4 ms·

> Because to them it's obvious. And to them it's wrong! Much is said in this book which is difficult (but not impossible) to rigourously justify. It took centu

by taejo 10y ago

> Because to them it's obvious. And to them it's wrong! Much is said in this book which is difficult (but not impossible) to rigourously justify. It took centuries for calculus to be placed on a rigourous mathematical foundation; this foundation (called "real analysis", largely developed in the 19th century) is quite different from the intuitive ideas presented in this book. The presentation here (in particular the idea that dx² is negligible) can be made rigourous through "non-standard analysis", but this 20th century development is less-known (in fact, unknown to many mathematicians) and perhaps even more difficult than real analysis. Rigour is not necessary to understanding, and can even be fatal to understanding, but it's how mathematicians work, and generally mathematics is taught by mathematicians.

jacquesm 10y ago
> in fact, unknown to many mathematicians > and generally mathematics is taught by mathematicians So that leaves the door wide-open to a whole army of mathematicians to who it is unknown that are teachers. And those are the ones for who all of the above does not apply and who still treat their 'entry level problems' as obvious to all their students even when the evidence strongly suggests that to the students it is not at all obvious. And I'm writing this as a high school kid who was pretty good in math but totally lost interest due to such teachers (that, and computers were far more interesting).
sundarurfriend 10y ago
> this foundation (called "real analysis", largely developed in the 19th century) is quite different from the intuitive ideas presented in this book. ... (in particular the idea that dx² is negligible) Thank you for posting this! Such things always bothered me in high school, seemed like approximations that ought to bite you in the behind at least in some corner cases. Another example from TLA: > dy = 2cos(θ + 1/2 dθ) · sin 1/2 dθ > But if we regard dθ as indefinitely small, then in the limit we may neglect 1/2 dθ by comparison with θ, and may also take sin 1/2 dθ as being the same as 1/2 dθ. The equation then becomes: > dy = 2cosθ × 1/2 dθ This again seems like very sloppy and careless kind of approximation that ought to bite you in the back - but knowing there are just (supposed-to-be) intuitive non-rigorous methods, and that these have actual rigorous backing, somehow soothes me.
- thanatropism 10y ago
  But this is true of every level of mathematics. See humorous proofs that 2=3 amounting to manipulations like 0a = 0b, cut the zeros, a=b. This is no fault of "nonrigorous elementary algebra", it's a matter of remembering all notations are abbreviations and you have to know how to manipulate them. You can get away with dy and dx if you just think "I'm not writing integral signs because they're annoying" and remember the rules of working with integrals. This is how stochastic differential equations work -- they're not even differential equations because Brownian motion is not differentiable, they're just notation.
thanatropism 10y ago
"dx^2 is negligible" can be alternatively (rigorously) obtained in standard analysis/calculus in the following sequence: 0) Intermediate value theorem (continuous function with image between b and c must be such that there is x such that f(x)=d for d between b and c. This can be first proved where d=0 -- it's the method of bisection by nested intervals, really -- and then generalized) 0.5) The actual definition of derivatives known to everyone. 1) Rolle's theorem, i.e. the Intermediate Value Theorem on first derivatives (if something goes up and down then it must have a critical pont). 2) Mean Value Theorem (Rolle's theorem but rotated) 3) Taylor's theorem.

by taejo 10y ago

jacquesm 10y ago

> in fact, unknown to many mathematicians > and generally mathematics is taught by mathematicians So that leaves the door wide-open to a whole army of mathematicians to who it is unknown that are teachers. And those are the ones for who all of the above does not apply and who still treat their 'entry level problems' as obvious to all their students even when the evidence strongly suggests that to the students it is not at all obvious. And I'm writing this as a high school kid who was pretty good in math but totally lost interest due to such teachers (that, and computers were far more interesting).

sundarurfriend 10y ago

> this foundation (called "real analysis", largely developed in the 19th century) is quite different from the intuitive ideas presented in this book. ... (in particular the idea that dx² is negligible) Thank you for posting this! Such things always bothered me in high school, seemed like approximations that ought to bite you in the behind at least in some corner cases. Another example from TLA: > dy = 2cos(θ + 1/2 dθ) · sin 1/2 dθ > But if we regard dθ as indefinitely small, then in the limit we may neglect 1/2 dθ by comparison with θ, and may also take sin 1/2 dθ as being the same as 1/2 dθ. The equation then becomes: > dy = 2cosθ × 1/2 dθ This again seems like very sloppy and careless kind of approximation that ought to bite you in the back - but knowing there are just (supposed-to-be) intuitive non-rigorous methods, and that these have actual rigorous backing, somehow soothes me.

thanatropism 10y ago
But this is true of every level of mathematics. See humorous proofs that 2=3 amounting to manipulations like 0a = 0b, cut the zeros, a=b. This is no fault of "nonrigorous elementary algebra", it's a matter of remembering all notations are abbreviations and you have to know how to manipulate them. You can get away with dy and dx if you just think "I'm not writing integral signs because they're annoying" and remember the rules of working with integrals. This is how stochastic differential equations work -- they're not even differential equations because Brownian motion is not differentiable, they're just notation.

thanatropism 10y ago

"dx^2 is negligible" can be alternatively (rigorously) obtained in standard analysis/calculus in the following sequence: 0) Intermediate value theorem (continuous function with image between b and c must be such that there is x such that f(x)=d for d between b and c. This can be first proved where d=0 -- it's the method of bisection by nested intervals, really -- and then generalized) 0.5) The actual definition of derivatives known to everyone. 1) Rolle's theorem, i.e. the Intermediate Value Theorem on first derivatives (if something goes up and down then it must have a critical pont). 2) Mean Value Theorem (Rolle's theorem but rotated) 3) Taylor's theorem.