Re: doing radix conversions in floats

Bill:

Thanks for the quick right up for hexadecimal numbers. I adapted your 
procedures into working with general radix numbers base 2 to base 16. 
(it could be expanded, say up to base 35 or so)

Here's the code I came up with:

{R2D(N,rdx=16)=
my(m,a,b,s,t,u);
m=Map(["0",0; "1",1; "2",2; "3",3; "4",4; "5",5; "6",6; "7",7; "8",8; 
"9",9; "a",10; "b",11; "c",12; "d",13; "e",14; "f",15]);
[a,b]=strsplit(N,".");
if(rdx==16,
  u=eval(Str("0x",a)) + eval(Str("0x",b))/rdx^#b*bitprecision(1.,4*#b);
  return(u);
);
a=Vecrev(strsplit(a));
b=strsplit(b);
\\ convert integer part
s=0;
for(i=1,#a,s+=mapget(m,a[i])*(rdx^(i-1)));
t=0;
for(i=1,#b,t+=mapget(m,b[i])/(rdx^i));
u=s+t*1.0;
return(u);
}
addhelp(R2D,"R2D(n,{radix=16}) converts radix number into a decimal at 
current bit precision");

{D2R(N,rdx=16)=
my(m,a,b,c,d,A,s,t,u);
m=["0","1","2","3","4","5","6","7","8","9","a","b","c","d","e","f"];
a=truncate(N);
b=N-a;
if(rdx==16,
   s=strprintf("%x.%x",a,b*rdx^((bitprecision(b)+exponent(b))\4));
   return(s);
);
A=digits(a,rdx);
s="";
for(i=1,#A,s=strexpand(s,m[A[i]+1]););
r=ceil(default(realprecision)*log(10)/log(rdx));
t="";
for(i=1,r,
  c=b*rdx;
  d=truncate(c);
  b=c-d;
  t=strexpand(t, m[d+1]);
);
u=strexpand(s, ".",  t);
return(u);
}
addhelp(D2R,"D2R(n,{radix=16}) converts a decimal number into a radix 
number at current bit precision");

This seems to work fairly well.

- Randall


On 4/16/25 01:22, Bill Allombert wrote:
> On Tue, Apr 15, 2025 at 11:42:11PM -0700, American Citizen wrote:
>> Has anyone ever written a radix conversion program for converting decimal
>> floats to other radixes such a base 16?
>>
>> I wrote a dec2hex and hex2dec for integers which seems to work, if I
>> stipulated that the output of the numbers in other bases than 10 is a
>> delimited string "0xnnn".
> In GP you can do
> ? strprintf("0x%x",7123456)
> %89 = "0x6cb200"
> ? eval("0x6cb200")
> %90 = 7123456
>
>> I post the code for both below, but ask if anyone has come up with radix
>> conversions to the right of the decimal point (and in the current working
>> precision of gp-pari) ?
>>
>> For example the square root of 2 to 39 digits is
>> 1.41421356237309504880168872420969807857 but the equivalent base 16
>> representation is 1.6a09e667f3bcc908b2fb1366ea957d3e3adec1751277509a when I
>> make an attempt to have the equivalent hexadecimal number.
> You can start with something like
>
> d2h(x)=my(f=frac(x));strprintf("%x.%x",truncate(x-f),f*16^((bitprecision(f)+exponent(f))\4))
> h2d(s)=my([a,b]=strsplit(s,"."));eval(Str("0x",a))+eval(Str("0x",b))/16^#b*bitprecision(1.,4*#b)
>
> ? d2h(sqrt(2))
> %123 = "1.6a09e667f3bcc908b2fb1366ea957d3"
> ? h2d(%)
> %124 = 1.4142135623730950488016887242096980785
>
> Cheers,
> Bill