Eliminations and echelon forms in exact linear algebra
|
|
- Douglas Watts
- 6 years ago
- Views:
Transcription
1 Eliminations and echelon forms in exact linear algebra Clément PERNET, INRIA-MOAIS, Grenoble Université, France East Coast Computer Algebra Day, University of Waterloo, ON, Canada, April 9, 2011 Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 1 / 59
2 Introduction Gaussian elimination in Computer Algebra Linear system solving: over Z p, Z, Q Polynomial system solving: Gröbner basis (Crypto, Number Theory) (Robotics, Crypto) Linear dependencies: rank, basis (of vector spaces, free modules, Krylov spaces,...) (Number Theory) Determinants: certificate of similarity (Graph Theory) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 2 / 59
3 Motivation Designing efficient dense gaussian elimination routines over an exact ring/field. Extensively studied for numerical computations Specificities of exact computations: No partial/full pivoting Rank profile matters size of coefficients (e.g. compressed in GF(2)) Study originating from, the design of the libraries FFLAS-FFPACK: word size finite fields M4RI: GF(2) asymmetry Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 3 / 59
4 Outline 1 Decompositions and factorizations Gaussian elimination based matrix decompositions Relations between decompositions 2 Algorithms Block recursive gaussian elimination Memory allocations Time complexity 3 Parallelization 4 Algorithms into practice: the case of GF(2) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 4 / 59
5 Outline 1 Decompositions and factorizations Gaussian elimination based matrix decompositions Relations between decompositions 2 Algorithms Block recursive gaussian elimination Memory allocations Time complexity 3 Parallelization 4 Algorithms into practice: the case of GF(2) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 5 / 59
6 Outline 1 Decompositions and factorizations Gaussian elimination based matrix decompositions Relations between decompositions 2 Algorithms Block recursive gaussian elimination Memory allocations Time complexity 3 Parallelization 4 Algorithms into practice: the case of GF(2) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 6 / 59
7 LU decomposition A = L U L unit lower triangular, U non-sing upper triangular Exists for matrices having the generic rank profile (every leading principal minor is non zero) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 7 / 59
8 LUP, PLU decomposition A U = L P P a permutation matrix Exists for Any non-singular matrix Or any matrix with generic row rank profile Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 8 / 59
9 LUP, PLU decomposition A U = L P A = P L U P a permutation matrix Exists for Any non-singular matrix Or any matrix with generic row rank profile Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 8 / 59
10 LSP, LQUP, PLUQ decompositions A = L S P S: semi-upper triangular, Q permutation matrix Exists for any m n matrix Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 9 / 59
11 LSP, LQUP, PLUQ decompositions A = L S P A = L Q U P S: semi-upper triangular, Q permutation matrix Exists for any m n matrix Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 9 / 59
12 LSP, LQUP, PLUQ decompositions A = L S P A = L Q U P A = P L U Q S: semi-upper triangular, Q permutation matrix Exists for any m n matrix Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 9 / 59
13 Echelon form decomposition Row Echelon Form XA = R X A = R X, Y : non-singular transformation matrices R, C: matrices in row/col echelon form Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 10 / 59
14 Echelon form decomposition Row Echelon Form XA = R X A = R Column Echelon Form AY = C A Y = C X, Y : non-singular transformation matrices R, C: matrices in row/col echelon form Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 10 / 59
15 Reduced echelon form decomposition Row Reduced Echelon Form XA = R X A = R X, Y : non-singular transformation matrices R, C: matrices in reduced row/col echelon form Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 11 / 59
16 Reduced echelon form decomposition Row Reduced Echelon Form XA = R X A = R Column Reduced Echelon Form AY = C A Z = C X, Y : non-singular transformation matrices R, C: matrices in reduced row/col echelon form Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 11 / 59
17 Turing factorization A = P L U R PLU decomposition of X 1, the inverse of the transformation matrix to REF Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 12 / 59
18 CUP and PLE decompositions A = C U P C: column echelon form E: row echelon form Exists for any m n matrix Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 13 / 59
19 CUP and PLE decompositions A = C U P A = P L E C: column echelon form E: row echelon form Exists for any m n matrix Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 13 / 59
20 Outline 1 Decompositions and factorizations Gaussian elimination based matrix decompositions Relations between decompositions 2 Algorithms Block recursive gaussian elimination Memory allocations Time complexity 3 Parallelization 4 Algorithms into practice: the case of GF(2) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 14 / 59
21 Relations: up to permutations From LSP to LQUP S = QU A = L S P A = L Q U P Fact The first r = rank(a) values of the permutation Q are monotonically increasing. Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 15 / 59
22 Relations: up to permutations From LQUP to CUP A = L Q U P C = LQ A = L Q U P A = C U P And to PLE using transposition: PLE(A T ) = CUP(A) T Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 16 / 59
23 Relations: From LQUP to PLUQ A = L Q U P P Q, L = Q T LQ A = C U P A = P L U Q Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 17 / 59
24 Relations: From CUP to ColumnEchelon form A = C U P Y = P T [ U I n r ] 1 A = C U Id P A Y = Similarly, from PLE to RowEchelon form C Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 18 / 59
25 Relations: From Column Echelon form to Reduced Column Echelon form A Y = C M [ M Z = Y I n r ] 1 A Y = A Y = Q Q Id M Id A Z = C Similarly, from PLE to RowEchelon form Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 19 / 59
26 Relations: From PLE to Turing A = P L E U = V, [ ] 1 V R = E I n r A = P L A = P L V V Q Id Q A = P L U R Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 20 / 59
27 Outline 1 Decompositions and factorizations Gaussian elimination based matrix decompositions Relations between decompositions 2 Algorithms Block recursive gaussian elimination Memory allocations Time complexity 3 Parallelization 4 Algorithms into practice: the case of GF(2) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 21 / 59
28 Algorithms: main types Three ways to group operations: 1 simple iterative Apply the standard Gaussian elimination in dimension n Main loop for i=1 to n 2 block algorithms 1 block iterative (Tile) Apply Gaussian elimination in dimension n/k over blocks of size k Main loop: for i=1 to n/k 2 block recursive Apply Gaussian elimination in dimension 2 recursively on blocks of size n/2 i Main loop: for i=1 to 2 Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 22 / 59
29 Type of algorithms Data locality: prefer block algorithms cache aware: block iterative cache oblivious: block recursive Base case efficieny: simple iterative Asymptotic time complexity: block recursive Parallelization: block iterative Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 23 / 59
30 Outline 1 Decompositions and factorizations Gaussian elimination based matrix decompositions Relations between decompositions 2 Algorithms Block recursive gaussian elimination Memory allocations Time complexity 3 Parallelization 4 Algorithms into practice: the case of GF(2) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 24 / 59
