|
Size: 2506
Comment:
|
Size: 3463
Comment:
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 5: | Line 5: |
| ## <<TableOfContents>> | <<TableOfContents>> == GIAC Factoring == '''People:''' Thomas, Burcin, Richard, William Stein (total anarchy, no leader!) |
| Line 9: | Line 13: |
| '''People:''' Burcin Erocal | '''People:''' '''Burcin''', Erocal, Felix |
| Line 15: | Line 19: |
| '''People:''' Flavia Stan, Karen Kohl, Fredrik Johansson | '''People:''' Flavia Stan, Karen Kohl, '''Fredrik Johansson''', Zaf Add a hypergeometric function class + simplifications |
| Line 19: | Line 25: |
| '''People:''' Burcin Erocal | ---- /!\ '''Edit conflict - other version:''' ---- '''People:''' Burcin Erocal, Simon King, Oleksandr, Alex D., Burkhard ---- /!\ '''Edit conflict - your version:''' ---- '''People:''' Burcin Erocal, Simon King, Alex, Alex D., Burkhard (total anarchy!) ---- /!\ '''End of edit conflict''' ---- |
| Line 23: | Line 36: |
| == Locapal support == '''People:''' Burcin Erocal There is experimental support for computing Groebner bases over certain localizations of operator algebras in Singular. See [[http://www.math.rwth-aachen.de/~Viktor.Levandovskyy/filez/singular/levandovskyy_kl.pdf|this presentation]] for more details. Support for arithmetic needs to be provided in Sage. |
|
| Line 31: | Line 38: |
| '''People:''' Burcin Erocal | '''People:''' '''Stefan Boethner''', Ralf, Burkhard, Burcin Erocal |
| Line 38: | Line 45: |
| == Algorithmic/Automatic Differentiation == | == Function Fields == |
| Line 40: | Line 47: |
| '''People:''' William Stein | The goal of this project is to get the basic infrastructure for function fields into Sage. See [[daysff/curves|Hess's papers and talks]]. |
| Line 42: | Line 49: |
| I never thought much of this topic, but there is [[http://www.euroscipy.org/talk/2045|a talk at Euroscipy]] suggesting it could be useful. Examples of what you could imagine doing: | People: '''William Stein''', Sebastian P. * Trac 9054: [[http://trac.sagemath.org/sage_trac/ticket/9054|Create a class for basic function_field arithmetic for Sage]] * Trac 9069: [[http://trac.sagemath.org/sage_trac/ticket/9069|Weak Popov Form (reduction algorithm)]] * Trac 9094: [[http://trac.sagemath.org/sage_trac/ticket/9094|is_square and sqrt for polynomials and fraction fields]] * Trac 9095: [[http://trac.sagemath.org/sage_trac/ticket/9095|Heights of points on elliptic curves over function fields]] Make sure to see [[daysff/curves|this page for more links]]. == Fast linear algebra over small extensions of GF(2) == '''People''': '''Martin Albrecht''', Ciaran Mullan, Robert Miller, Sebastian P., Thomas Here is how long Sage currently takes to compute the reduced row echelon form over GF(2^4) on a Macbook Pro (2nd generation): || n || Sage || NTL *2 || Magma || M4RIE || || 1000 || 49.49 || 18.84 || 0.090 || 0.097 || || 2000 || 429.05 || 149.11 || 0.510 || 0.529 || || 3000 || 1494.33 || 526.57 || 1.640 || 2.315 || Note that over GF(2^8) this code is already faster than Magma |
| Line 44: | Line 72: |
| sage: f(x,y,z)=sqrt(x^3+y)/(x-y^3*z)^(3/2) sage: a = fast_callable(f,CDF) sage: timeit('a(2,3,4)') 625 loops, best of 3: 24.2 µs per loop sage: g = f.derivative(x,10) sage: a = fast_callable(g,CDF) sage: timeit('a(2,3,4)') 625 loops, best of 3: 1.21 ms per loop }}} I imagine that one could instead do: {{{ sage: f(x,y,z)=sqrt(x^3+y)/(x-y^3*z)^(3/2) sage: a = fast_callable(g,CDF) sage-fantasy: b = a.derivative(x,10) sage-fantasy: timeit('b(2,3,4)') 625 loops, best of 3: 24.2 µs per loop |
