Главная
Study mode:
on
1
Welcome
2
Introducing the speaker
3
What are tricks and tips?
4
Differentiation with(out) a difference
5
V-shape curve is a result of floating-point evaluation (cancelation) errors dominating truncation errors
6
"Automatic differentiation "
7
Complex step method
8
Example: derivative of atan(x)/(1 + e^(-x^2)) at x = 2
9
Computing principal logarithm in a complex plane, a multi-valued function
10
Computing the principle logarithm in the 1960s
11
Logarithm of the product of numbers, complex case
12
Arcsin and Arccos in complex plane
13
Unwinding number
14
Roundtrip relations
15
Accurate difference
16
Low rank updated of n x n real matrix A
17
Why Sherman-Morrison formula holds?
18
World's Most Fundamental Matrix Equation
19
Computing a product
20
Matrix chain multiplication problem (MCMP)
21
Chain rule of differentiation and MCMP
22
Randomization
23
1985 IEEE Standard 754 and it 2008 Revision
24
Model for rounding errors analysis
25
This model is weaker than what IEEE Standard actually says
26
Model vs correctly rounded result
27
Prevision versus accuracy
28
Accuracy is not limited by the precision
29
Photocopying errors
30
Typing errors
31
Low precision arithmetic
32
Applications of half-precision (fp16, floating point 16 bits)
33
Error analysis in low precision arithmetic
34
What you can do to reduce error in fp16?
35
Can we obtain more information bounds?
36
Conclusions
37
Q&A: how to avoid the case when randomization makes the problem worse?
38
Q&A: how to choose between methods like contour integral and higher precision arithmetic?
39
Q&A: does half-precision allow a brute force analysis of the distribution of operations?
40
Q&A: can you comment on low precision and power consumption?
Description:
Dive into a comprehensive 56-minute keynote from JuliaCon 2018 featuring Nick Higham, Royal Society Research Professor and Richardson Professor of Applied Mathematics at the University of Manchester. Explore a wealth of tricks and tips in numerical computing, covering topics such as differentiation techniques, complex plane computations, matrix operations, and low precision arithmetic. Learn about the complex step method, unwinding numbers, roundtrip relations, and the matrix chain multiplication problem. Gain insights into IEEE Standard 754, rounding error analysis, and the applications of half-precision arithmetic. The talk concludes with a Q&A session addressing randomization, method selection, and the relationship between low precision and power consumption.

Tricks and Tips in Numerical Computing - Keynote

The Julia Programming Language
Add to list
#Computer Science #Scientific Computing #Numerical Computing #Mathematics #Algebra #Linear Algebra #Matrix Multiplication #Machine Learning #Automatic Differentiation
00:00
-00:32
1 of 40

1 Welcome