Questions tagged [floating-point]
Approximate representation of numbers as a fixed number of digits multiplied by a logarithmic scale.
230
questions
0
votes
1
answer
34
views
Absolute difference between largest IEEE754 number and its predecesor
In simple precision format, the largest possible positive number is
$A = 0 ~~~ 11111110 ~~~ 111\ldots 111$
Its predecessor is
$B = 0 ~~~ 11111110 ~~~ 111 \ldots 110$
But what is the absolute ...
2
votes
1
answer
51
views
Comparison of different algorithms for summing floating point numbers
I am exploring several approaches to summing floating point values, such as:
Naive summation, for comparison
Summing sorted values
summing with numpy, again for comparison
Kahan's algorithm
Pairwise ...
1
vote
2
answers
191
views
Convert a rational number to a floating-point number exactly
We have two integers, $n$ and $d$. They are coprime (the only positive integer that is a divisor of both of them is $1$). They may be implemented as something that fits in a machine register, or they ...
2
votes
2
answers
175
views
Multiplication of subnormal and normal numbers under IEEE 754
As far as I understand, when doing this operation, we first need to identify the subnormal value, normalise it, adjust the exponent, and then multiply the significands. Wouldn't it be advantageous to ...
3
votes
2
answers
216
views
Floating-point modular multiplication algorithm
Is there a well-known algorithm for modular multiplication of floating-point numbers?
I would like to multiply some large angle in single precision (6-7 significant digits) and wrap it back to 360 ...
1
vote
1
answer
59
views
FFT of logarithmic input data
Is there a reasonably accurate method of computing an FFT of logarithmically-represented input data (with a sign bit, that is $±2^{\text{double-precision value}}$)?
The naive method (convert to linear ...
1
vote
1
answer
29
views
Floating-point rounding - bit patterns of values that are halfway between two possible results
I am working through the book Computer Systems: A Programmer's Perspective.
The authors explain that round-to-even rounding can be applied for values that are halfway between two possible results. For ...
0
votes
0
answers
242
views
Why XMM register width is 128-bit while double precision floating point is 64-bit?
As far as I know, XMM is used to store floating point. But Highest width floating point according IEEE754 standard is double precision (64-bit). But why XMM register width is 128-bit. Is another 64-...
0
votes
1
answer
39
views
How much larger is the next representable value if 2^59 is stored in a double?
This is an exam question I couldn't solve
If I store 2^59 as double, that would give me
1 * 2^58. Is the answer just 2? I.e. next value is 2^60??
2
votes
0
answers
50
views
Best way to constrain a complex number to being within the unit circle?
What is the best way of implementing the following function,
$$f(x) = \frac{x}{\max(1, |x|)},$$
where $x$ is complex, using a Cartesian representation of $x$ with IEEE 754 floating point ...
3
votes
0
answers
316
views
What are the use cases for the IEEE 754 inexact flag?
The IEEE 754 standard for floating point numbers defines a flag that is set when a result from floating point calculation isn't exact, i.e. has to be rounded. What algorithms are there that utilize ...
4
votes
3
answers
2k
views
Is significand same as mantissa in IEEE754?
I'm trying to understand IEEE 754 floating point. when I try convert 0.3 from decimal to binary with online calculator, it said the significand value was ...
0
votes
1
answer
38
views
How does CPU determine Reserved Exponent cases?
Using IEEE 754 algorithm i assume, that it can be implemented in a branchless way.
But how does CPU determine special cases (Reserved Exponent values):
Exponent
Significand
is
11111111
000000000...
...
1
vote
1
answer
102
views
What does it mean unambiguously that a number is value 0 up to numerical precision?
I was reading that a quantity $x$ is $0$ upt to numerical precision. What does this statement formally mean -- especially in the context of numerical methods or real computers.
I looked up in google ...
1
vote
0
answers
32
views
Standard for representing a float scaled to a particular range?
TL;DR Is there a "standard" way to represent a float scaled to a particular range, such that we get maximum precision for the given bit depth, within that range?
I'll start with my general ...
0
votes
2
answers
70
views
(Numerical Analysis) What is the largest double float represented for the gamma function and $n!$
Consider that
\begin{align}
\Gamma(n+1) = n!
\end{align}
for any integers. I then got the following two questions:
What is the largest value of $n$ for which $Γ(n+1)$ and $n!$ can be exactly ...
3
votes
1
answer
82
views
Bisecting Intervals of floating point numbers containing 0 and infinity fairly
It is seldom considered that floating points are not evenly distributed in the real number line. I've been working with interval arithmetic and noticed when bisecting $[a,b]$ on the real number line ...
