Skip to main content

compile - How to boost the performance of my own BSplineFunction


In my CAGD package, I implemented a CAGDBSplineFunction[] like built-in BSplineFunction[]. Here is a performance comparison:


pts = Table[{ Cos[2 Pi u/6] Cos[v], Sin[2 Pi u/6] Cos[v], v}, {u, 6}, {v, -1, 1, 1/2}];
f = BSplineFunction[pts]
g = CAGDBSplineFunction[pts]

ParametricPlot3D[f[u, v], {u, 0, 1}, {v, 0, 1}] // AbsoluteTiming
ParametricPlot3D[g[u, v], {u, 0, 1}, {v, 0, 1}] // AbsoluteTiming


enter image description here


Obviously, my CAGDBSplineFunction[] is far slower(about 20X) than the built-in BSplineFunction[]


The CAGDBSplineFunction[] mainly uses an auxiliary function functionalNonzeroBasis[] to compute the coordinate of a parameter pair $(u,v)$.


To improve the performance, I refactor the functionalNonzeroBasis[] to compiledNonzeroBasis[]


searchSpan[{deg_, knots_}, u_] := 
With[{un = knots[[-(deg + 1)]]},
If[u == un,
Position[knots, un][[1, 1]] - 2,
Ordering[UnitStep[u - knots], 1][[1]] - 2

]
]

coeff[u_, U_, i_, p_] :=
If[U[[i + p + 1]] != U[[i + 1]],
(u - U[[i + 1]])/(U[[i + p + 1]] - U[[i + 1]]), 0]

compiledNonzeroBasis=
ReleaseHold[
Hold@Compile[{{i, _Integer}, {p, _Integer}, {u, _Real, 1}, {u0, _Real}},

Module[
{lst = Table[0., {p + 2}, {p + 1}], cc = i - 2},
lst[[2, 1]] = 1.0;
Do[
lst[[m + n - cc, n + 1]] =
coeff[u0, u, m, n] lst[[m + n - 1 - cc, n]] + (1 -
coeff[u0, u, m + 1, n]) lst[[m + n - cc, n]],
{n, 1, p}, {m, i - n, i}
];
lst // Transpose // Last // Rest

], CompilationTarget -> "C", RuntimeOptions -> "Speed"
] /. DownValues[coeff]
]

compiledBSplineSurf[
ctrlnets_, {{deg1_, knots1_}, {deg2_, knots2_}},
u_?NumericQ, v_?NumericQ] :=
Module[{i, j, validnets, row, col},
i = searchSpan[{deg1, knots1}, u];
j = searchSpan[{deg2, knots2}, v];

validnets =
Take[ctrlnets, {i - deg1 + 1, i + 1}, {j - deg2 + 1, j + 1}];
row = compiledNonzeroBasis[i, deg1, knots1, u];
col = compiledNonzeroBasis[j, deg2, knots2, v];
row.Transpose[validnets, {1, 3, 2}].col
]

Test


(*knot vector*)
k1 = {0., 0., 0., 0., 0.333333, 0.666667, 1., 1., 1., 1.};

k2 = {0., 0., 0., 0., 0.5, 1., 1., 1., 1.};

ParametricPlot3D[
compiledBSplineSurf[pts, {{3, k1}, {3, k2}}, u, v],
{u, 0, 1}, {v, 0, 1}] // AbsoluteTiming

This time the performance improves about 2X


enter image description here


Another trial is trying Compile`GetElement


With[{get = Compile`GetElement},

basisCoeff[deg_, U_, i_, u_] :=
If[U[[i + deg + 1]] != U[[i + 1]],
(u - get[U, i + 1])/(get[U, i + deg + 1] - get[U, i + 1]),
(*case for ui=uj*)
0.0
];
optimizedNonzeroBasis =
ReleaseHold[
Hold@Compile[{{i, _Integer}, {deg, _Integer}, {knots, _Real, 1}, {u0, _Real}},
Module[{lst = Table[0., {deg + 2}, {deg + 1}], c = i - 2},

lst[[2, 1]] = 1.0;
Do[
With[{mn = m + n},
lst[[mn - c, n + 1]] =
basisCoeff[n, knots, m, u0] get[lst, mn - c - 1, n] +
(1 - basisCoeff[n, knots, m + 1, u0]) get[lst,
mn - c, n]
],
{n, 1, deg}, {m, i - n, i}
];

Rest@Last[Transpose[lst]
]
], CompilationTarget -> "C", RuntimeOptions -> "Speed"
] /. DownValues[basisCoeff]]
]

ParametricPlot3D[
optimizedBSplineSurf[pts, {{3, k1}, {3, k2}}, u, v],
{u, 0, 1}, {v, 0, 1}] // AbsoluteTiming


enter image description here


So I would like to know:



  • Is it possible to further improve/boost the performance of compiledBSplineSurf[]?