> K<a> := GF(2^8); > for i := 1000 to 10001 by 1000 do for> A:=RandomMatrix(K,i,i); for> t:=Cputime(); for> E:=EchelonForm(A); for> print i, Cputime(t); for> end for; 1000 1.290 2000 9.870 3000 33.560 |
| Line 62: | Line 84: |
| This is just a random example I made up, but it's the sort of massive performance one might hope for from AD; the literature should have better examples. There are *tons* of potential Python libraries to consider, e.g., [[http://github.com/b45ch1/algopy|algopy]] and [[http://www.coin-or.org/CppAD/|CppAD]] (which has Python bindings). | {{{ gf(2^8), 1000 x 1000: wall time: 0.865 gf(2^8), 2000 x 2000: wall time: 4.306 gf(2^8), 3000 x 3000: wall time: 14.029 }}} == Generating Stuff == '''People:''' Robert Miller (self-determination!) === Fix sage.functions === '''People:''' '''Frederik''', William Stein, Harald == Easy ripping apart of symbolic expression trees == '''People:''' '''Burcin''', Thomas, Stefan, Frederik |
Sage Days 24 Coding Sprint Projects
This is a list of projects suitable for Sage Days 24. Feel free to add your favourite ideas/wishes, and to put your name down for something you're interested in (you'll need to get an account on the wiki to do this).
Contents
GIAC Factoring
People: Thomas, Burcin, Richard, William Stein (total anarchy, no leader!)
Kovacic's Algorithm
People: Burcin, Erocal, Felix
Implement Kovacic's algorithm in Sage.
Hypergeometric Functions
People: Flavia Stan, Karen Kohl, Fredrik Johansson, Zaf
Add a hypergeometric function class + simplifications
Plural support
![/!\](/moin_static1911/modernized/img/alert.png "/!\"){kind=small}
Edit conflict - other version:
People: Burcin Erocal, Simon King, Oleksandr, Alex D., Burkhard
Edit conflict - your version:
People: Burcin Erocal, Simon King, Alex, Alex D., Burkhard (total anarchy!)
End of edit conflict
Add support for Singular's noncommutative component Plural, finish #4539.
Parallel Integration
People: Stefan Boethner, Ralf, Burkhard, Burcin Erocal
Integrate Stefan Boettner's parallel integration code in Sage. There are several prerequisites for this, such as
algebraic function fields (transcendence degree > 1)
- differential rings/fields
- proper to_polynomial(), to_rational() functions for symbolic expressions
Function Fields
The goal of this project is to get the basic infrastructure for function fields into Sage. See Hess's papers and talks.
People: William Stein, Sebastian P.
Trac 9054: Create a class for basic function_field arithmetic for Sage
Trac 9069: Weak Popov Form (reduction algorithm)
Trac 9094: is_square and sqrt for polynomials and fraction fields
Trac 9095: Heights of points on elliptic curves over function fields
Make sure to see this page for more links.
Fast linear algebra over small extensions of GF(2)
People: Martin Albrecht, Ciaran Mullan, Robert Miller, Sebastian P., Thomas
Here is how long Sage currently takes to compute the reduced row echelon form over GF(2^4) on a Macbook Pro (2nd generation):
n |
Sage |
NTL *2 |
Magma |
M4RIE |
1000 |
49.49 |
18.84 |
0.090 |
0.097 |
2000 |
429.05 |
149.11 |
0.510 |
0.529 |
3000 |
1494.33 |
526.57 |
1.640 |
2.315 |
Note that over GF(2^8) this code is already faster than Magma
> K<a> := GF(2^8); > for i := 1000 to 10001 by 1000 do for> A:=RandomMatrix(K,i,i); for> t:=Cputime(); for> E:=EchelonForm(A); for> print i, Cputime(t); for> end for; 1000 1.290 2000 9.870 3000 33.560
gf(2^8), 1000 x 1000: wall time: 0.865 gf(2^8), 2000 x 2000: wall time: 4.306 gf(2^8), 3000 x 3000: wall time: 14.029
Generating Stuff
People: Robert Miller (self-determination!)
Fix sage.functions
People: Frederik, William Stein, Harald
Easy ripping apart of symbolic expression trees
People: Burcin, Thomas, Stefan, Frederik