31 Block recursive gaussian elimination Author Year Computation Requirement Strassen 69 Inverse gen. rank prof. Bunch, Hopcroft 74 LUP gen. row rank prof. Ibarra, Moran, Hui 82 LSP, LQUP none Schönage, Keller-Gerig 85 StepForm none Storjohann 00 Echelon, RedEch none here 11 CUP,PLE,PLUQ none Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 25 / 59
32 Block recursive gaussian elimination Author Year Computation Requirement Strassen 69 Inverse gen. rank prof. Bunch, Hopcroft 74 LUP gen. row rank prof. Ibarra, Moran, Hui 82 LSP, LQUP none Schönage, Keller-Gerig 85 StepForm none Storjohann 00 Echelon, RedEch none here 11 CUP,PLE,PLUQ none Comparison according to No requirement on the input matrix Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 25 / 59
33 Block recursive gaussian elimination Author Year Computation Requirement Strassen 69 Inverse gen. rank prof. Bunch, Hopcroft 74 LUP gen. row rank prof. Ibarra, Moran, Hui 82 LSP, LQUP none Schönage, Keller-Gerig 85 StepForm none Storjohann 00 Echelon, RedEch none here 11 CUP,PLE,PLUQ none Comparison according to No requirement on the input matrix Rank sensitive complexity Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 25 / 59
34 Block recursive gaussian elimination Author Year Computation Requirement Strassen 69 Inverse gen. rank prof. Bunch, Hopcroft 74 LUP gen. row rank prof. Ibarra, Moran, Hui 82 LSP, LQUP none Schönage, Keller-Gerig 85 StepForm none Storjohann 00 Echelon, RedEch none here 11 CUP,PLE,PLUQ none Comparison according to No requirement on the input matrix Rank sensitive complexity Memory allocations Constant factor in the time complexity Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 25 / 59
35 Outline 1 Decompositions and factorizations Gaussian elimination based matrix decompositions Relations between decompositions 2 Algorithms Block recursive gaussian elimination Memory allocations Time complexity 3 Parallelization 4 Algorithms into practice: the case of GF(2) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 26 / 59
36 Memory requirements: Definition In place = output overrides the input and computation does not need extra memory (considering Matrix multiplication C C + AB as a black box) Remark: a unit lower triangular and an upper triangular matrix can be stored on the same m n storage! Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 27 / 59
37 Preliminaries TRSM: TRiangular Solve with Matrix [ ] 1 [ ] [ ] [ ] [ A B D A 1 I B I = C E I I ] [ ] D C 1 E Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 28 / 59
38 Preliminaries TRSM: TRiangular Solve with Matrix [ ] 1 [ ] [ ] [ ] [ A B D A 1 I B I = C E I I Compute F = C 1 E Compute G = D BF Compute [ H ] = A 1 G H Return F ] [ ] D C 1 E (Recursive call) (MM) (Recursive call) A B C D E Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 28 / 59
39 Preliminaries TRSM: TRiangular Solve with Matrix [ ] 1 [ ] [ ] [ ] [ A B D A 1 I B I = C E I I Compute F = C 1 E Compute G = D BF Compute [ H ] = A 1 G H Return F ] [ ] D C 1 E (Recursive call) (MM) (Recursive call) A B C D F Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 28 / 59
40 Preliminaries TRSM: TRiangular Solve with Matrix [ ] 1 [ ] [ ] [ ] [ A B D A 1 I B I = C E I I Compute F = C 1 E Compute G = D BF Compute [ H ] = A 1 G H Return F ] [ ] D C 1 E (Recursive call) (MM) (Recursive call) A B C G F Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 28 / 59
41 Preliminaries TRSM: TRiangular Solve with Matrix [ ] 1 [ ] [ ] [ ] [ A B D A 1 I B I = C E I I Compute F = C 1 E Compute G = D BF Compute [ H ] = A 1 G H Return F ] [ ] D C 1 E (Recursive call) (MM) (Recursive call) A B C H F Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 28 / 59
42 Preliminaries TRSM: TRiangular Solve with Matrix [ ] 1 [ ] [ ] [ ] [ A B D A 1 I B I = C E I I Compute F = C 1 E Compute G = D BF Compute [ H ] = A 1 G H Return F ] [ ] D C 1 E (Recursive call) (MM) (Recursive call) A B C H F O (n ω ) In place Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 28 / 59
43 The LSP algorithm S A1 A2 1 Split A Row-wise L Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 29 / 59
44 The LSP algorithm S S 1 A21 V A 22 1 Split A Row-wise 2 Recursive call on A 1 L L1 Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 29 / 59
45 The LSP algorithm S S 1 V A 22 1 Split A Row-wise 2 Recursive call on A 1 3 G A 21 U 1 1 (trsm) U 1 L L1 G Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 29 / 59
46 The LSP algorithm S L S 1 L1 G V H 1 Split A Row-wise 2 Recursive call on A 1 3 G A 21 U 1 1 (trsm) 4 H A 22 G V (MM) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 29 / 59
47 The LSP algorithm S L S 1 L1 G S 2 L 2 1 Split A Row-wise 2 Recursive call on A 1 3 G A 21 U 1 1 (trsm) 4 H A 22 G V (MM) 5 Recursive call on H Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 29 / 59
48 The LQUP algorithm U A1 A2 1 Split A Row-wise L Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 30 / 59
49 The LQUP algorithm U U 1 A21 V A 22 1 Split A Row-wise 2 Recursive call on A 1 L L1 Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 30 / 59
50 The LQUP algorithm U 1 U V A 22 1 Split A Row-wise 2 Recursive call on A 1 3 G A 21 U 1 1 (trsm) L L1 G Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 30 / 59
51 The LQUP algorithm U L U 1 L1 G V H 1 Split A Row-wise 2 Recursive call on A 1 3 G A 21 U 1 1 (trsm) 4 H A 22 G V (MMl) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 30 / 59
52 The LQUP algorithm L U 1 U 1 Split A Row-wise L1 G U 2 L 2 2 Recursive call on A 1 3 G A 21 U 1 1 (trsm) 4 H A 22 G V (MMl) 5 Recursive call on H Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 30 / 59
53 The LQUP algorithm U 1 U U 2 1 Split A Row-wise 2 Recursive call on A 1 3 G A 21 U 1 1 (trsm) 4 H A 22 G V (MMl) L L1 G L 2 5 Recursive call on H Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 30 / 59
54 The CUP decomposition A1 A2 1 Split A Row-wise Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 31 / 59
55 The CUP decomposition 1 Split A Row-wise C U 1 1 V A21 A22 2 Recursive call on A 1 Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 31 / 59
56 The CUP decomposition 1 Split A Row-wise C U 1 1 G V A 22 2 Recursive call on A 1 3 G A 21 U 1 1 (trsm) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 31 / 59
57 The CUP decomposition 1 Split A Row-wise C U 1 1 G V H 2 Recursive call on A 1 3 G A 21 U 1 1 (trsm) 4 H A 22 G V (MM) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 31 / 59
58 The CUP decomposition 1 Split A Row-wise C U 1 1 U 2 G C 2 2 Recursive call on A 1 3 G A 21 U 1 1 (trsm) 4 H A 22 G V (MM) 5 Recursive call on H Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 31 / 59
59 The CUP decomposition 1 Split A Row-wise U1 C1 U 2 G C 2 2 Recursive call on A 1 3 G A 21 U 1 1 (trsm) 4 H A 22 G V (MM) 5 Recursive call on H 6 Row permutations Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 31 / 59
60 Memory: LSP vs LQUP vs PLUQ vs CUP Decomposition In place LSP N LQUP N PLUQ Y CUP Y Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 32 / 59
61 Echelon forms From CUP to ColumnEchelon form [ ] 1 Y = P T U1 U 2 I n r = P T [ U 1 1 U 1 1 U 2 I n r ] A = A = C C A Y = U U Id C P P Additional operations: U 1 U 2 trsm (triangular system solve) in-place Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 33 / 59
62 Echelon forms From CUP to ColumnEchelon form [ ] 1 Y = P T U1 U 2 I n r = P T [ U 1 1 U 1 1 U 2 I n r ] A = A = C C A Y = U U Id C P P Additional operations: U 1 U 2 trsm (triangular system solve) in-place U 1 1 : trtri (triangular inverse) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 33 / 59
