[HN Gopher] Logarithms yearning to be free
       ___________________________________________________________________
        
       Logarithms yearning to be free
        
       Author : _Microft
       Score  : 68 points
       Date   : 2022-04-13 13:56 UTC (9 hours ago)
        
 (HTM) web link (www.johndcook.com)
 (TXT) w3m dump (www.johndcook.com)
        
       | CyberRabbi wrote:
       | A link between zeroth powers and the logarithm function is such a
       | simple idea but potentially fruitful. Exponentials and logarithms
       | often translate between additive groups and multiplicative
       | groups. Maybe there are interesting isomorphisms where the
       | logarithm can be generalized as a zeroth power.
        
       | ewewr wrote:
       | integral{from t=0, to t=x} t^{-1+0} dx
       | 
       | but you calculate instead
       | 
       | f(x) = lim_{e->0} [ integral{from t=0, to t=x} t^{-1+e} dx ]
       | 
       | that is just
       | 
       | = lim_{e->0} [ (t^e - 1) / e ]
       | 
       | replacing t^e
       | 
       | = lim_{e->0} [ (e^(ln(t) _e) - 1) / e ]
       | 
       | by https://en.wikipedia.org/wiki/L%27H%C3%B4pital%27s_rule
       | 
       | = lim_{e->0} [ ln(t)_e^(ln(t) _e) / 1 ]
       | 
       | that is easy to calculate
       | 
       | = ln(t)_e^(ln(t) _0) / 1
       | 
       | = ln(t)_1 / 1
       | 
       | = ln(t)
       | 
       | So f(t)=ln(t)
       | 
       | reply
        
       | codeflo wrote:
       | If you want to make a statement like that, then logarithms might
       | not be the zeroth power, but the epsilonth power, where e is an
       | "infinitesimal".
       | 
       | Bear with me here, I know infinitesimal math isn't a fully
       | coherent thing. But there's a reason why Newton used it, it
       | sometimes works surprisingly well to make intuitive analogies.
       | Maybe there's a way to make it work here.
       | 
       | For example: The function ln(x) grows strictly slower than x^a
       | for any positive real a, but faster than x^0. Hence, it's x^e,
       | where 0 < e < a for any positive real a.
        
         | gradschool wrote:
         | Where does that leave superpolylogarithmic subexponential
         | functions (sung to the tune from Mary Poppins)?
         | 
         | https://www.csee.umbc.edu/~sherman/Papers/superpoly.ps
        
         | voxl wrote:
         | Infinitesimal math is completely coherent, it just doesn't have
         | one theory that is "thee algebra" like the real numbers do.
        
       | _Microft wrote:
       | You might want to appeal to Alexandra, patron saint of the
       | paywalled scientist, for she might hear you.
        
         | CodesInChaos wrote:
         | Searching for the title of the paper, DDG shows scihub as the
         | first result, while google only has a libgen hit on the second
         | page (and it's the author's list of papers, not the paper
         | itself).
        
       | westurner wrote:
       | > _The author opens with the example of finding the
       | antiderivative of xn. When n [?] -1 the antiderivative is another
       | power function, but when n = -1 it's a logarithm._
       | 
       | What a neat limit. Probably best to leave the powerfn/logfn() as
       | a dumb symbolic symbol until the end (until after later parameter
       | substitution)?
        
         | nh23423fefe wrote:
         | I don't follow. The antiderivative fails because of division by
         | zero. What limit? And what does symbolic manip do?
        
           | gus_massa wrote:
           | The idea is that you want to calculate
           | 
           | integral{from t=0, to t=x} t^{-1+0} dx
           | 
           | but you calculate instead
           | 
           | f(x) = lim_{e->0} [ integral{from t=0, to t=x} t^{-1+e} dx ]
           | 
           | that is just
           | 
           | = lim_{e->0} [ (t^e - 1) / e ]
           | 
           | replacing t^e
           | 
           | = lim_{e->0} [ (e^(ln(t)*e) - 1) / e ]
           | 
           | by https://en.wikipedia.org/wiki/L%27H%C3%B4pital%27s_rule
           | 
           | = lim_{e->0} [ ln(t)*e^(ln(t)*e) / 1 ]
           | 
           | that is easy to calculate
           | 
           | = ln(t)*e^(ln(t)*0) / 1
           | 
           | = ln(t)*1 / 1
           | 
           | = ln(t)
           | 
           | So f(t)=ln(t)
        
             | _Microft wrote:
             | Just for the sake of correctness: the differential has to
             | be dt instead of dx here.
        
               | gus_massa wrote:
               | I agree, thanks. But I saw your comment just now and it's
               | too late to edit. :(
        
             | westurner wrote:
             | So the type of the return value changes at
             | asymptotes/limits (already) and thus that's not a _pure
             | function_ in terms of math. If a [math] function returns a
             | more complex type signature instead of throwing a
             | ZeroDivisionError (as Python core does) what is that then
             | called? Is it differentiable or no, etc?
             | 
             | We throw ZeroDivisionError instead of axiomatically
             | defining a ranking for                 scalar*parameter*inf
             | if x > 0:           2*x*inf > x*inf           # because
             | 2 > 1
             | 
             | But basically every CAS just prematurely throws away all
             | terms next to infinity (by replacing the information in
             | that expression with just infinity)? And nothing yet
             | implements e.g. Conway's _Surreal numbers_ infinities?
             | 
             | Is negative infinity to the infinity greater or lesser than
             | infinity?                 assert (-1*math.inf)**math.inf ==
             | math.inf            assert (-1*sympy.oo)**sympy.oo ==
             | sympy.oo
             | 
             | Here's a dumb Real/Function instead of prematurely
             | discarding information that could be useful:
             | from sympy import symbol       from sympy.abc import x
             | Infinity = symbol('Infinity', real=True)  # *       #
             | from sympy.symbols import Wild
             | 
             | All the axioms just change there.
             | limit(-x**-1)
             | 
             | An uphill battle for certain.
             | 
             | "[Python-ideas] Re: 'Infinity' constant in Python"
             | https://mail.python.org/archives/list/python-
             | ideas@python.or...
        
       ___________________________________________________________________
       (page generated 2022-04-13 23:01 UTC)