lattice/latticegpu.jl
1
0
Fork 0
mirror of https://igit.ific.uv.es/alramos/latticegpu.jl.git synced 2025-09-18 12:17:14 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Small optimization in OR

Browse source
This commit is contained in:
Alberto Ramos 2025-09-04 17:37:58 +02:00
commit eabbfe6307
1 changed files with 22 additions and 27 deletions
Download patch file Download diff file Expand all files Collapse all files

49
src/YM/YMupdate.jl
View file

@ -37,9 +37,7 @@ function krnl_or_wilson!(U::AbstractArray{SU3{T}}, id, icl, ztw, lp::SpaceParm)
clr = point_color((b, r), lp) clr = point_color((b, r), lp)
if clr % 2 == icl if clr % 2 == icl
stp = sum_staples(U, id, (b, r), ztw, lp) stp = sum_staples(U, id, (b, r), ztw, lp)
M = convert(M3x3{T},U[b,id,r]) U[b,id,r] = do_or(U[b,id,r], stp)
M = do_or(M, stp)
U[b,id,r] = SU3{T}(M.u11, M.u12, M.u13, M.u21, M.u22, M.u23)
end end
return nothing return nothing
@ -96,35 +94,36 @@ end
return stp return stp
end end
@inline function do_or(M::M3x3{T}, stp) where {T} @inline function do_or(U::SU3{T}, stp) where {T}
M = do_or12(M, stp) M = do_or12(U, stp)
M = do_or13(M, stp) M = do_or13(M, stp)
M = do_or23(M, stp) U = do_or23(M, stp)
U = unitarize(U)
return M
return U
end end
@inline function do_or12(M::M3x3{T}, stp) where {T} @inline function do_or12(U::SU3{T}, stp) where {T}
w11 = M.u11*stp.u11 + M.u12*stp.u21 + M.u13*stp.u31 w11 = U.u11*stp.u11 + U.u12*stp.u21 + U.u13*stp.u31
w12 = M.u11*stp.u12 + M.u12*stp.u22 + M.u13*stp.u32 w12 = U.u11*stp.u12 + U.u12*stp.u22 + U.u13*stp.u32
w22 = M.u21*stp.u12 + M.u22*stp.u22 + M.u23*stp.u32 w22 = U.u21*stp.u12 + U.u22*stp.u22 + U.u23*stp.u32
w21 = M.u21*stp.u11 + M.u22*stp.u21 + M.u23*stp.u31 w21 = U.u21*stp.u11 + U.u22*stp.u21 + U.u23*stp.u31
w = tuple(complex(real(w11+w22)/2, imag(w11-w22)/2), complex(real(w12-w21)/2,imag(w12+w21)/2)) w = tuple(complex(real(w11+w22)/2, imag(w11-w22)/2), complex(real(w12-w21)/2,imag(w12+w21)/2))
h = overrelax_cd(w) h = overrelax_cd(w)
u11 = h[1]*M.u11 + h[2]*M.u21 u11 = h[1]*U.u11 + h[2]*U.u21
u12 = h[1]*M.u12 + h[2]*M.u22 u12 = h[1]*U.u12 + h[2]*U.u22
u13 = h[1]*M.u13 + h[2]*M.u23 u13 = h[1]*U.u13 + h[2]*U.u23
u21 = -conj(h[2])*M.u11 + conj(h[1])*M.u21 u21 = -conj(h[2])*U.u11 + conj(h[1])*U.u21
u22 = -conj(h[2])*M.u12 + conj(h[1])*M.u22 u22 = -conj(h[2])*U.u12 + conj(h[1])*U.u22
u23 = -conj(h[2])*M.u13 + conj(h[1])*M.u23 u23 = -conj(h[2])*U.u13 + conj(h[1])*U.u23
u31 = M.u31 u31 = conj(U.u12*U.u23 - U.u13*U.u22)
u32 = M.u32 u32 = conj(U.u13*U.u21 - U.u11*U.u23)
u33 = M.u33 u33 = conj(U.u11*U.u22 - U.u12*U.u21)
return M3x3{T}(u11, u12, u13, u21, u22, u23, u31, u32, u33) return M3x3{T}(u11, u12, u13, u21, u22, u23, u31, u32, u33)
end end
@ -170,11 +169,7 @@ end
u22 = h[1]*M.u22 + h[2]*M.u32 u22 = h[1]*M.u22 + h[2]*M.u32
u23 = h[1]*M.u23 + h[2]*M.u33 u23 = h[1]*M.u23 + h[2]*M.u33
u31 = -conj(h[2])*M.u21 + conj(h[1])*M.u31 return SU3{T}(u11, u12, u13, u21, u22, u23)
u32 = -conj(h[2])*M.u22 + conj(h[1])*M.u32
u33 = -conj(h[2])*M.u23 + conj(h[1])*M.u33
return M3x3{T}(u11, u12, u13, u21, u22, u23, u31, u32, u33)
end end