63 Echelon forms From CUP to ColumnEchelon form [ ] 1 Y = P T U1 U 2 I n r = P T [ U 1 1 U 1 1 U 2 I n r ] A = A = C C A Y = U U Id C P P Additional operations: U 1 U 2 trsm (triangular system solve) in-place U 1 1 : trtri (triangular inverse) in-place Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 33 / 59
64 From LQUP to Column Echelon TRTRI: triangular inverse 1: if n = 1 then 2: U U 1 [ ] 1 [ U1 U 2 U 1 = 1 U 1 U 3 1 U 2U 1 ] 3 U 1 3 3: else 4: U 2 U 1 3 U 2 TRSM 5: U 2 U 2 U 1 3 TRSM 6: U 1 U 1 1 TRTRI 7: U 3 U 1 3 TRTRI 8: end if Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 34 / 59
65 Reduced Echelon forms From Column Echelon form to Reduced Column Echelon form A Y = C M [ M Z = Y I n r ] 1 A Y = A Y = Q Q Id M Id A Z = C Similarly, from PLE to RowEchelon form Again reduces to: U 1 X: TRSM, in-place U 1 : TRTRI, in-place Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 35 / 59
66 Reduced Echelon forms From Column Echelon form to Reduced Column Echelon form A Y = C M [ M Z = Y I n r ] 1 A Y = A Y = Q Q Id M Id A Z = C Similarly, from PLE to RowEchelon form Again reduces to: U 1 X: TRSM, in-place U 1 : TRTRI, in-place UL: TRTRM, Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 35 / 59
67 Reduced Echelon forms From Column Echelon form to Reduced Column Echelon form A Y = C M [ M Z = Y I n r ] 1 A Y = A Y = Q Q Id M Id A Z = C Similarly, from PLE to RowEchelon form Again reduces to: U 1 X: TRSM, in-place U 1 : TRTRI, in-place UL: TRTRM, Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 35 / 59
68 Reduced Echelon forms From Column Echelon form to Reduced Column Echelon form A Y = C M [ M Z = Y I n r ] 1 A Y = A Y = Q Q Id M Id A Z = C Similarly, from PLE to RowEchelon form Again reduces to: U 1 X: TRSM, in-place U 1 : TRTRI, in-place UL: TRTRM, in-place Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 35 / 59
69 From Echelon to Reduced Echelon TRTRM: triangular triangular multiplication [ ] [ ] [ ] U1 U 2 L1 U1 L = 1 + U 2 L 2 U 2 L 3 U 3 L 2 L 3 U 3 L 2 U 3 L 3 1: X 1 U 1 L 1 TRTRM 2: X 1 X 1 + U 2 L 2 MM 3: X 2 U 2 L 3 TRMM 4: X 3 U 3 L 2 TRMM 5: X 4 U 3 L 3 TRTRM Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 36 / 59
70 From Echelon to Reduced Echelon TRTRM: triangular triangular multiplication [ ] [ ] [ ] U1 U 2 L1 U1 L = 1 + U 2 L 2 U 2 L 3 U 3 L 2 L 3 U 3 L 2 U 3 L 3 1: X 1 U 1 L 1 TRTRM 2: X 1 X 1 + U 2 L 2 MM 3: X 2 U 2 L 3 TRMM 4: X 3 U 3 L 2 TRMM 5: X 4 U 3 L 3 TRTRM U L 1 1 L L U 2 U Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 36 / 59
71 From Echelon to Reduced Echelon TRTRM: triangular triangular multiplication [ ] [ ] [ ] U1 U 2 L1 U1 L = 1 + U 2 L 2 U 2 L 3 U 3 L 2 L 3 U 3 L 2 U 3 L 3 1: X 1 U 1 L 1 TRTRM 2: X 1 X 1 + U 2 L 2 MM 3: X 2 U 2 L 3 TRMM 4: X 3 U 3 L 2 TRMM 5: X 4 U 3 L 3 TRTRM X 1 L L U 2 U Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 36 / 59
72 From Echelon to Reduced Echelon TRTRM: triangular triangular multiplication [ ] [ ] [ ] U1 U 2 L1 U1 L = 1 + U 2 L 2 U 2 L 3 U 3 L 2 L 3 U 3 L 2 U 3 L 3 1: X 1 U 1 L 1 TRTRM 2: X 1 X 1 + U 2 L 2 MM 3: X 2 U 2 L 3 TRMM 4: X 3 U 3 L 2 TRMM 5: X 4 U 3 L 3 TRTRM X 1 L L U 2 U Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 36 / 59
73 From Echelon to Reduced Echelon TRTRM: triangular triangular multiplication [ ] [ ] [ ] U1 U 2 L1 U1 L = 1 + U 2 L 2 U 2 L 3 U 3 L 2 L 3 U 3 L 2 U 3 L 3 1: X 1 U 1 L 1 TRTRM 2: X 1 X 1 + U 2 L 2 MM 3: X 2 U 2 L 3 TRMM 4: X 3 U 3 L 2 TRMM 5: X 4 U 3 L 3 TRTRM X 1 L L X 2 U Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 36 / 59
74 From Echelon to Reduced Echelon TRTRM: triangular triangular multiplication [ ] [ ] [ ] U1 U 2 L1 U1 L = 1 + U 2 L 2 U 2 L 3 U 3 L 2 L 3 U 3 L 2 U 3 L 3 1: X 1 U 1 L 1 TRTRM 2: X 1 X 1 + U 2 L 2 MM 3: X 2 U 2 L 3 TRMM 4: X 3 U 3 L 2 TRMM 5: X 4 U 3 L 3 TRTRM X 1 X L X 2 U Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 36 / 59
75 From Echelon to Reduced Echelon TRTRM: triangular triangular multiplication [ ] [ ] [ ] U1 U 2 L1 U1 L = 1 + U 2 L 2 U 2 L 3 U 3 L 2 L 3 U 3 L 2 U 3 L 3 1: X 1 U 1 L 1 TRTRM 2: X 1 X 1 + U 2 L 2 MM 3: X 2 U 2 L 3 TRMM 4: X 3 U 3 L 2 TRMM 5: X 4 U 3 L 3 TRTRM X 1 X 3 X 2 X 4 O (n ω ) In place Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 36 / 59
76 Example: in place matrix inversion A Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 37 / 59
77 Example: in place matrix inversion A L U CUP decomposition A = LU Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 37 / 59
78 Example: in place matrix inversion U A L L U 1 CUP decomposition Echelon AU 1 = L Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 37 / 59
79 Example: in place matrix inversion U A L L U 1 A 1 CUP decomposition Echelon Reduced Echelon A(U 1 L 1 ) = I Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 37 / 59
80 Experiments./test-invert A ,996,125,000 bytes x ms bytes 16M x804aa7e:_z10read_fieldi 14M 12M x8049a98:main 10M x804edc0:void%fflas::win 8M 6M x804edee:void%fflas::win 4M x804edd4:void%fflas::win 2M 0M ms./test-redechelon A ,663,687,500 bytes x ms bytes 8M x804a47e:_z10read_fieldi 6M x804b0ec:void%fflas::win 4M x804b11a:void%fflas::win 2M x804b100:void%fflas::win 0M ms Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 38 / 59
81 Direct computation of the Reduced Echelon form Strassen 69: inverse of generic matrices Storjohann 00: Gauss-Jordan generalization for any rank profile Matrix Inversion [Strassen 69] [ ] 1 [ A B A 1 = C D I ] [ I B I ] [ I (D CA 1 B) 1 ] [ I CA 1 I ] Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 39 / 59
82 Direct computation of the Reduced Echelon form Strassen 69: inverse of generic matrices Storjohann 00: Gauss-Jordan generalization for any rank profile Matrix Inversion [Strassen 69] [ ] 1 [ A B A 1 = C D I ] [ I B I ] [ I (D CA 1 B) 1 ] [ I CA 1 1: Compute E = A 1 (Recursive call) 2: Compute F = D CEB (MM) 3: Compute G = F 1 (Recursive call) 4: Compute H = EB (MM) 5: Compute J = HG (MM) 6: Compute K = CE (MM) 7: Compute L = E + JK (MM) 8: Compute [ M = GK ] (MM) E J 9: Return M G I ] Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 39 / 59
83 Strassen-Storjohann s Gauss-Jordan elimination Problem Needs to perform operations of the form A AB not doable in place by a usual matrix multiplication algorithm Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 40 / 59
84 Strassen-Storjohann s Gauss-Jordan elimination Problem Needs to perform operations of the form A AB not doable in place by a usual matrix multiplication algorithm Workaround [Storjohann]: 1 Decompose B = LU LU 2 A AL trmm 3 A AU trmm Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 40 / 59
85 Outline 1 Decompositions and factorizations Gaussian elimination based matrix decompositions Relations between decompositions 2 Algorithms Block recursive gaussian elimination Memory allocations Time complexity 3 Parallelization 4 Algorithms into practice: the case of GF(2) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 41 / 59
86 Rank sensitive time complexity Fact Algorithms LSP, CUP, LQUP, PLUQ,... have a rank sensitive computation time: O ( mnr ω 2) n=3000, PIII 1.6Ghz, 512Mb RAM Block algorithm Iterative algorithm 100 Time (s) Rank Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 42 / 59
87 Rank sensitivity and Left/Right/Crout Looking variants Right looking / Left looking/ Crout variants does not affect 2x2 splitting for block iterative: always the same rank sensitive complexity: 2n 2 r 2nr 2 + 2/3r 3 Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 43 / 59
