. 2013 Jan 28;40(2):023301. doi: 10.1118/1.4774361

Table .

Algorithm IV. Implementation of kernel K_srt ⟨⟨⟨ (N_v, N_t), (N_r, 1, 1) ⟩⟩⟩.

Input:

D_g

T_α_{int}

Output:

D_g

\bar{t} = (blockIdx . y) c_{0} Δ_{t} + c_{0} t_{\min}

;

θ^{s} = (threadIdx . x) Δ_{θ^{s}} + θ_{\min}^{s}

;

ϕ^{s} = (blockIdx . x) Δ_{ϕ^{s}} + ϕ_{\min}^{s}

θ_{\max}^{'} = \arccos (({(R^{s})}^{2} + {\bar{t}}^{2} - R^{2}) / (2 \bar{t} R^{s}))

3: Σ = 0;

θ^{'} = θ_{\max}^{'}

4: while θ^′ > 0 do

z^{'} = \bar{t} \cos θ^{'}

;

r^{'} = \bar{t} \sin θ^{'}

; ϕ^′ = 0

6: while ϕ^′ < 2π do

7: x^′ = r^′cos ϕ^′; y^′ = r^′sin ϕ^′

8: x = −x^′sin θ^′ − (z^′ − R^s)cos θ^′; y = y^′; z = x^′cos θ^′ − (z^′ − R^s)sin θ^′ {Convert to global coordinate system}

9: x_n = (x − x_min)/Δ_s; y_n = (y − y_min)/Δ_s; z_n = (z − z_min)/Δ_s

10: Σ + = tex3D(x_n, y_n, z_n) {Tri-linear interpolation}

11: ϕ^′ = ϕ^′ + Δ_s/r^′

12: endwhile

13:

θ^{'} = θ^{'} - Δ_{s} / \bar{t}

14: endwhile

15:

D_g [threadIdx . x] [blockIdx . x] [blockIdx . y] = Σ Δ_{s}^{2}