1
vote
0
answers
247
views
Can Radix Sort be modified for signed ints and/or floats?
A few months ago I learned about the magic that allows radix sort to run in O(n) time and space. Most tutorials on radix sort say it is useful for very large ...
0
votes
1
answer
50
views
IEEE 754 conversion
I'm trying to convert 3.2 into IEEE 754 format. We find that $(3)_2=11$ and we also find that
$0.2*2=0.4 -0$
$0.4*2=0.8 -0$
$0.8*2=1.6 -1$
$0.6*2=1.2 -1$
and this cycle repeats so $.2=00110011...$
...
4
votes
0
answers
106
views
Uniformly random decimal numbers
Due to finite precision of number representations, we face situations like:
In: 0.1+0.1+0.1==0.3
Out: False
(on my ...
1
vote
0
answers
45
views
(Branchless) Bitonic Sorting Network for a Set of Floating Point Numbers
In the past I've implemented a branchless Bitonic Sorting Network on a gpu using CUDA, for integers.
I am facing a related problem:
In my Order Independent Transparency implementation, I would like to ...
0
votes
1
answer
57
views
How can vector angle comparison between lattice points be done without using floating-points? (Convex Hull)
Let's say I have a point $(x_0, y_0)$, and some other points $(x_1, y_1), (x_2, y_2) ... (x_n, y_n)$, such that all of them are lattice points; all have integer coordinates. Let's further assume that ...
1
vote
2
answers
730
views
How many Integers can be represent in Double-Precision floating-point form
How to calculate the number of Integers that can be represent in Double-Precision floating-point form?
1
vote
1
answer
312
views
Prove every number in double precision 32-bit floating-point format can be represented in 64-bit format
Theorem: Prove every number in double precision 32-bit floating-point format can be represented in double precision 64-bit floating point-format.
64-bit format:
Attempt: Let $ b = b_0 ,...,b_{31} $ ...
1
vote
1
answer
45
views
Why does floating point become less accurate as the powers of 2 increase?
https://fabiensanglard.net/floating_point_visually_explained/
I was reading this article where the exponent and the mantissa are explained as the window and offset respectively. As the gap between ...
0
votes
0
answers
123
views
Is there a way to convert FLOPS to bit operation per second
My problem is the following: I have $N$ inner products to compute in parallel every second.
Each of the vectors in those inner product is composed of $7$ bits.
I want to know for which $N$ it starts ...
2
votes
1
answer
203
views
Unit conversion - Better to divide by an integer or multiply by a double?
I currently have a long timestamp measured in units of 100ns elapsed since January 1st, 1900. I need to convert it to milliseconds.
I have the choice of either ...
1
vote
2
answers
875
views
Half precision floating point question -- smallest non-zero number
There's a floating point question that popped up and I'm confused about the solution. It states that
IEEE 754-2008 introduces half precision, which is a binary
floating-point representation that uses ...
1
vote
1
answer
483
views
Floating Point Arithmetic with 3 bits mantissa
Find all values of $ x ∈ R $ such that x + 1 = 1 in floating point arithmetic with 3 bits mantissa.
How do we represent number 1 in floating point arithmetic with 3 bits mantissa I wonder? After that, ...
3
votes
2
answers
172
views
Python versus Matlab on the quantity 1/0
Python and Matlab seem to disagree on the division by 0.
Python:
...
1
vote
3
answers
4k
views
Negative Numbers in 32 bit Floating Point IEEE Numbers
So I understand the logic behind converting positive decimal numbers to IEEE 32 bit floating numbers but I'm not completely sure behind the negative one's. If for example we have a decimal number say -...
1
vote
2
answers
336
views
Adding two numbers in base 2(floating point) vs Multiplying two numbers in base 2(floating point)
Is it true that adding two numbers in base 2 is more complex than multiplying them? If so can someone please explain why this is the case?
1
vote
2
answers
149
views
Prove that $1^\text{nan} = 1.00$
I know that for most computation involve nan (not a number) the result is a nan itself except for some cases.
For example, $1^{\text{nan}} = 1.00$ which proven by mathematicians to be true.
I tried to ...
2
votes
2
answers
27
views
Floating point bitwise comparator. If f1 and f2 are floating point numbers with the following properties can we always say f1 > f2?
Recall floating-point representation:
Suppose $f$ is a floating-point number then we can express f as,
If $f$ is normal:
$$(-1)^{s}\cdot2^{e-127}(1 + \sum\limits_{k=1}^{23} b_{23-k}\cdot 2^{-k})$$
If $...