88 Time complexity: comparing constants O (n ω ) = C ω n 3 Algorithm Constant C ω C 3 C log2 7 in-place MM C ω 2 6 C TRSM ω ω 1 2 C TRTRI ω 1 8 (2 ω 1 2)(2 ω 1 1) 3 5 TRTRM, CUP C ω PLUQ LQUP, ω ω C Echelon ω ω ω 2 5 C RedEchelon ω(2 ω 1 +2) 44 2 (2 ω 1 2)(2 ω 1 1) 5 5C StepForm ω C ω ω 1 1 (2 ω 1 1)(2 ω 2 1) 5 GJ 2 8 Cω 2 ω 2 1 : GJ: GaussJordan alg of [Storjohann00] computing the reduced echelon form Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 44 / 59
89 Applications to standard linalg problems Problem Using C ω C 3 C log2 7 In place Rank RankProfile IsSingular Det Solve GJ CUP C ω 2 ω 2 1 C ω Cω 2 ω ω Inverse GJ CUP C ω 2 ω 2 1 C ω(2 ω 1 +2) (2 ω 1 2)(2 ω 1 1) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 45 / 59
90 Summary - Det - Rank - RankProfile - Solve - Echelon - Nullspace basis - Reduced Echelon form - Inverse A 2/3 L U U 1/3 L -1 1/3 L -1U-1 2/3 LUDecomp TRTRI TRTRI TRTRM Gauss 1 GaussJordan 2 LU decomp A=LU Echelon form decomp (TA=E) A -1 Reduced Echelon form decomp TA=R Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 46 / 59
91 Outline 1 Decompositions and factorizations Gaussian elimination based matrix decompositions Relations between decompositions 2 Algorithms Block recursive gaussian elimination Memory allocations Time complexity 3 Parallelization 4 Algorithms into practice: the case of GF(2) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 47 / 59
92 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
93 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
94 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
95 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
96 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
97 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
98 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
99 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
100 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
101 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
102 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
103 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
104 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
105 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
106 Parallelization Using block iterative algorithm Parallelizing the matrix multiplication updates Always a critical path of n on the diagonal Scheduling tasks DAG, static scheduling,... Dynamic scheduling, work-stealing Advantage of exact computations: No need to update pivots (local pivoting) Multi-frontal approach made easier less ops Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 48 / 59
107 Dynamic scheduling with Kaapi hread0 hread1 hread2 hread3 hread4 hread5 hread6 hread7 hread8 hread9 hread10 hread11 hread12 hread13 hread14 hread15 hread16 hread17 hread18 hread19 hread20 hread21 hread22 hread23 hread24 hread25 hread26 hread27 hread28 hread29 hread30 hread31 hread32 hread33 hread34 hread35 hread36 hread37 hread38 hread39 hread40 hread41 hread42 hread43 hread44 hread45 hread46 hread47 ROOT Thread0 Thread1 Thread2 Thread3 Thread4 Thread5 Thread6 Thread7 Thread8 Thread9 Thread10 Thread11 Thread12 Thread13 Thread14 Thread15 Thread16 Thread17 Thread18 Thread19 Thread20 Thread21 Thread22 Thread23 Thread24 Thread25 Thread26 Thread27 Thread28 Thread29 Thread30 Thread31 Thread32 Thread33 Thread34 Thread35 Thread36 Thread37 Thread38 Thread39 Thread40 Thread41 Thread42 Thread43 Thread44 Thread45 Thread min: 0 max: min: 0 max: Thread0 Thread1 Thread2 Thread3 Thread4 Thread5 Thread6 Thread7 Thread8 Thread9 Thread10 Thread11 Thread12 Thread13 Thread14 Thread15 Thread16 Thread17 Thread18 Thread19 Thread20 Thread21 Thread22 Thread23 Thread24 Thread25 Thread26 Thread27 Thread28 Thread29 Thread30 Thread31 Thread32 Thread33 Thread34 Thread35 Thread36 Thread37 Thread38 Thread39 Thread40 Thread41 Thread42 Thread43 Thread44 Thread45 Thread46 Thread47 min: 0 max: Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 49 / 59
108 Outline 1 Decompositions and factorizations Gaussian elimination based matrix decompositions Relations between decompositions 2 Algorithms Block recursive gaussian elimination Memory allocations Time complexity 3 Parallelization 4 Algorithms into practice: the case of GF(2) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 50 / 59
109 Dedicated implementations over GF(2) The M4RI library [M Albrecht & Al.]: Dense basic linear algebra over GF(2) Packed representation: unsigned long long is a vector of 64 coefficients Matrix multiplication: Gray code table lookup (Methode of the 4 russians) O ( n 3 / log n ) Strassen on the coarse grain. O ( (n/k) ω k 3 / log k ) =O (n ω ) Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 51 / 59
110 Design of the gaussian elimination routine PLE vs CUP Row major storage easier to permute pivots along columns PLE rather than CUP PLE vs PLUQ PLUQ involves more back and forth pivoting Compact LAPACK representation of permutation: product of transpositions Not possible to maintain for P in O (1) PLUQ PLE rater than PLUQ Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 52 / 59
111 Design of the gaussian elimination routine Structure of the algorithm block recursive PLE on the coarse grain (O (n ω )) block iterative PLE with block size k = log n simple iterative PLE on cache fitting blocks Additional tricks: With gray code tables: inverting means reverse table look-up TRSM as efficient as MM... Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 53 / 59
112 Visualisation Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 54 / 59
113 Visualisation Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 54 / 59
114 Visualisation Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 54 / 59
115 Visualisation Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 54 / 59
116 Visualisation Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 54 / 59
117 Visualisation Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 54 / 59
118 Visualisation Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 54 / 59
119 Visualisation Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 54 / 59
120 Visualisation Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 54 / 59
121 Visualisation Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 54 / 59
122 Visualisation Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 54 / 59
123 Visualisation Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 54 / 59
124 Results: Reduced Row Echelon Form execution time t Magma Sage Magma/Sage Elimination log 2 relation Figure: 2.66 Ghz Intel i7, 4GB RAM Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 55 / 59
125 Results: Row Echelon Form Using one core we can compute the echelon form of a 500, , 000 dense random matrix over F 2 in seconds = 2.7 hours. Using four cores decomposition we can compute the echelon form of a random dense 500, , 000 matrix in seconds = 1.05 hours. Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 56 / 59
126 Work-in-Progress: Sensitivity to Sparsity M4RI M+P 0.15 PLS Magma execution time t nonzero elements per row on average Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 57 / 59
127 Work-in-Progress: Gröbner Basis Linear Algebra 64-bit Debian/GNU Linux, 2.6Ghz Opteron) Problem Matrix Density Magma M4RI PLS M+P 0.15 M+P 0.20 Dimension HFE 25 12, , s 3.28s 3.45s 3.03s 3.21s HFE 30 19, , s 23.72s 25.42s 23.84s 25.09s HFE 35 29, , s s s s s MXL 26, , s 23.03s 19.04s 17.91s 18.00s Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 58 / 59
128 Work-in-Progress: Multi-core Support M4RI BOpS & Speed-up PLE BOpS & Speed-up Clément Pernet (INRIA-MOAIS, Grenoble) Eliminations in exact linear algebra ECCAD 11, U of Waterloo 59 / 59
Linear Algebra 1: Computing canonical forms in exact linear algebra
Linear Algebra : Computing canonical forms in exact linear algebra Clément PERNET, LIG/INRIA-MOAIS, Grenoble Université, France ECRYPT II: Summer School on Tools, Mykonos, Grece, June st, 202 Introduction
