Description of fast matrix multiplication algorithm: ⟨3×5×8:93⟩

Algorithm type

2⁢X2⁢Y3⁢Z2+2⁢X3⁢Y⁢Z2+23⁢X2⁢Y2⁢Z2+2⁢X2⁢Y⁢Z3+2⁢X⁢Y4⁢Z+3⁢X⁢Y2⁢Z3+8⁢X3⁢Y⁢Z+4⁢X2⁢Y2⁢Z+5⁢X⁢Y3⁢Z+4⁢X⁢Y⁢Z3+14⁢X2⁢Y⁢Z+8⁢X⁢Y2⁢Z+16⁢X⁢Y⁢Z2X2Y3Z22X3YZ223X2Y2Z22X2YZ32XY4Z3XY2Z38X3YZ4X2Y2Z5XY3Z4XYZ314X2YZ8XY2Z16XYZ2*X^2*Y^3*Z^2+2*X^3*Y*Z^2+23*X^2*Y^2*Z^2+2*X^2*Y*Z^3+2*X*Y^4*Z+3*X*Y^2*Z^3+8*X^3*Y*Z+4*X^2*Y^2*Z+5*X*Y^3*Z+4*X*Y*Z^3+14*X^2*Y*Z+8*X*Y^2*Z+16*X*Y*Z

Algorithm definition

The algorithm ⟨3×5×8:93⟩ could be constructed using the following decomposition:

⟨3×5×8:93⟩ = ⟨3×5×2:25⟩ + ⟨3×5×6:68⟩.

This decomposition is defined by the following equality:

