3
$\begingroup$

I would like to integrate over a dilationed region in higher dimensions.

In 3 dimensions it works well.

(*Nintegrate over a 3 dim region*)
regS = Simplex[{{1, 0, 0}, {0, 1, 0}, {0, 0, 
     1}, {-(1/3), -(1/3), -(1/3)}}];
regSdil = RegionDilation[regS, 0.1];
NIntegrate[1, {x, y, z} \[Element] regS]  (*0.333*)
NIntegrate[1, {x, y, z} \[Element] regSe] (*0.764*)

But in 4 dimensions I got the following error message: "Invalid integration variable or limit(s)"

(*Nintegrate over a 4 dim region*)
regS = Simplex[{{1, 0, 0, 0}, {0, 1, 0, 0}, {0, 0, 1, 0}, {0, 0, 0, 
     1}, {1/4 (1 - Sqrt[5]), 1/4 (1 - Sqrt[5]), 1/4 (1 - Sqrt[5]), 
     1/4 (1 - Sqrt[5])}}];
regSdil = RegionDilation[regS, 0.1];
NIntegrate[1, {x, y, z, k} \[Element] regS]  (*0.0931*) 
NIntegrate[1, {x, y, z, k} \[Element] 
  regSdil] (***"Invalid integration variable or limit(s)"***)

I don't know why. Is there a way to solve this. Or is there a better way to do such a calculation?

Improve this question
$\endgroup$
1
  • 1
    $\begingroup$ The problem is with RegionDilation, not the integration. If you evaluate RegionQ[regSdil], you'll get False. So RegionDilation does not produce a 4-dimensional region. It works in simpler 4D cases, e.g., RegionDilation[Ball[{0,0,0,0}],0.1] gives a larger 4D ball. $\endgroup$
    – tad
    Commented Jan 23 at 21:46

1 Answer 1

Reset to default
6
$\begingroup$
  • RegionDilation seems only work for simple region in higher dimensions.
RegionDilation[Ball[{0, 0, 0, 0}, 1], .1]
RegionDilation[Ball[{0, 0, 0, 0}, 1], Ball[{0, 0, 0, 0}, .1]]

Ball[{0, 0, 0, 0}, 1.1]

  • We try to add sphere points around the 5 points of the simplex regS and construct a ConvexHullRegion and integrate on such dilation region of regS.

  • Version 13.3.1, Version 14.0.0

Clear["Global`*"];
regS = Simplex[{{1, 0, 0, 0}, {0, 1, 0, 0}, {0, 0, 1, 0}, {0, 0, 0, 
     1}, {1/4   (1 - Sqrt[5]), 1/4   (1 - Sqrt[5]), 
     1/4   (1 - Sqrt[5]), 1/4   (1 - Sqrt[5])}}];
pts = regS[[1]];
r = .1;
regdilation = 
  ConvexHullRegion[
   Flatten[Outer[Plus, pts, RandomPoint[Sphere[{0, 0, 0, 0}, r], 200],
      1], 1]];
ineqs = First@regdilation /. 
   Thread[{Indexed[\[FormalX], {1}], Indexed[\[FormalX], {2}], 
      Indexed[\[FormalX], {3}], Indexed[\[FormalX], {4}]} -> {x, y, z,
       w}];
NIntegrate[
 Boole[ineqs], {x, y, z, w} ∈ 
  Cuboid @@ 
   Transpose[{First@# - 2 r, Last@# +2 r} & /@ RegionBounds[regS]]]

0.264839.

  • Version 11.3 , Version 12.2
Clear["Global`*"];
regS = Simplex[{{1, 0, 0, 0}, {0, 1, 0, 0}, {0, 0, 1, 0}, {0, 0, 0, 
     1}, {1/4   (1 - Sqrt[5]), 1/4   (1 - Sqrt[5]), 
     1/4   (1 - Sqrt[5]), 1/4   (1 - Sqrt[5])}}];
r = .1;
{{x1, x2}, {y1, y2}, {z1, z2}, {w1, 
    w2}} = {First@# - 2 r, Last@# + 2 r} & /@ RegionBounds[regS];
NIntegrate[
 Boole[RegionDistance[regS, {x, y, z, w}] <= r], {x, x1, x2}, {y, y1, 
  y2}, {z, z1, z2}, {w, w1, w2}]

0.277332.

Improve this answer
$\endgroup$
5
  • $\begingroup$ It is a pity that this Integrating a function over a region defined as the epsilon-neighborhood of a triangle cannot be generalized in this case! Any idea why? $\endgroup$
    – Ulrich Neumann
    Commented Jan 24 at 14:02
  • $\begingroup$ @UlrichNeumann I found that theRegionDistance approach work in 11.3 and 12.2 but not in 14.0 version. $\endgroup$
    – cvgmt
    Commented Jan 24 at 15:28
  • $\begingroup$ Thanks! I tried ball4D = ImplicitRegion[ u1^2 + u2^2 + u3^2 + u4^2 <= .1, {u1, u2, u3, u4}] regS = Simplex[{{1, 0, 0, 0}, {0, 1, 0, 0}, {0, 0, 1, 0}, {0, 0, 0, 1}, {1/4 (1 - Sqrt[5]), 1/4 (1 - Sqrt[5]), 1/4 (1 - Sqrt[5]), 1/4 (1 - Sqrt[5])}}]; pts = regS[[1]] sol = Resolve[ Exists[{v1, v2, v3, v4, u1, u2, u3, u4}, {v1, v2, v3, v4} \[Element] regS && {u1, u2, u3, u4} \[Element] ball4D, {x1, x2, x3, x4} == {v1, v2, v3, v4} + {u1, u2, u3, u4}] , Reals] reg = ImplicitRegion[sol, {x1, x2, x3, x4}]; without success $\endgroup$
    – Ulrich Neumann
    Commented Jan 24 at 16:00
  • $\begingroup$ Thanks! I like the idea, how you generate the dilationed region using ConvexHullRegion! Unfortunately the Nintegrate gives me error messages: (1) "Numerical integration converging too slowly; suspect one of the following: singularity, value of the integration is 0, highly oscillatory integrand, or WorkingPrecision too small." (2) "The global error of the strategy GlobalAdaptive has increased more than 2000 times" instead of decreasing." I have tried increasing the number of points, but the integral does not seem to converge. $\endgroup$
    – Zsombor
    Commented Jan 25 at 15:26
  • $\begingroup$ As I increase the number of random points of the spheres, these are the results with errors: 50 points: 0.266 + /- 0.061 100 points: 0.27 + /- 0.051 200 points: 0.282 + /- 0.064 500 points: 0.288 + /- 0.066 Is there any idea how to reduce errors? $\endgroup$
    – Zsombor
    Commented Jan 25 at 15:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.