|
By now, yoda concluded better 'Thoughts of the Day' would be posted.
Perhaps you should sythe down and think this through.
| "The difference between genius and stupidity is that genius has its limits." - Albert Einstein | | "As far as we know, our computer has never had an undetected error." - Weisert | | "If you are searching for perfection in others, then you seek disappointment. If you are seek perfection in yourself, then you will find failure." - Balboos HaGadol Mar 2010 |
|
|
|
|
|
|
His pc was broken due to the force update to windows 10
|
|
|
|
|
(waves hand)
These are not the documents you are looking for.
If you have an important point to make, don't try to be subtle or clever. Use a pile driver. Hit the point once. Then come back and hit it again. Then hit it a third time - a tremendous whack.
--Winston Churchill
|
|
|
|
|
In C#, Math.Sign(0) is 0. OK, that's one way to solve that.
But I was amused by this comment in SO regarding why there isn't a Sign function in Python:
Indeed there was a patch which included sign() in math, but it wasn't accepted, because they didn't agree on what it should return in all the edge cases (+/-0, +/-nan, etc)
Instead, they created copysign
math.copysign(x, y)
Return x with the sign of y. On a platform that supports signed zeros, copysign(1.0, -0.0) returns -1.0.
So if you do:
math.copysign(15, -313)
you get -15.0.
Or more amusingly:
math.copysign(0, -313)
Answer: -0.0
Then try this:
-0.0 < 0
Answer: False
-0.0 == 0
Answer: True
Anyways, I found that weird / interesting. How can 0 be negative?
Thoughts, on this Monday morning?
Marc
|
|
|
|
|
I'd like to know how they represent -0 internally...
I mean, either they're cutting one number off the max range of the numeric type (e.g. -127->127 instead of -128->127, so 11111111 could represent -0 instead of -1), and changing all of the low-level arithmetic to compensate (unlikely), or...
They're wasting a whole byte on the sign, just so they can represent something that, 99.999% of the time, doesn't matter...
If they're going to go that route, I think they should figure out 254 more ways to represent zero, just so they're not wasting bits 
|
|
|
|
|
In Python, copysign returns a floating point value, so they use the sign bit. There is no integer representation of -0. If you try a = -0, then print a shows 0.
Marc
|
|
|
|
|
Ian Shlasko wrote: I'd like to know how they represent -0 internally...
That's exactly where is the problem. I don't know C# (I'm currently a javascript programmer), but a Number is internally stored as IEEE754 floating point, which have a bit for sign. It is not an integer. Zero can be represented without the sign bit, therefore when sign bit is 0, you get +0.0, when it is 1, you get -0.0. It is just a bit interpretation problem.
If you are interesting why -0.0 work as it is, there is very good mathematical theory for infinitesimals. But as a rule of a thumb, never use == on floating-point number. It returns false more often than you expect to.
|
|
|
|
|
I don't see a problem?
-0 == -1*0 == 0
it ain’t broke, it doesn’t have enough features yet.
modified 20-Oct-19 21:02pm.
|
|
|
|
|
But in Python (note the floating point):
-1.0 * 0.0
-0.0
Marc
|
|
|
|
|
And in C
printf("%f", -1.0 * 0.0)
prints -0.000000
see Signed zero[^] (Wikipedia)
Edit:
should have read the rest of the thread before hitting post .. 
|
|
|
|
|
Not only Python.
Try it in SQL Server 2012:
declare @A as float;
declare @B as float;
declare @C as float;
set @A = 0.0;
set @B = -1.0;
set @C = @A * @B;
print @C;
or PHP:
$A = -1.0 * 0.0;
echo $A . "\n";
|
|
|
|
|
The Taurus would be the closest sign.
|
|
|
|
|
Actually, I believe the correct answer is Gemini! (I know I married them)
|
|
|
|
|
|
OriginalGriff wrote: Being annoyingly boring:
An interesting read actually. Starts to make sense. Interestingly, in C#:
double a = -0.0;
a is "0". C# makes no distinction between positive and negative 0.
Marc
|
|
|
|
|
But for which reason?
Is it the parser in the compiler?
Is it the output routine you used? (The VS debugger?)