More informationComputing the Rank Profile Matrix
Computing the Rank Profile Matrix Jean-Guillaume Dumas, Clément Pernet, Ziad Sultan To cite this version: Jean-Guillaume Dumas, Clément Pernet, Ziad Sultan. Computing the Rank Profile Matrix. Kazuhiro
More informationarxiv: v2 [cs.sc] 14 May 2018
Fast Computation of the Rank Profile Matrix and the Generalized Bruhat Decomposition arxiv:1601.01798v2 [cs.sc 14 May 2018 Jean-Guillaume Dumas Université Grenoble Alpes, Laboratoire LJK, umr CNRS, BP53X,
More informationHigh Performance and Reliable Algebraic Computing
High Performance and Reliable Algebraic Computing Soutenance d habilitation à diriger des recherches Clément Pernet Université Joseph Fourier (Grenoble 1) November 25, 2014 Rapporteurs : Daniel Augot,
More informationAlgorithmic Algebraic Techniques and their Application to Block Cipher Cryptanalysis
Algorithmic Algebraic Techniques and their Application to Block Cipher Cryptanalysis Martin Albrecht M.R.Albrecht@rhul.ac.uk Thesis submitted to Royal Holloway, University of London for the degree of Doctor
More informationMatrices and systems of linear equations
Matrices and systems of linear equations Samy Tindel Purdue University Differential equations and linear algebra - MA 262 Taken from Differential equations and linear algebra by Goode and Annin Samy T.
More informationPractical Linear Algebra: A Geometry Toolbox
Practical Linear Algebra: A Geometry Toolbox Third edition Chapter 12: Gauss for Linear Systems Gerald Farin & Dianne Hansford CRC Press, Taylor & Francis Group, An A K Peters Book www.farinhansford.com/books/pla
More informationPreliminary/Qualifying Exam in Numerical Analysis (Math 502a) Spring 2012
Instructions Preliminary/Qualifying Exam in Numerical Analysis (Math 502a) Spring 2012 The exam consists of four problems, each having multiple parts. You should attempt to solve all four problems. 1.
More informationOn the use of polynomial matrix approximant in the block Wiedemann algorithm
On the use of polynomial matrix approximant in the block Wiedemann algorithm ECOLE NORMALE SUPERIEURE DE LYON Laboratoire LIP, ENS Lyon, France Pascal Giorgi Symbolic Computation Group, School of Computer
More informationSTAT 309: MATHEMATICAL COMPUTATIONS I FALL 2018 LECTURE 13
STAT 309: MATHEMATICAL COMPUTATIONS I FALL 208 LECTURE 3 need for pivoting we saw that under proper circumstances, we can write A LU where 0 0 0 u u 2 u n l 2 0 0 0 u 22 u 2n L l 3 l 32, U 0 0 0 l n l
More informationLinear Algebra, Summer 2011, pt. 2
Linear Algebra, Summer 2, pt. 2 June 8, 2 Contents Inverses. 2 Vector Spaces. 3 2. Examples of vector spaces..................... 3 2.2 The column space......................... 6 2.3 The null space...........................
More informationMaximal Matching via Gaussian Elimination
Maximal Matching via Gaussian Elimination Piotr Sankowski Uniwersytet Warszawski - p. 1/77 Outline Maximum Matchings Lovásza s Idea, Maximum matchings, Rabin and Vazirani, Gaussian Elimination, Simple
More informationSparse BLAS-3 Reduction
Sparse BLAS-3 Reduction to Banded Upper Triangular (Spar3Bnd) Gary Howell, HPC/OIT NC State University gary howell@ncsu.edu Sparse BLAS-3 Reduction p.1/27 Acknowledgements James Demmel, Gene Golub, Franc
More informationLinear Algebra Section 2.6 : LU Decomposition Section 2.7 : Permutations and transposes Wednesday, February 13th Math 301 Week #4
Linear Algebra Section. : LU Decomposition Section. : Permutations and transposes Wednesday, February 1th Math 01 Week # 1 The LU Decomposition We learned last time that we can factor a invertible matrix
More informationConceptual Questions for Review
Conceptual Questions for Review Chapter 1 1.1 Which vectors are linear combinations of v = (3, 1) and w = (4, 3)? 1.2 Compare the dot product of v = (3, 1) and w = (4, 3) to the product of their lengths.
More informationExact Linear Algebra Algorithmic: Theory and Practice
Exact Linear Algebra Algorithmic: Theory and Practice ISSAC 15 Tutorial Clément Pernet Université Grenoble Alpes, Inria, LIP-AriC July 6, 2015 C. Pernet Exact Linear Algebra Algorithmic July 6, 2015 1
More informationHybrid static/dynamic scheduling for already optimized dense matrix factorization. Joint Laboratory for Petascale Computing, INRIA-UIUC
Hybrid static/dynamic scheduling for already optimized dense matrix factorization Simplice Donfack, Laura Grigori, INRIA, France Bill Gropp, Vivek Kale UIUC, USA Joint Laboratory for Petascale Computing,
More informationLinear Algebraic Equations
Linear Algebraic Equations 1 Fundamentals Consider the set of linear algebraic equations n a ij x i b i represented by Ax b j with [A b ] [A b] and (1a) r(a) rank of A (1b) Then Axb has a solution iff
More informationMatrix Inversion (or LUP-Factorization) via the Method of Four Russians, in Θ(n 3 / log n) Time
DEFINE ISSUE using \ issue Matrix Inversion (or LUP-Factorization) via the Method of Four Russians, in Θ(n 3 / log n) Time Gregory V. Bard Abstract We present an algorithm for reducing dense matrices over
More informationAMS526: Numerical Analysis I (Numerical Linear Algebra for Computational and Data Sciences)
AMS526: Numerical Analysis I (Numerical Linear Algebra for Computational and Data Sciences) Lecture 19: Computing the SVD; Sparse Linear Systems Xiangmin Jiao Stony Brook University Xiangmin Jiao Numerical
More information5.7 Cramer's Rule 1. Using Determinants to Solve Systems Assumes the system of two equations in two unknowns
5.7 Cramer's Rule 1. Using Determinants to Solve Systems Assumes the system of two equations in two unknowns (1) possesses the solution and provided that.. The numerators and denominators are recognized
More informationAlgebra C Numerical Linear Algebra Sample Exam Problems
Algebra C Numerical Linear Algebra Sample Exam Problems Notation. Denote by V a finite-dimensional Hilbert space with inner product (, ) and corresponding norm. The abbreviation SPD is used for symmetric
More informationLinear Systems of n equations for n unknowns
Linear Systems of n equations for n unknowns In many application problems we want to find n unknowns, and we have n linear equations Example: Find x,x,x such that the following three equations hold: x
More informationExtra Problems: Chapter 1
MA131 (Section 750002): Prepared by Asst.Prof.Dr.Archara Pacheenburawana 1 Extra Problems: Chapter 1 1. In each of the following answer true if the statement is always true and false otherwise in the space
More informationA Review of Matrix Analysis
Matrix Notation Part Matrix Operations Matrices are simply rectangular arrays of quantities Each quantity in the array is called an element of the matrix and an element can be either a numerical value
More informationANALYTICAL MATHEMATICS FOR APPLICATIONS 2018 LECTURE NOTES 3
ANALYTICAL MATHEMATICS FOR APPLICATIONS 2018 LECTURE NOTES 3 ISSUED 24 FEBRUARY 2018 1 Gaussian elimination Let A be an (m n)-matrix Consider the following row operations on A (1) Swap the positions any
More informationMTH 464: Computational Linear Algebra
MTH 464: Computational Linear Algebra Lecture Outlines Exam 2 Material Prof. M. Beauregard Department of Mathematics & Statistics Stephen F. Austin State University February 6, 2018 Linear Algebra (MTH
More informationforms Christopher Engström November 14, 2014 MAA704: Matrix factorization and canonical forms Matrix properties Matrix factorization Canonical forms
Christopher Engström November 14, 2014 Hermitian LU QR echelon Contents of todays lecture Some interesting / useful / important of matrices Hermitian LU QR echelon Rewriting a as a product of several matrices.
