
This paper considers the work of combining the proper orthogonal decomposition (POD) reduced-order method with the discontinuous Galerkin (DG) method to solve three-dimensional time-domain Euler equations. The POD-DG formulation is established by constructing the POD base vector space, based on POD technology one can apply the Galerkin projection of the DG scheme to this dimension reduction space for calculation. Its overall goal is to overcome the disadvantages of high computational cost and memory requirement in the DG algorithm, reduce the degrees of freedom (DOFs) of the calculation model, and save the calculation time while ensuring acceptable accuracy. Numerical experiments verify these advantages of the proposed POD-DG method.
Citation: Lan Zhu, Li Xu, Jun-Hui Yin, Shu-Cheng Huang, Bin Li. A discontinuous Galerkin Method based on POD model reduction for Euler equation[J]. Networks and Heterogeneous Media, 2024, 19(1): 86-105. doi: 10.3934/nhm.2024004
[1] | Emad Solouma, Ibrahim Al-Dayel, Meraj Ali Khan, Youssef A. A. Lazer . Characterization of imbricate-ruled surfaces via rotation-minimizing Darboux frame in Minkowski 3-space E31. AIMS Mathematics, 2024, 9(5): 13028-13042. doi: 10.3934/math.2024635 |
[2] | Ibrahim AL-Dayel, Emad Solouma, Meraj Khan . On geometry of focal surfaces due to B-Darboux and type-2 Bishop frames in Euclidean 3-space. AIMS Mathematics, 2022, 7(7): 13454-13468. doi: 10.3934/math.2022744 |
[3] | A. A. Abdel-Salam, M. I. Elashiry, M. Khalifa Saad . Tubular surface generated by a curve lying on a regular surface and its characterizations. AIMS Mathematics, 2024, 9(5): 12170-12187. doi: 10.3934/math.2024594 |
[4] | Bahar UYAR DÜLDÜL . On some new frames along a space curve and integral curves with Darboux q-vector fields in E3. AIMS Mathematics, 2024, 9(7): 17871-17885. doi: 10.3934/math.2024869 |
[5] | Ayman Elsharkawy, Clemente Cesarano, Abdelrhman Tawfiq, Abdul Aziz Ismail . The non-linear Schrödinger equation associated with the soliton surfaces in Minkowski 3-space. AIMS Mathematics, 2022, 7(10): 17879-17893. doi: 10.3934/math.2022985 |
[6] | Fatemah Mofarreh, Rashad A. Abdel-Baky . Singularities of swept surfaces in Euclidean 3-space. AIMS Mathematics, 2024, 9(9): 26049-26064. doi: 10.3934/math.20241272 |
[7] | Yanlin Li, Ali. H. Alkhaldi, Akram Ali, R. A. Abdel-Baky, M. Khalifa Saad . Investigation of ruled surfaces and their singularities according to Blaschke frame in Euclidean 3-space. AIMS Mathematics, 2023, 8(6): 13875-13888. doi: 10.3934/math.2023709 |
[8] | Masatomo Takahashi, Haiou Yu . On generalised framed surfaces in the Euclidean space. AIMS Mathematics, 2024, 9(7): 17716-17742. doi: 10.3934/math.2024861 |
[9] | Kemal Eren, Soley Ersoy, Mohammad Nazrul Islam Khan . Simultaneous characterizations of alternative partner-ruled surfaces. AIMS Mathematics, 2025, 10(4): 8891-8906. doi: 10.3934/math.2025407 |
[10] | Nadia Alluhaibi . Circular surfaces and singularities in Euclidean 3-space E3. AIMS Mathematics, 2022, 7(7): 12671-12688. doi: 10.3934/math.2022701 |
This paper considers the work of combining the proper orthogonal decomposition (POD) reduced-order method with the discontinuous Galerkin (DG) method to solve three-dimensional time-domain Euler equations. The POD-DG formulation is established by constructing the POD base vector space, based on POD technology one can apply the Galerkin projection of the DG scheme to this dimension reduction space for calculation. Its overall goal is to overcome the disadvantages of high computational cost and memory requirement in the DG algorithm, reduce the degrees of freedom (DOFs) of the calculation model, and save the calculation time while ensuring acceptable accuracy. Numerical experiments verify these advantages of the proposed POD-DG method.
