an infinitely accurate version is pi + e. That's about it. So what we are actually looking for in this computation is an estimation algorithm, one which can be made as accurate as we wish, but finitely so. The natural way to express this is with rational intervals as rationals are precise and intervals give a containment of the numbers.
For arithmetic, we can have a mechanism for figuring out how precise the input approximations need to be in order to get a given precision for the final computation. The perspective presented here naturally leads to that as a matter of defining and establishing the arithmetic of oracles.
As for playing around with physical representations of irrational numbers, keep in mind that there is no way to prove that, say, something that looks like a unit square to us is really a perfect square down to infinite precision. And without that, we can easily have that the unit square is only very close to such a figure, but is actually a rational rectangle with a rational diagonal that very closely approximates the square root of 2.
The lazy evaluation is that defining the real number is about having a way of answering the question when presented with the rational interval, but one does not actually need to have the answers until asked.
From this perspective, a program that computes Yes or No when given a rational interval via the rule "a<b Yes iff a^2< 2 < b^2" can be said to be the square root of 2. For the usual presentations of the other definitions, they cannot be embodied in a computer. One cannot literally have all the infinite elements of a Cauchy sequence, or the infinite sequence of digits in a decimal representation, or the uncountably many elements of a Dedekind cut, represented in a computer memory. One can have a function in memory.
The other definitions can also be presented, in various ways, as functions as well, but I think, fundamentally, what we want in practice from a real number is an interval of small enough size to be of use in whatever we are doing with the real number. That is what the oracles facilitate.
Dedekind cuts can similarly by recast as a function that gives -1,0, or 1 if a number is below, the same, or above the real number. This would be a great improvement over the usual presentation.
But beyond that difference, consider the solution to f(x) = 0 in terms of the intermediate value theorem for an increasing function. The Dedekind cut solution is A = {x | f(x) < 0}. That is the answer. From the perspective of Dedekind cuts, there is nothing further to compute and it has provided nothing to compute with.
The oracle answer would be the rule that says yes to a<b if f(a)*f(b) <=0. This gives a direct mechanism for testing intervals and the idea of narrowing intervals by picking a point inside a Yes interval is very natural from this perspective. I do not see that suggestion popping out of the Dedekind cut approach.
The rational interval approach specifically is targeting narrowing solutions to the answer with error bounds built into it. If I tell you that 3 is in the set A, you have no intrinsic sense of how good of an approximation that is. But if I tell you that the target is in the interval 3 to 3.1, then you have a sense of the precision. Oracles promote having accuracy as foundational.
This is what motivated me to come up with a new definition of real numbers, namely, they are objects (I call them oracles) that answer Yes or No when asked if the number ought to be between two given rational numbers. Abstracting out what properties such an object should have, one can come up with a space of these oracles, define an arithmetic, and prove that they satisfy the axioms of real numbers.
For details: https://github.com/jostylr/Reals-as-Oracles/
In many ways, this is giving a definitional support to the use of interval analysis which is, of course, a very practical concern. It also brings our some cool stuff about mediants and continued fractions (nothing new about that, but nicely motivated).
It also fits in with the adjacent post about busy beaver numbers and its conclusion about knowing a number is in an interval.
which tells you what is happening right there. So if you eliminate the data outside the component, as one should, I think it becomes a very nice readable interface.
Avoiding Ultraviolet Divergence by Means of Interior-Boundary Conditions https://arxiv.org/abs/1506.00497 This is perhaps the first of the papers and so may be a good place to start.
Bohmian Trajectories for Hamiltonians with Interior-Boundary Condition https://arxiv.org/abs/1809.10235 This is the Bohmian part of the story.
Multi-Time Wave Functions https://arxiv.org/abs/1702.05282 This explains the trickiness of having interactions with multi-time wave functions (space-time suggests having multi-time wave functions and no single time).
Creation Rate of Dirac Particles at a Point Source https://arxiv.org/abs/2211.16606 This seems to suggest that there is a kind of spiral approach from/to the point of creation/annihilation.
---
The authors have done a variety of papers on this. A key phrase they use is Interior-Boundary Conditions.
They also released a book based on a course covering some of these ideas: Multi-time Wave Functions, An Introduction https://link.springer.com/book/10.1007/978-3-030-60691-6
1) Bohmian mechanics seems to require some kind of simultaneity. Various proposals have been put forth for making natural foliations, possibly using the wave function to do so, that allow one to evaluate the positions of all the particles at a given time so as to know which configuration point to use in obtaining the velocity of the particle. This is possible, but it does not feel philosophically satisfactory yet. GRWf does not require a foliation to be relativistic which is a nice feature.
2) Quantum field theory naturally models particle creation and annihilation. Most QFTs are presented in a mathematically incoherent way. While conclusions can be made via renormalization, etc., driving an actual dynamics is tricky from that stuff. By not doing perturbations but rather defining the operators by taking into account the shifting of the probability from one sector of n particles to another of n+1 particles, QFTs can actually be made to make mathematical sense. This has been accomplished in some of the simpler models. It is not a problem whatsoever to define a Bohmian evolution where particles appear and disappear in a probabilistic fashion. The difficulty is purely in having a properly defined wave function evolution and that is on its way to being solved.
3) One needs to define wave functions, their evolution, and the particle evolutions in a curved space-time. This is not a problem whatsoever. It is very easy to translate and interpret what we need into differential geometric language. QM has issues with the usual observables/operators translating (such as a momentum operator), but since these are derived concepts in Bohmian mechanics, no fundamental difficulty arises.
4) In general relativity, the mass distribution is part of the evolution of the space-time metric. The mass is based on where the particles are. To date, the wave function tells the particles what to do, but the particles do not have any impact on the wave function. The wave function is impacted by the space-time metric. Also, there are some suggestions that mass might be entirely a part of the wave function and not associated with the particle, i.e., the particles are really just undecorated points moving about.
Basically, gravity and the wave function need to work it out and the particles will then be guided by both as the space-time metric is what takes the gradient of the wave function or, for Dirac style motion, something analogous to a square root of the metric is floating around. Bohmian mechanics does have the advantage that it just has to be concerned with the evolution of the particle lines and not having to figure out how to define various observables.
In other words, a known space-time metric interfaces just fine with Bohmian mechanics, but figuring out how to evolve the space-time metric is still an open question.
At the current time, I am not aware of any novel ideas coming from a Bohmian or GRWf point of view towards resolving the problem of gravity.
If interested in a good discussion of all these things and much, much more, I highly recommend the recent book Foundations of Quantum Mechanics by Roderich Tumulka which covers what the title says, but also discuss Bohmian mechanics and other interpretations.