So I tried the following:
double a = -0.0;
Console.WriteLine("a = {0}, 1.0/a = {1}", a, 1.0 / a);
double b = double.Parse("-0.0");
Console.WriteLine("b = {0}, 1.0/b = {1}", b, 1.0 / b);
byte[] abytes = BitConverter.GetBytes(a);
Console.WriteLine("a bytes=" + string.Join(",", abytes.Select(z => z.ToString("X2"))));
byte[] bbytes = BitConverter.GetBytes(b);
Console.WriteLine("b bytes=" + string.Join(",", bbytes.Select(z => z.ToString("X2"))));
The output was surprising:
a = 0, 1.0/a = -Infinity
b = 0, 1.0/b = Infinity
a bytes=00,00,00,00,00,00,00,80
b bytes=00,00,00,00,00,00,00,00
So the compiler handles the -0.0 correctly, but double.Parse doesn't and the output is converting the negative 0.0 to "0.0" with no sign.
"Fairy tales do not tell children the dragons exist. Children already know that dragons exist. Fairy tales tell children the dragons can be killed."
- G.K. Chesterton
|
|
|
|
|
Another trick for C#: use the Ceiling function.
static void Main(string[] args)
{
double a = Math.Ceiling(-0.1);
double b = 1 * 0;
double c = -1 * 0;
Console.WriteLine("a = Math.Ceiling(-0.1), a=" + a.ToString());
byte[] bytes;
bytes = BitConverter.GetBytes(a);
Console.WriteLine("Bytes of a:");
Console.WriteLine(BitConverter.ToString(bytes));
Console.WriteLine("b = 1 * 0, b=" + b.ToString());
bytes = BitConverter.GetBytes(b);
Console.WriteLine("Bytes of b:");
Console.WriteLine(BitConverter.ToString(bytes));
Console.WriteLine("c = -1 * 0, c=" + c.ToString());
bytes = BitConverter.GetBytes(c);
Console.WriteLine("Bytes of c:");
Console.WriteLine(BitConverter.ToString(bytes));
}
Output:
a = Math.Ceiling(-0.1), a=0
Bytes of a:
00-00-00-00-00-00-00-80
b = 1 * 0, b=0
Bytes of b:
00-00-00-00-00-00-00-00
c = -1 * 0, c=0
Bytes of c:
00-00-00-00-00-00-00-00
I wrote about this some years ago for our SAS programming users (What's the difference between 0 and -0?[^]), as a similar IEEE 754 behavior exists there.
|
|
|
|
|
The flaw in this is that b and c are both calculated as integer values (at compile time) then converted to double. Integer doesn't have a negative zero.
"Fairy tales do not tell children the dragons exist. Children already know that dragons exist. Fairy tales tell children the dragons can be killed."
- G.K. Chesterton
|
|
|
|
|
So they decided to look at C/C++ and implement the copysign - C++ Reference[^] but not the signbit - C++ Reference[^]?
That is indeed half-baked.
Marc Clifton wrote: Anyways, I found that weird / interesting. How can 0 be negative?
This can occur when an operation underflows. If you have for example two very small number where one is negative and multiply them, the result can underflow and will be set to zero but the sign bit is preserved.
|
|
|
|
|
|
Zero means nothing to me...
|
|
|
|
|
Marc Clifton wrote: How can 0 be negative?
Well it certainly isn't positive when it's the amount in your glass or your bank account!
I am not a number. I am a ... no, wait!
|
|
|
|
|
Dear Mr. Clifton,
Please stop belittling me.
Signed,
0
|
|
|
|
|
Sander Rossel wrote: Please stop belittling me.
Dear Mr. Zero (or is it Mrs. Zero, there is some ambiguity here!),
I'm sorry, I had no intention of being n-aught-y. In fact, you are very important! Without you, nothing could not be expressed mathematically. You fulfill a central role in mathematics as the additive identity of the integers, real numbers, and many other algebraic structures.
And besides, you are at least 3756 years old, and in ancient Egypt you were given a designation that means "beautiful", which is much better than what the Babylonian's did, which was to represent you with a space, a blank, a nothing!
Of course, the Greeks weren't sure about you -- how could nothing be something!
Well, I know that you are something indeed, and certainly, based on your age, much wiser than I.
Your humble servant,
Marc
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