More informationLecture 2 Systems of Linear Equations and Matrices, Continued
Lecture 2 Systems of Linear Equations and Matrices, Continued Math 19620 Outline of Lecture Algorithm for putting a matrix in row reduced echelon form - i.e. Gauss-Jordan Elimination Number of Solutions
More informationAlgorithms for exact (dense) linear algebra
Algorithms for exact (dense) linear algebra Gilles Villard CNRS, Laboratoire LIP ENS Lyon Montagnac-Montpezat, June 3, 2005 Introduction Problem: Study of complexity estimates for basic problems in exact
More information12/1/2015 LINEAR ALGEBRA PRE-MID ASSIGNMENT ASSIGNED BY: PROF. SULEMAN SUBMITTED BY: M. REHAN ASGHAR BSSE 4 ROLL NO: 15126
12/1/2015 LINEAR ALGEBRA PRE-MID ASSIGNMENT ASSIGNED BY: PROF. SULEMAN SUBMITTED BY: M. REHAN ASGHAR Cramer s Rule Solving a physical system of linear equation by using Cramer s rule Cramer s rule is really
More informationDEN: Linear algebra numerical view (GEM: Gauss elimination method for reducing a full rank matrix to upper-triangular
form) Given: matrix C = (c i,j ) n,m i,j=1 ODE and num math: Linear algebra (N) [lectures] c phabala 2016 DEN: Linear algebra numerical view (GEM: Gauss elimination method for reducing a full rank matrix
More informationNote Set 3 Numerical Linear Algebra
Note Set 3 Numerical Linear Algebra 3.1 Overview Numerical linear algebra includes a class of problems which involves multiplying vectors and matrices, solving linear equations, computing the inverse of
More informationMatrix Solutions to Linear Equations
Matrix Solutions to Linear Equations Augmented matrices can be used as a simplified way of writing a system of linear equations. In an augmented matrix, a vertical line is placed inside the matrix to represent
More informationMatrix decompositions
Matrix decompositions Zdeněk Dvořák May 19, 2015 Lemma 1 (Schur decomposition). If A is a symmetric real matrix, then there exists an orthogonal matrix Q and a diagonal matrix D such that A = QDQ T. The
More informationAlgorithms for Fast Linear System Solving and Rank Profile Computation
Algorithms for Fast Linear System Solving and Rank Profile Computation by Shiyun Yang A thesis presented to the University of Waterloo in fulfillment of the thesis requirement for the degree of Master
More informationLU Factorization. Marco Chiarandini. DM559 Linear and Integer Programming. Department of Mathematics & Computer Science University of Southern Denmark
DM559 Linear and Integer Programming LU Factorization Marco Chiarandini Department of Mathematics & Computer Science University of Southern Denmark [Based on slides by Lieven Vandenberghe, UCLA] Outline
More informationFast Matrix Multiplication Over GF3
Fast Matrix Multiplication Over GF3 Matthew Lambert Advised by Dr. B. David Saunders & Dr. Stephen Siegel February 13, 2015 Matrix Multiplication Outline Fast Matrix Multiplication Outline Fast Matrix
More informationProcess Model Formulation and Solution, 3E4
Process Model Formulation and Solution, 3E4 Section B: Linear Algebraic Equations Instructor: Kevin Dunn dunnkg@mcmasterca Department of Chemical Engineering Course notes: Dr Benoît Chachuat 06 October
More informationMATH 2360 REVIEW PROBLEMS
MATH 2360 REVIEW PROBLEMS Problem 1: In (a) (d) below, either compute the matrix product or indicate why it does not exist: ( )( ) 1 2 2 1 (a) 0 1 1 2 ( ) 0 1 2 (b) 0 3 1 4 3 4 5 2 5 (c) 0 3 ) 1 4 ( 1
More informationReview of Matrices and Block Structures
CHAPTER 2 Review of Matrices and Block Structures Numerical linear algebra lies at the heart of modern scientific computing and computational science. Today it is not uncommon to perform numerical computations
More informationLINEAR SYSTEMS (11) Intensive Computation
LINEAR SYSTEMS () Intensive Computation 27-8 prof. Annalisa Massini Viviana Arrigoni EXACT METHODS:. GAUSSIAN ELIMINATION. 2. CHOLESKY DECOMPOSITION. ITERATIVE METHODS:. JACOBI. 2. GAUSS-SEIDEL 2 CHOLESKY
More informationExploiting Fill-in and Fill-out in Gaussian-like Elimination Procedures on the Extended Jacobian Matrix
2nd European Workshop on AD 1 Exploiting Fill-in and Fill-out in Gaussian-like Elimination Procedures on the Extended Jacobian Matrix Andrew Lyons (Vanderbilt U.) / Uwe Naumann (RWTH Aachen) 2nd European
More informationIntroduction to Parallel Programming in OpenMP Dr. Yogish Sabharwal Department of Computer Science & Engineering Indian Institute of Technology, Delhi
Introduction to Parallel Programming in OpenMP Dr. Yogish Sabharwal Department of Computer Science & Engineering Indian Institute of Technology, Delhi Lecture - 33 Parallel LU Factorization So, now, how
More informationSolving Consistent Linear Systems
Solving Consistent Linear Systems Matrix Notation An augmented matrix of a system consists of the coefficient matrix with an added column containing the constants from the right sides of the equations.
