OpenEuphoria

1. bigatom in The Archive

• Posted by cargoan Dec 21, 2014
• 1902 views

Now available in The Archive:

http://rapideuphoria.com/bigatom.zip

... And if it is useful, use it, modify it, adapt it ... whatever you want.

See you soon. And thank you all. Yes to all, everyone.

Sometimes it's hard to continue, you take away the desire ... but I did not despair.

Greetings

cargoan

3. Re: bigatom in The Archive

• Posted by cargoan Dec 22, 2014
• 1746 views

Also I have trigonometric functions (sin, cos, arctan), but I can not work well with my bigatoms. If someone who knows more than I want to review and perhaps find and fix.

This is what I have:

 
 
-- ##################################### 
-- ###   FUNCIONES TRIGONOMÉTRICAS   ### 
-- ##################################### 
-- 
 
-- sin(x) = x - x^3/3! + x^5/5! - x^7/7! ...  
override function sin(atom x) 
   integer n = 3 
   atom curr = x, res = x 
   atom nfact = 6, x2 = -x * x 
   atom xpow = x2 * x 
    
   while curr do 
      curr   = xpow / nfact 
      res   += curr 
      n     += 2 
      nfact *= n * n - n 
      xpow  *= x2 
   end while 
 
   return res 
end function 
 
 
export function ba_sin(object x) 
 
   if not bigatom(x) then 
      if atom(x) then 
         return ba_new(sin(x)) 
      end if 
      x = ba_new(x) 
   end if 
 
   integer n = 1 
   sequence curr, prev, res 
   sequence nfact, mx2 
   sequence mx, xpow 
 
   sequence sc = scale(SCALE + 2, 0) 
    
   nfact = {1, 0, {6}} 
   mx   = x 
   mx[SIGN] = -mx[SIGN] 
   res  = ZERO 
   prev = ZERO 
   curr = x 
   mx2   = ba_multiply(mx, x) 
   xpow =  x   -- ba_multiply(x, mx2) 
   while not eu:equal(prev, curr) do 
--   while curr[SIGN] do 
      prev  = curr 
      res   = ba_add(res, curr) 
      n    += 2 
      nfact = ba_sub(ba_multiply(n, n), n) 
      xpow  = ba_multiply(xpow, mx2) 
      curr  = ba_divide(xpow, nfact) 
   end while 
   scale(sc) 
   res[DIGITS] = remove(res[DIGITS], 
                          res[EXPONENT] + 2 + SCALE, 
                          length(res[DIGITS])) 
 
   return normalize(res) 
end function 
 
 
-- cos(x) = 1 - x^2/2! + x^4/4! - x^6/6! + x^8/8! - ...  
override function cos(atom x) 
   integer n = 2 
   atom curr = 1, res = x 
   atom nfact = 2, x2 = -x * x 
   atom xpow = x2 
    
   while curr do 
      curr = xpow / nfact 
      res   += curr 
      n     += 2 
      nfact *= n * n - n 
      xpow  *= x2 
   end while 
 
   return res 
end function 
 
 
-- sacada de bc 
-- Usa la fórmula: 
--   atan(x) = atan(c) + atan((x-c)/(1+xc)) para c pequeño (0.2 en este caso) 
-- Por debajo de 0.2, usa la serie: 
--   atan(x) = x - x^3/3 + x^5/5 - x^7/7 + ...  
--  
override function arctan(atom x) 
   integer sg = 1, mult = 0, n = 3 
   atom  curr, res, xpow, x2   
   atom ap2 = 0.0 
 
   if x < 0 then 
     sg = -1 
     x  = -x 
   end if 
    
   if x > 0.2 then 
      ap2 = arctan(0.2)  
   end if 
    
   -- hacer x <= 0.2 
   while x > 0.2 do 
      mult += 1 
      x = (x - 0.2) / (1 + x * 0.2) 
   end while 
    
   curr  = x 
   xpow = x 
   res  = x 
   x2   = -x * x 
   while curr do 
      xpow *= x2 
      curr  = xpow / n 
      res  += curr 
      n    += 2 
   end while 
 
   return sg * (res + mult * ap2) 
end function 
 
 
export function ba_arctan(object x) 
 
   if not bigatom(x) then 
      if atom(x) then 
         return ba_new(arctan(x)) 
      end if 
      x = ba_new(x) 
   end if 
    
   sequence sc = scale(SCALE + (SCALE + 1)/ 2, 0) 
    