Update:


For the optimizedBSplineSurf[], just replace compiledNonzeroBasis[] with optimizedNonzeroBasis[]



Answer



You just need to fully compile your function:



fullycompiledBSplineSurf = 
Hold@Compile[{{ctrlnets, _Real, 3}, {deg1, _Integer}, {deg2, _Integer},
{knots1, _Real, 1}, {knots2, _Real, 1}, {u, _Real}, {v, _Real}},
Module[{i, j, validnets, row, col},
i = searchSpan[{deg1, knots1}, u];
j = searchSpan[{deg2, knots2}, v];
validnets = Take[ctrlnets, {i - deg1 + 1, i + 1}, {j - deg2 + 1, j + 1}];
row = optimizedNonzeroBasis[i, deg1, knots1, u];
col = optimizedNonzeroBasis[j, deg2, knots2, v];
row.Transpose[validnets, {1, 3, 2}].col

],
CompilationOptions -> {"InlineExternalDefinitions" -> True},
"RuntimeOptions" -> {"EvaluateSymbolically" -> False},
CompilationTarget -> C, RuntimeOptions -> "Speed"
] /. DownValues@searchSpan // ReleaseHold;

ParametricPlot3D[
fullycompiledBSplineSurf[pts, 3, 3, k1, k2, u, v],
{u, 0, 1}, {v, 0, 1}] // AbsoluteTiming


enter image description here


Now the function is about half as fast as the built-in:


ParametricPlot3D[f[u, v], {u, 0, 1}, {v, 0, 1}] // AbsoluteTiming

enter image description here


I think I don't need to make any explanation, because all the techniques above have been used in answering your previous questions.


Comments

Popular posts from this blog

plotting - How to draw lines between specified dots on ListPlot?

I would like to create a plot where I have unconnected dots and some connected. So far, I have figured out how to draw the dots. My code is the following: ListPlot[{{1, 1}, {2, 2}, {3, 3}, {4, 4}, {1, 4}, {2, 5}, {3, 6}, {4, 7}, {1, 7}, {2, 8}, {3, 9}, {4, 10}, {1, 10}, {2, 11}, {3, 12}, {4,13}, {2.5, 7}}, Ticks -> {{1, 2, 3, 4}, None}, AxesStyle -> Thin, TicksStyle -> Directive[Black, Bold, 12], Mesh -> Full] I have thought using ListLinePlot command, but I don't know how to specify to the command to draw only selected lines between the dots. Do have any suggestions/hints on how to do that? Thank you. Answer One possibility would be to use Epilog with Line : ListPlot[ {{1, 1}, {2, 2}, {3, 3}, {4, 4}, {1, 4}, {2, 5}, {3, 6}, {4, 7}, {1, 7}, {2, 8}, {3, 9}, {4, 10}, {1, 10}, {2, 11}, {3, 12}, {4, 13}, {2.5, 7}}, Ticks -> {{1, 2, 3, 4}, None}, AxesStyle -> Thin, TicksStyle -> Directive[Black, Bold, 12], Mesh -> Full, Epilog -> { Line[ ...

calculus and analysis - Integral of the Sinc product

Let us consider the following integral $$ B_n = \int_0^\infty \prod_{k=1,3,5,\dots}^n\frac{\sin (x/k)}{x/k}dx $$ By definition, Sinc[x] == Sin[x]/x , therefore B[n_?OddQ] := Integrate[Product[Sinc[x/k], {k, 1, n, 2}], {x, 0, Infinity}] B /@ Range[1, 13, 2] {Pi/2, Pi/2, Pi/2, Pi/2, Pi/2, Pi/2, Pi/2} OK, everything is fine. But... B[15] 467807924713440738696537864469 Pi/935615849440640907310521750000 What's going on? Answer As Eckhard wrote in comments B[n] is the n-th Borwein integral . (The letter B was not accidental :) ) This funny properties of Borwein integrals is related to the Fourier transform of Sinc function FourierTransform[Sinc[x], x, k] 1/2 Sqrt[Pi/2] (Sign[1 - k] + Sign[1 + k]) Plot[%, {k, -2, 2}, Filling -> 0] which is the box function. The result is $\pi/2$ while the sum $$ 1/3+1/5+\dots+1/n $$ B_n = \frac{\pi}{2} - \pi \bigg(\sum_{k=3,5,\ldots}^n\frac{1}{k} -1\biggr)^\frac{n-1}{2}\prod_{k=3,5,\dots}^n\frac{k}{k-1}. $$ For $n=15$ it is equal to $$ \frac{46780...

list manipulation - Selecting multiple columns from a matrix?

Sample data: data = { {{2013, 1, 1}, 24.13, 167.67, 231.82}, {{2013, 1, 2}, 32.15, 170.92, 225.99}, {{2013, 1, 3}, 35.43, 172.68, 221.67}, {{2013, 1, 4}, 36.73, 173.05, 218.32}, {{2013, 1, 5}, 58.19, 165.96, 197.05}, {{2013, 1, 6}, 69.99, 163.50, 187.52}, {{2013, 1, 7}, 71.37, 154.21, 175.58}, {{2013, 1, 8}, 72.51, 149.66, 163.25}}; I want a DateListPlot with three graphs, so for a matrix formed by columns 1 and 2, one for columns 1 and 3, and 1 for columns 1 and 4. At the moment I'm using this code: data2 = Transpose[{data[[All, 1]], data[[All, 2]]}]; data3 = Transpose[{data[[All, 1]], data[[All, 3]]}]; data4 = Transpose[{data[[All, 1]], data[[All, 4]]}]; DateListPlot[{data2, data3, data4}, Joined -> True, Filling -> {3 -> {1}}] but I have a hunch that this can be done more efficiently. I don't like the Transpose s in particular. Any ideas? edit (for extra credit) What if I need to multiply the second column by 2, which in my solution is simp...