More informationIncomplete Cholesky preconditioners that exploit the low-rank property
anapov@ulb.ac.be ; http://homepages.ulb.ac.be/ anapov/ 1 / 35 Incomplete Cholesky preconditioners that exploit the low-rank property (theory and practice) Artem Napov Service de Métrologie Nucléaire, Université
More informationOptimized LU-decomposition with Full Pivot for Small Batched Matrices S3069
Optimized LU-decomposition with Full Pivot for Small Batched Matrices S369 Ian Wainwright High Performance Consulting Sweden ian.wainwright@hpcsweden.se Based on work for GTC 212: 1x speed-up vs multi-threaded
More informationParallelization of Multilevel Preconditioners Constructed from Inverse-Based ILUs on Shared-Memory Multiprocessors
Parallelization of Multilevel Preconditioners Constructed from Inverse-Based ILUs on Shared-Memory Multiprocessors J.I. Aliaga 1 M. Bollhöfer 2 A.F. Martín 1 E.S. Quintana-Ortí 1 1 Deparment of Computer
More informationSolution of Linear Equations
Solution of Linear Equations (Com S 477/577 Notes) Yan-Bin Jia Sep 7, 07 We have discussed general methods for solving arbitrary equations, and looked at the special class of polynomial equations A subclass
More informationConsider the following example of a linear system:
LINEAR SYSTEMS Consider the following example of a linear system: Its unique solution is x + 2x 2 + 3x 3 = 5 x + x 3 = 3 3x + x 2 + 3x 3 = 3 x =, x 2 = 0, x 3 = 2 In general we want to solve n equations
More information18.06SC Final Exam Solutions
18.06SC Final Exam Solutions 1 (4+7=11 pts.) Suppose A is 3 by 4, and Ax = 0 has exactly 2 special solutions: 1 2 x 1 = 1 and x 2 = 1 1 0 0 1 (a) Remembering that A is 3 by 4, find its row reduced echelon
More informationFast, deterministic computation of the Hermite normal form and determinant of a polynomial matrix
Fast, deterministic computation of the Hermite normal form and determinant of a polynomial matrix George Labahn, Vincent Neiger, Wei Zhou To cite this version: George Labahn, Vincent Neiger, Wei Zhou.
More informationReview Questions REVIEW QUESTIONS 71
REVIEW QUESTIONS 71 MATLAB, is [42]. For a comprehensive treatment of error analysis and perturbation theory for linear systems and many other problems in linear algebra, see [126, 241]. An overview of
More informationPivoting. Reading: GV96 Section 3.4, Stew98 Chapter 3: 1.3
Pivoting Reading: GV96 Section 3.4, Stew98 Chapter 3: 1.3 In the previous discussions we have assumed that the LU factorization of A existed and the various versions could compute it in a stable manner.
More informationLinear Systems. Math A Bianca Santoro. September 23, 2016
Linear Systems Math A4600 - Bianca Santoro September 3, 06 Goal: Understand how to solve Ax = b. Toy Model: Let s study the following system There are two nice ways of thinking about this system: x + y
More informationMath 102, Winter Final Exam Review. Chapter 1. Matrices and Gaussian Elimination
Math 0, Winter 07 Final Exam Review Chapter. Matrices and Gaussian Elimination { x + x =,. Different forms of a system of linear equations. Example: The x + 4x = 4. [ ] [ ] [ ] vector form (or the column
More informationCME 302: NUMERICAL LINEAR ALGEBRA FALL 2005/06 LECTURE 6
CME 302: NUMERICAL LINEAR ALGEBRA FALL 2005/06 LECTURE 6 GENE H GOLUB Issues with Floating-point Arithmetic We conclude our discussion of floating-point arithmetic by highlighting two issues that frequently
More informationMH1200 Final 2014/2015
MH200 Final 204/205 November 22, 204 QUESTION. (20 marks) Let where a R. A = 2 3 4, B = 2 3 4, 3 6 a 3 6 0. For what values of a is A singular? 2. What is the minimum value of the rank of A over all a
More informationDense LU factorization and its error analysis
Dense LU factorization and its error analysis Laura Grigori INRIA and LJLL, UPMC February 2016 Plan Basis of floating point arithmetic and stability analysis Notation, results, proofs taken from [N.J.Higham,
More informationLinear Equations and Matrix
1/60 Chia-Ping Chen Professor Department of Computer Science and Engineering National Sun Yat-sen University Linear Algebra Gaussian Elimination 2/60 Alpha Go Linear algebra begins with a system of linear
More informationSaving Energy in the LU Factorization with Partial Pivoting on Multi-Core Processors
20th Euromicro International Conference on Parallel, Distributed and Network-Based Special Session on Energy-aware Systems Saving Energy in the on Multi-Core Processors Pedro Alonso 1, Manuel F. Dolz 2,
More informationSolving Sparse Rational Linear Systems. Pascal Giorgi. University of Waterloo (Canada) / University of Perpignan (France) joint work with
Solving Sparse Rational Linear Systems Pascal Giorgi University of Waterloo (Canada) / University of Perpignan (France) joint work with A. Storjohann, M. Giesbrecht (University of Waterloo), W. Eberly
More informationJ.I. Aliaga 1 M. Bollhöfer 2 A.F. Martín 1 E.S. Quintana-Ortí 1. March, 2009
Parallel Preconditioning of Linear Systems based on ILUPACK for Multithreaded Architectures J.I. Aliaga M. Bollhöfer 2 A.F. Martín E.S. Quintana-Ortí Deparment of Computer Science and Engineering, Univ.
More informationComputing least squares condition numbers on hybrid multicore/gpu systems
Computing least squares condition numbers on hybrid multicore/gpu systems M. Baboulin and J. Dongarra and R. Lacroix Abstract This paper presents an efficient computation for least squares conditioning
More informationAccelerating linear algebra computations with hybrid GPU-multicore systems.
Accelerating linear algebra computations with hybrid GPU-multicore systems. Marc Baboulin INRIA/Université Paris-Sud joint work with Jack Dongarra (University of Tennessee and Oak Ridge National Laboratory)
More informationCS412: Lecture #17. Mridul Aanjaneya. March 19, 2015
CS: Lecture #7 Mridul Aanjaneya March 9, 5 Solving linear systems of equations Consider a lower triangular matrix L: l l l L = l 3 l 3 l 33 l n l nn A procedure similar to that for upper triangular systems
More informationLinear Algebra Primer
Linear Algebra Primer David Doria daviddoria@gmail.com Wednesday 3 rd December, 2008 Contents Why is it called Linear Algebra? 4 2 What is a Matrix? 4 2. Input and Output.....................................
More informationMAC Module 2 Systems of Linear Equations and Matrices II. Learning Objectives. Upon completing this module, you should be able to :
MAC 0 Module Systems of Linear Equations and Matrices II Learning Objectives Upon completing this module, you should be able to :. Find the inverse of a square matrix.. Determine whether a matrix is invertible..
More informationLinear algebra I Homework #1 due Thursday, Oct Show that the diagonals of a square are orthogonal to one another.
Homework # due Thursday, Oct. 0. Show that the diagonals of a square are orthogonal to one another. Hint: Place the vertices of the square along the axes and then introduce coordinates. 2. Find the equation
More information(17) (18)
Module 4 : Solving Linear Algebraic Equations Section 3 : Direct Solution Techniques 3 Direct Solution Techniques Methods for solving linear algebraic equations can be categorized as direct and iterative
More informationChapter 1: Systems of Linear Equations and Matrices
: Systems of Linear Equations and Matrices Multiple Choice Questions. Which of the following equations is linear? (A) x + 3x 3 + 4x 4 3 = 5 (B) 3x x + x 3 = 5 (C) 5x + 5 x x 3 = x + cos (x ) + 4x 3 = 7.
More informationComputational Linear Algebra
Computational Linear Algebra PD Dr. rer. nat. habil. Ralf Peter Mundani Computation in Engineering / BGU Scientific Computing in Computer Science / INF Winter Term 2017/18 Part 2: Direct Methods PD Dr.
More informationEigenvalues and Eigenvectors
5 Eigenvalues and Eigenvectors 5.2 THE CHARACTERISTIC EQUATION DETERMINANATS n n Let A be an matrix, let U be any echelon form obtained from A by row replacements and row interchanges (without scaling),
More informationSolving Ax = b, an overview. Program
Numerical Linear Algebra Improving iterative solvers: preconditioning, deflation, numerical software and parallelisation Gerard Sleijpen and Martin van Gijzen November 29, 27 Solving Ax = b, an overview
More informationElementary Linear Algebra
Matrices J MUSCAT Elementary Linear Algebra Matrices Definition Dr J Muscat 2002 A matrix is a rectangular array of numbers, arranged in rows and columns a a 2 a 3 a n a 2 a 22 a 23 a 2n A = a m a mn We
More informationHW2 - Due 01/30. Each answer must be mathematically justified. Don t forget your name.