   integer sg = 1, mult = 0, n = 3 
   sequence  curr, res, xpow, x2   
   sequence ap2 = ZERO, p2 = {1,-1,{2}} 
 
   if x[SIGN] < 0 then 
     sg = -1 
     x[SIGN] = -x[SIGN] 
   end if 
    
   if eu:compare(x, p2) = SG_PLUS then 
      ap2 = ba_arctan(0.2)  
   end if 
 
   -- hacer x <= 0.2 
   sequence a, b 
   while eu:compare(x, p2) = SG_PLUS do 
      mult += 1 
    --  x = (x - 0.2) / (1 + x * 0.2) 
      a = ba_sub(x, p2) 
      b = ba_add(ONE, ba_multiply(x, p2)) 
      x = ba_divide(a, b) 
   end while 
    
   sequence limit = {x[SIGN], -(SCALE + 1), {1}}   -- ba_power({x[SIGN],1,{1}}, -(SCALE + 6))  
--   sequence prev = ZERO 
   curr  = x 
   xpow = x 
   res  = x 
   x2   = ba_multiply(x, x) 
   x2[SIGN] = -x2[SIGN] 
--   while eu:compare(curr, prev) do      -- esto es interminable 
--   while curr[SIGN] do                  -- como esto, es lo mismo 
   while eu:compare(curr, limit) > 0 do   -- esto no es correcto, sale antes de tiempo 
--      prev = curr 
      xpow = ba_multiply(x, x2) 
      curr = ba_divide(xpow, n) 
      res  = ba_add(res, curr) 
      n   += 2 
      limit[SIGN] = -limit[SIGN] 
   end while 
   scale(sc) 
   res = ba_add(res, ba_multiply(mult, ap2)) 
 
   return ba_multiply(res, sg) 
end function 
-- 

Atoms seem to go well with, but not bigatom. I do not know more, maybe later when I learn more ... but it would be very good to include sin, cos and arctan to complete the library. (arctan is a very important function)

Greets.

cargoan

5. Re: bigatom in The Archive

• Posted by cargoan Dec 22, 2014
• 1693 views

Hola _tom.

Agradezco mucho que te hayas molestado tanto, yo no habría sido capaz.

Si os interesa, yo no quiero nada, ni siquiera reconocimiento alguno, muy al contrario. Soy yo el nunca podré agradecer a los desarrolladores de Euphoria su gran trabajo y enorme esfuerzo para que podamos disfrutar de esta maravilla, llamada Euphoria.

Yo conocí Euphoria en un cd hace mucho, había un pequeño archivo zip llamado euphor14.zip, creo, en ms-dos claro. Y ponía: Lenguaje de programación sencillo y potente, me llamó la atención y probé, imprimí el manual, aunque estuviese en inglés, le eché una ojeada y... esto es lo que busco, yo no soy programador y el basic y otros lenguajes interpretados eran muy lentos si querías exprimirlos un poco y C... bueno C me gusta, pero antes era mucho más sencillo, ahora entre typedefs, macros, extensiones y... seguro que es mejor y más seguro pero para un simple aficionado al final ya no sé ni lo que estaba viendo, se me hace grande.

Como no me gustaba windows, siempre usaba msdos y Dos Navigator, y me pasé a OS/2 y estuve tanteando con derivados de pascal (Modula y Oberon), al final me pasé a linux y en cuanto ví que tenía en el repositorio una versión de 64 bits cloné el repositorio... y aquí estamos en mí Archlinux64 (liquorix+BFS) y con el repo mercurial en mi home.

Muchísimas gracias y hasta pronto.

cargoan

English:

Hello _tom.

I appreciate you've bothered you so much, I would not have been able.

If you are interested, I want nothing, not even any recognition, quite the contrary. Is I can never thank the developers Euphoria their hard work and great effort for us to enjoy this wonder called Euphoria. Euphoria

I met a cd long ago, there was a small zip file called euphor14.zip, I think, in ms-dos clear. And read: language simple and powerful programming caught my attention and I tried, I printed the manual, even if it was in English, I took a look and ... this is what I want, I'm not a programmer and basic and other languages interpreted were very slow if you wanted to squeeze a little and C ... good C I like, but it was much easier before, now between typedefs, macros, extensions and ... it sure is better and safer but for an amateur to end do not even know what I was seeing, it makes me great.

As I did not like windows, always wore msdos and Dos Navigator, and I switched to OS / 2 and was groping derived from pascal (Modula and Oberon), at the end I switched to linux and as soon as I saw that I had in the repository a 64-bit I cloned the repository ... and we're here with me Archlinux64 (liquorix+BFS) and the mercurial repo in my home.