Trace⁡Mul⁡A_1_1A_1_2A_1_3A_1_4A_1_5A_2_1A_2_2A_2_3A_2_4A_2_5A_3_1A_3_2A_3_3A_3_4A_3_5B_1_1B_1_2B_1_3B_1_4B_1_5B_1_6B_1_7B_1_8B_2_1B_2_2B_2_3B_2_4B_2_5B_2_6B_2_7B_2_8B_3_1B_3_2B_3_3B_3_4B_3_5B_3_6B_3_7B_3_8B_4_1B_4_2B_4_3B_4_4B_4_5B_4_6B_4_7B_4_8B_5_1B_5_2B_5_3B_5_4B_5_5B_5_6B_5_7B_5_8C_1_1C_1_2C_1_3C_2_1C_2_2C_2_3C_3_1C_3_2C_3_3C_4_1C_4_2C_4_3C_5_1C_5_2C_5_3C_6_1C_6_2C_6_3C_7_1C_7_2C_7_3C_8_1C_8_2C_8_3=Trace⁡Mul⁡A_1_1A_1_2A_1_3A_1_4A_1_5A_2_1A_2_2A_2_3A_2_4A_2_5A_3_1A_3_2A_3_3A_3_4A_3_5B_1_1B_1_2B_2_1B_2_2B_3_1B_3_2B_4_1B_4_2B_5_1B_5_2C_1_1C_1_2C_1_3C_2_1C_2_2C_2_3+Trace⁡Mul⁡A_1_1A_1_2A_1_3A_1_4A_1_5A_2_1A_2_2A_2_3A_2_4A_2_5A_3_1A_3_2A_3_3A_3_4A_3_5B_1_3B_1_4B_1_5B_1_6B_1_7B_1_8B_2_3B_2_4B_2_5B_2_6B_2_7B_2_8B_3_3B_3_4B_3_5B_3_6B_3_7B_3_8B_4_3B_4_4B_4_5B_4_6B_4_7B_4_8B_5_3B_5_4B_5_5B_5_6B_5_7B_5_8C_3_1C_3_2C_3_3C_4_1C_4_2C_4_3C_5_1C_5_2C_5_3C_6_1C_6_2C_6_3C_7_1C_7_2C_7_3C_8_1C_8_2C_8_3TraceMulA_1_1A_1_2A_1_3A_1_4A_1_5A_2_1A_2_2A_2_3A_2_4A_2_5A_3_1A_3_2A_3_3A_3_4A_3_5B_1_1B_1_2B_1_3B_1_4B_1_5B_1_6B_1_7B_1_8B_2_1B_2_2B_2_3B_2_4B_2_5B_2_6B_2_7B_2_8B_3_1B_3_2B_3_3B_3_4B_3_5B_3_6B_3_7B_3_8B_4_1B_4_2B_4_3B_4_4B_4_5B_4_6B_4_7B_4_8B_5_1B_5_2B_5_3B_5_4B_5_5B_5_6B_5_7B_5_8C_1_1C_1_2C_1_3C_2_1C_2_2C_2_3C_3_1C_3_2C_3_3C_4_1C_4_2C_4_3C_5_1C_5_2C_5_3C_6_1C_6_2C_6_3C_7_1C_7_2C_7_3C_8_1C_8_2C_8_3TraceMulA_1_1A_1_2A_1_3A_1_4A_1_5A_2_1A_2_2A_2_3A_2_4A_2_5A_3_1A_3_2A_3_3A_3_4A_3_5B_1_1B_1_2B_2_1B_2_2B_3_1B_3_2B_4_1B_4_2B_5_1B_5_2C_1_1C_1_2C_1_3C_2_1C_2_2C_2_3TraceMulA_1_1A_1_2A_1_3A_1_4A_1_5A_2_1A_2_2A_2_3A_2_4A_2_5A_3_1A_3_2A_3_3A_3_4A_3_5B_1_3B_1_4B_1_5B_1_6B_1_7B_1_8B_2_3B_2_4B_2_5B_2_6B_2_7B_2_8B_3_3B_3_4B_3_5B_3_6B_3_7B_3_8B_4_3B_4_4B_4_5B_4_6B_4_7B_4_8B_5_3B_5_4B_5_5B_5_6B_5_7B_5_8C_3_1C_3_2C_3_3C_4_1C_4_2C_4_3C_5_1C_5_2C_5_3C_6_1C_6_2C_6_3C_7_1C_7_2C_7_3C_8_1C_8_2C_8_3Trace(Mul(Matrix(3, 5, [[A_1_1,A_1_2,A_1_3,A_1_4,A_1_5],[A_2_1,A_2_2,A_2_3,A_2_4,A_2_5],[A_3_1,A_3_2,A_3_3,A_3_4,A_3_5]]),Matrix(5, 8, [[B_1_1,B_1_2,B_1_3,B_1_4,B_1_5,B_1_6,B_1_7,B_1_8],[B_2_1,B_2_2,B_2_3,B_2_4,B_2_5,B_2_6,B_2_7,B_2_8],[B_3_1,B_3_2,B_3_3,B_3_4,B_3_5,B_3_6,B_3_7,B_3_8],[B_4_1,B_4_2,B_4_3,B_4_4,B_4_5,B_4_6,B_4_7,B_4_8],[B_5_1,B_5_2,B_5_3,B_5_4,B_5_5,B_5_6,B_5_7,B_5_8]]),Matrix(8, 3, [[C_1_1,C_1_2,C_1_3],[C_2_1,C_2_2,C_2_3],[C_3_1,C_3_2,C_3_3],[C_4_1,C_4_2,C_4_3],[C_5_1,C_5_2,C_5_3],[C_6_1,C_6_2,C_6_3],[C_7_1,C_7_2,C_7_3],[C_8_1,C_8_2,C_8_3]]))) = Trace(Mul(Matrix(3, 5, [[A_1_1,A_1_2,A_1_3,A_1_4,A_1_5],[A_2_1,A_2_2,A_2_3,A_2_4,A_2_5],[A_3_1,A_3_2,A_3_3,A_3_4,A_3_5]]),Matrix(5, 2, [[B_1_1,B_1_2],[B_2_1,B_2_2],[B_3_1,B_3_2],[B_4_1,B_4_2],[B_5_1,B_5_2]]),Matrix(2, 3, [[C_1_1,C_1_2,C_1_3],[C_2_1,C_2_2,C_2_3]])))+Trace(Mul(Matrix(3, 5, [[A_1_1,A_1_2,A_1_3,A_1_4,A_1_5],[A_2_1,A_2_2,A_2_3,A_2_4,A_2_5],[A_3_1,A_3_2,A_3_3,A_3_4,A_3_5]]),Matrix(5, 6, [[B_1_3,B_1_4,B_1_5,B_1_6,B_1_7,B_1_8],[B_2_3,B_2_4,B_2_5,B_2_6,B_2_7,B_2_8],[B_3_3,B_3_4,B_3_5,B_3_6,B_3_7,B_3_8],[B_4_3,B_4_4,B_4_5,B_4_6,B_4_7,B_4_8],[B_5_3,B_5_4,B_5_5,B_5_6,B_5_7,B_5_8]]),Matrix(6, 3, [[C_3_1,C_3_2,C_3_3],[C_4_1,C_4_2,C_4_3],[C_5_1,C_5_2,C_5_3],[C_6_1,C_6_2,C_6_3],[C_7_1,C_7_2,C_7_3],[C_8_1,C_8_2,C_8_3]])))

N.B.: for any matrices A, B and C such that the expression Tr(Mul(A,B,C)) is defined, one can construct several trilinear homogeneous polynomials P(A,B,C) such that P(A,B,C)=Tr(Mul(A,B,C)) (P(A,B,C) variables are A,B and C's coefficients). Each trilinear P expression encodes a matrix multiplication algorithm: the coefficient in C_i_j of P(A,B,C) is the (i,j)-th entry of the matrix product Mul(A,B)=Transpose(C).

Algorithm description

These encodings are given in compressed text format using the maple computer algebra system. In each cases, the last line could be understood as a description of the encoding with respect to classical matrix multiplication algorithm. As these outputs are structured, one can construct easily a parser to its favorite format using the maple documentation without this software.

Back to main table