0
votes
0
answers
103
views
Convert $8.75×10^{6}$ to IEEE-32 format?
There is a similar question already asked on this site but does not have an answer as to how the 10x was converted into 2y. I know how to convert 8.75 or 875 into IEEE representation. But what about ...
1
vote
1
answer
465
views
What is the machine epsilon and number of mantissa bits for TI-83?
I am trying to determine how many bits the TI-83 Plus uses to store floating point numbers. I am using the algorithm for approximating the machine epsilon given in "Numerical Mathematics and ...
1
vote
0
answers
51
views
How close are current computer technologies in terms of energy efficiency to the Landauer Limit?
I'm trying to figure out how close (in orders of magnitude) current computer technologies are in terms of energy efficiency to the Landauer Limit. However, I'm finding (seemingly) conflicting ...
1
vote
3
answers
143
views
How to determine the set of real numbers corresponding to a given floating point number?
Let's say we consider IEEE 754 double precision floating-point numbers, and we use RNTE - Round To Nearest, Ties to Even - rounding.
I know that the RNTE rounding works this way:
given two consecutive ...
4
votes
1
answer
570
views
Logic behind choosing the exponent bias as $2^7 -1$ instead of $2^7$ in $32$ bits IEEE-754 floating point representation
The $\text{IEEE-754}$ uses $32$ bits to represent single precision floating point numbers. The partitions of the register are as follows:
...
0
votes
0
answers
38
views
What is the currently know fastest algorithms to convert each individual Java primitive and Number types to a Char Array/String
What is the currently know fastest algorithms to convert each individual Java primitive and Number types to a Char Array/String number representation while ...
2
votes
1
answer
8k
views
How to represent zero as floating point number?
For the floating-point number, we have the form
$\pm d_0.d_1d_2···d_{P-1}\times\beta^E$
$\pm$ --------------------------- sign
$d_0.d_1d_2···d_{p-1}$ --------- significant
$\beta$ --------------------...
0
votes
0
answers
41
views
Are floating-point numbers normalised by computers or humans?
I keep seeing posts and articles about how to normalise a floating-point number, why that's done and how such a number is represented in binary. But no one seems to mention who/what the normalisation ...
-1
votes
3
answers
650
views
How are bitwise operators used in normalisation of floating-point numbers?
After having spent a significant amount of time googling this topic, I am still struggling to confidently answer the following question:
...
1
vote
1
answer
953
views
Converting Decimal Numbers between 0 and 1 to Binary
I've been playing around with a program I wrote that converts decimal numbers to binary numbers and i've noticed that eventually, after applying the algorithm (multiply by 2, subtract 1 if greater ...
1
vote
2
answers
49
views
Floating-point oblivious way to compute multiset numbers
I have to compute $R = \left(\!\!{n + 1\choose k}\!\!\right)$, which happens to be:
$$
R = \left(\!\!{n+1\choose k }\!\!\right) = \binom{n+k}{k} = \frac{(n + k)!}{n!k!} = \frac{(n+1)(n+2)\cdots(n+k)}{...
2
votes
2
answers
83
views
Validity of Algorithm for Testing Two Floating Point Numbers
This question is related to the epsilon- (or delta- if you prefer) test for floating point equality. But my question is not how to do it. Instead I have a related algorithm for testing equality, and I ...
0
votes
0
answers
43
views
New way of representing floating point
I want to create a new way of representing floating point. In standard IEEE floating point, we have 1 bit to represent sign, 8 bits to represent exponential and 23 bits to represent significand. In ...
0
votes
2
answers
217
views
IEEE 754 addition wrong result floating point numbers
I want to add two IEEE 754 numbers.
I followed the steps to add two 754 numbers. However the result it not correct.
Number 1:
S:0
E:01111111
M:11111111111111111111111
Number 2:
S:0
E:01111111
M:...
0
votes
2
answers
80
views
Cancellation of inequalities in floating point arithmetic
In finite precision floating point arithmetic the associative property of addition is not satisfied. This is, it is not always the case that $$(a+b)+c=a+(b+c)$$
Even $a=(a+b)-b$ is not always true.
To ...
0
votes
1
answer
60
views
How does value decompression work for Facebook's Gorilla in the case where count of leading zeroes is not stored
I am referring to this paper: http://www.vldb.org/pvldb/vol8/p1816-teller.pdf
My question is regarding section 4.1.2 where it says:
When XOR is non-zero, calculate the number of leading and ...