Thank you and goodbye.

cargoan

7. Re: bigatom in The Archive

• Posted by cargoan Dec 23, 2014
• 1635 views

Revisiting code I found an wrong? declaration (object) when it is an integer and when I added shortcuts for ten powers I forgot cut the mantissa in ba_root(). Effect is that you can receive more decimal places than the current scale. It's not an important bug but is not correct.

Updated in the Euphoria archive.

Greets.

Note: and added trigonometric functions, commented out, at end of file... if anyone is encouraged and solves it, I do not get more, convergence of arctan is very painful with these divisors in the McLaurin-Taylor serie. I had to watch a few videos of classes on the matter but ... nothing. Well, nothing not, I have learned enough, never time wasted if you learn something... and as singing, yesterday spoiled, the great Joe Cocker: "with a little help from my friends..."

9. Re: bigatom in The Archive

• Posted by cargoan Dec 27, 2014
• 1376 views
• Last edited Dec 28, 2014

Improved euler() function for speed.

Checked up to one million decimals.

┌──────────────────────────────────────────────────────[sábado, 27 de diciembre 22:45 CET] 
├─[cargoan en ~cargoan/Documentos/Programación/Euphoria/bigatom-a1] 
└─[bash 4.3]──[$]: euc -D TEST3 bigatom.e 
Build directory: build-739556/ 
Translating code, pass: 1 2 3 4 5 6 7 8 9 10 11  generating 
Compiling with GCC 
Compiling  12% init-.c 
Compiling  50% bigatom.c 
Compiling  75% main-.c 
Linking 100% ../bigatom 
┌──────────────────────────────────────────────────────[sábado, 27 de diciembre 22:59 CET] 
├─[cargoan en ~cargoan/Documentos/Programación/Euphoria/bigatom-a1] 
└─[bash 4.3]──[$]: ./bigatom  
'e' con 1000000 decimales en: 1105.750 seg. 
┌─────────────────────────────────────────────────────[domingo, 28 de diciembre 10:08 CET] 
├─[cargoan en ~cargoan/Documentos/Programación/Euphoria/bigatom-a1] 
└─[bash 4.3]──[$]: grep "model name" /proc/cpuinfo | sort -u | sed 's/.*: //'; cat /sys/de 
vices/system/cpu/cpu0/cpufreq/{scaling_governor,cpuinfo_max_freq}; uname -a; journalctl -b 
 | grep -i smpboot | grep -i mips | sed 's/.*:*: //'; zgrep ^CONFIG.*_BFS= /proc/config.gz 
AMD Athlon(tm) Dual Core Processor 4450e 
ondemand 
2300000 
Linux HP-s3733es 3.18.1-1-lqx #1 ZEN SMP PREEMPT Fri Dec 26 20:47:47 CET 2014 x86_64 GNU/L 
inux 
Total of 2 processors activated (9200.18 BogoMIPS) 
CONFIG_SCHED_BFS=y 

Updated in the Archive.

Thanks.

11. Re: bigatom in The Archive

• Posted by cargoan Dec 28, 2014
• 1440 views

I know the existence of unittest but only that, there it is... but it's something I have not reviewed yet ... and nor was it one of my priorities. I'm not a programmer, amateur only, then ... ...I'll have to discuss that topic.

I do not know any English, and yes is Translator but do not know English but Castilian is horrible most of the time, or is incomprehensible or gives a totally opposite to the translation.

is because as the Italians say: "traduttore, tradittore" (translator, traitor).

Greets.

Note: I checked with: http://apod.nasa.gov/htmltest/gifcity/e.1mil (Says a million but spare twenty)

[$]: diff e.1mil e-1millon  
12501c12501 
< 818883747115156239682713 
--- 
> 8188 
 

13. Re: bigatom in The Archive

• Posted by cargoan Dec 29, 2014
• 1320 views

a million and a half good

[$]: ./bigatom  
'e' with 1500000 decimal places in: 2419.070 secs. 
[$]: tail -n3 e-1millon5  
74825944780370926454043853803943744332806781271297919806419520177237478193466293 
67171721860429794706831810947471135277309638919039832745908266244429554005170557 
6339 
 
... 5277309638 9190398327 4590826624 4429554005 1705576339 : euler(1_500_000) 
... 5277309638 9190398327 4590826624 4429554005 1705576339 : 1,500,000 (numberworld) 
 

...looks like it goes well