HW2 - Due 0/30 Each answer must be mathematically justified. Don t forget your name. Problem. Use the row reduction algorithm to find the inverse of the matrix 0 0, 2 3 5 if it exists. Double check your
More informationCheat Sheet for MATH461
Cheat Sheet for MATH46 Here is the stuff you really need to remember for the exams Linear systems Ax = b Problem: We consider a linear system of m equations for n unknowns x,,x n : For a given matrix A
More informationGAUSSIAN ELIMINATION AND LU DECOMPOSITION (SUPPLEMENT FOR MA511)
GAUSSIAN ELIMINATION AND LU DECOMPOSITION (SUPPLEMENT FOR MA511) D. ARAPURA Gaussian elimination is the go to method for all basic linear classes including this one. We go summarize the main ideas. 1.
More informationComputational Techniques Prof. Sreenivas Jayanthi. Department of Chemical Engineering Indian institute of Technology, Madras
Computational Techniques Prof. Sreenivas Jayanthi. Department of Chemical Engineering Indian institute of Technology, Madras Module No. # 05 Lecture No. # 24 Gauss-Jordan method L U decomposition method
More informationUtilisation de la compression low-rank pour réduire la complexité du solveur PaStiX
Utilisation de la compression low-rank pour réduire la complexité du solveur PaStiX 26 Septembre 2018 - JCAD 2018 - Lyon Grégoire Pichon, Mathieu Faverge, Pierre Ramet, Jean Roman Outline 1. Context 2.
More informationReview. Example 1. Elementary matrices in action: (a) a b c. d e f = g h i. d e f = a b c. a b c. (b) d e f. d e f.
Review Example. Elementary matrices in action: (a) 0 0 0 0 a b c d e f = g h i d e f 0 0 g h i a b c (b) 0 0 0 0 a b c d e f = a b c d e f 0 0 7 g h i 7g 7h 7i (c) 0 0 0 0 a b c a b c d e f = d e f 0 g
More informationBlack Box Linear Algebra
Black Box Linear Algebra An Introduction to Wiedemann s Approach William J. Turner Department of Mathematics & Computer Science Wabash College Symbolic Computation Sometimes called Computer Algebra Symbols
More informationDETERMINANTS DEFINED BY ROW OPERATIONS
DETERMINANTS DEFINED BY ROW OPERATIONS TERRY A. LORING. DETERMINANTS DEFINED BY ROW OPERATIONS Determinants of square matrices are best understood in terms of row operations, in my opinion. Most books
More information30.3. LU Decomposition. Introduction. Prerequisites. Learning Outcomes
LU Decomposition 30.3 Introduction In this Section we consider another direct method for obtaining the solution of systems of equations in the form AX B. Prerequisites Before starting this Section you
More informationDense Linear Algebra over Finite Fields: the FFLAS and FFPACK packages
Dense Linear Algebra over Finite Fields: the FFLAS and FFPACK packages Jean-Guillaume Dumas, Thierry Gautier, Pascal Giorgi, Clément Pernet To cite this version: Jean-Guillaume Dumas, Thierry Gautier,
More informationToward High Performance Matrix Multiplication for Exact Computation
Toward High Performance Matrix Multiplication for Exact Computation Pascal Giorgi Joint work with Romain Lebreton (U. Waterloo) Funded by the French ANR project HPAC Séminaire CASYS - LJK, April 2014 Motivations
More informationhal , version 1-29 Feb 2008
Compressed Modular Matrix Multiplication Jean-Guillaume Dumas Laurent Fousse Bruno Salvy February 29, 2008 Abstract We propose to store several integers modulo a small prime into a single machine word.
More informationParallel Numerical Algorithms
Parallel Numerical Algorithms Chapter 13 Prof. Michael T. Heath Department of Computer Science University of Illinois at Urbana-Champaign CS 554 / CSE 512 Michael T. Heath Parallel Numerical Algorithms
More informationChapter 4. Solving Systems of Equations. Chapter 4
Solving Systems of Equations 3 Scenarios for Solutions There are three general situations we may find ourselves in when attempting to solve systems of equations: 1 The system could have one unique solution.
More informationThis can be accomplished by left matrix multiplication as follows: I
1 Numerical Linear Algebra 11 The LU Factorization Recall from linear algebra that Gaussian elimination is a method for solving linear systems of the form Ax = b, where A R m n and bran(a) In this method
More informationAMS 209, Fall 2015 Final Project Type A Numerical Linear Algebra: Gaussian Elimination with Pivoting for Solving Linear Systems
AMS 209, Fall 205 Final Project Type A Numerical Linear Algebra: Gaussian Elimination with Pivoting for Solving Linear Systems. Overview We are interested in solving a well-defined linear system given
More information22A-2 SUMMER 2014 LECTURE 5
A- SUMMER 0 LECTURE 5 NATHANIEL GALLUP Agenda Elimination to the identity matrix Inverse matrices LU factorization Elimination to the identity matrix Previously, we have used elimination to get a system
More informationFast Matrix Multiplication*
mathematics of computation, volume 28, number 125, January, 1974 Triangular Factorization and Inversion by Fast Matrix Multiplication* By James R. Bunch and John E. Hopcroft Abstract. The fast matrix multiplication
More informationHomework 2 Foundations of Computational Math 2 Spring 2019
Homework 2 Foundations of Computational Math 2 Spring 2019 Problem 2.1 (2.1.a) Suppose (v 1,λ 1 )and(v 2,λ 2 ) are eigenpairs for a matrix A C n n. Show that if λ 1 λ 2 then v 1 and v 2 are linearly independent.
More informationMAC Learning Objectives. Learning Objectives (Cont.) Module 10 System of Equations and Inequalities II
MAC 1140 Module 10 System of Equations and Inequalities II Learning Objectives Upon completing this module, you should be able to 1. represent systems of linear equations with matrices. 2. transform a
More informationMAC1105-College Algebra. Chapter 5-Systems of Equations & Matrices
MAC05-College Algebra Chapter 5-Systems of Equations & Matrices 5. Systems of Equations in Two Variables Solving Systems of Two Linear Equations/ Two-Variable Linear Equations A system of equations is
More informationNumerical Linear Algebra
Chapter 3 Numerical Linear Algebra We review some techniques used to solve Ax = b where A is an n n matrix, and x and b are n 1 vectors (column vectors). We then review eigenvalues and eigenvectors and
More informationMutantXL: Solving Multivariate Polynomial Equations for Cryptanalysis
MutantXL: Solving Multivariate Polynomial Equations for Cryptanalysis Johannes Buchmann 1, Jintai Ding 2, Mohamed Saied Emam Mohamed 1, and Wael Said Abd Elmageed Mohamed 1 1 TU Darmstadt, FB Informatik
More informationQR Factorization of Tall and Skinny Matrices in a Grid Computing Environment
QR Factorization of Tall and Skinny Matrices in a Grid Computing Environment Emmanuel AGULLO (INRIA / LaBRI) Camille COTI (Iowa State University) Jack DONGARRA (University of Tennessee) Thomas HÉRAULT
More informationIntroduction to communication avoiding algorithms for direct methods of factorization in Linear Algebra
Introduction to communication avoiding algorithms for direct methods of factorization in Linear Algebra Laura Grigori Abstract Modern, massively parallel computers play a fundamental role in a large and
More information