Sweeping surface is the surface swept out by the movement of a plane curve (the profile curve or generatrix) whilst the plane is moved through space in such away that the movement of the plane is always in the direction of the normal to the plane. Sweeping is a very substantial, strongly, and spread method in geometric modelling. The fundamental idea is to select various geometrical object (generators), which is then swept over a spine curve (trajectory) in the space. The result of such evolution, depend on movement through space and intrinsic shape deformation, is a sweep object. The sweep object type is given by the choice of the generator and the trajectory. Additional specifics on the sweeping surfaces can be found in [1,2,3,4]. In recent years, the ownerships of sweeping surfaces and their offset surfaces have been examined in Euclidean and non-Euclidean spaces (see e.g., [5,6,7,8,9]). There are several different names for the sweeping surface in previous written works, namely, tubular surface, pipe surface, string, and canal surface [10,11]. However, to the best of the authors knowledge, we can not find any literature on the study for regarding curves lying in surfaces as the original objects and considering the singularities and convexity of sweeping surfaces generated by these curves. Serve such a need and as the extension of the study[7], the current study focuses our attention on the geometrical properties of sweeping surfaces whose center curves in surfaces in Euclidean 3-space E3.
In this work, in analogous with the well known Bishop and Frenet-Serret frames, we define a new version of Darboux frame using a common tangent vector field to a surface along a curve and call this frame the rotation minimizing Darboux frame (RMDF). By using this frame, we give the parametric representation for a sweeping surface. We also show that the parametric curves on this surface are curvature lines. Then, we study the local singularities and convexity of a sweeping surface. Consequently, the necessary and sufficient condition for the sweeping surface to be a developable ruled surface is derived. In addition, some examples of application are introduced and explained in detail.
The ambient space is the Euclidean 3-space E3, and for our work we have used [10,11] as general references. Let
R(s,u)=(x1(s,u),x2(s,u),x3(s,u)), (s,u)∈D⊆R2, |
represent a regular surface M. The R′s tangent vectors are
Rs(s,u)=∂R∂s, Ru(s,u)=∂R∂u. |
The unit normal vector to the surface M is
u(s,u)=Rs×Ru‖Rs×Ru‖, |
where × denotes the cross product in E3. The metric (first fundamental form) I is defined by
I(s,u)=g11ds2+2g12dsdu+g22du2, |
where g11=<Rs,Rs>, g12=<Rs,Ru>, g22=<Ru,Ru>. We define the second fundamental form II of M by
II(s,u)=h11ds2+2h12dsdu+h22du2, |
where h11=<Rss,u>, h12=<Rsu,u>, h22=<Ruu,u>. The Gaussian K(s,u) and mean H(s,u) curvature are defined by
K(s,u)=h11h22−h212g11g22−g212 and H(s,u)=g11h22+g22h11−2g12h222(g11g22−g212). |
Let γ:I⊆R→M is a unit speed curve on M. Since γ(s) is a space curve, there exists the Serret-Frenet frame {t(s), n(s), b(s)}. The derivative formulas for Serret-Frenet frame are given by
(t′n′b′)=(0κ0−κ0τ0−τ0)(tnb), (′=dds), |
where κ(s) is the curvature and τ(s) is the torsion of γ(s). Due to γ(s) on the surface M, we have the moving Darboux frame {γ(s);e1,e2,e3}; t=e1(s), e3=e3(s) is the surface unit normal restricted to γ, and e2=e3×e1 be the unit tangent to the surface M. The the relationships between these frames is expressed as:
(e1e2e3)=(1000cosφsinφ0−sinφcosφ)(tnb);φ=φ(s). | (2.1) |
Then we have the following Frenet-Serret type formulae:
(e′1e′2e′3)=(0κgκn−κg0τg−κn−τg0)(e1e2e3)=˜ω(s)×(e1e2e3), | (2.2) |
where ˜ω(s)=τge1−κne2+κge3 is referred to as the Darboux vector. Here,
{κn(s)=κsinφ=<γ′′,e3>,κg(s)=κcosφ=det(γ′,γ′′,e3),τg(s)=τ−φ′=det(γ′,e3,e′3). | (2.3) |
We call κg(s) a geodesic curvature, κn(s) a normal curvature, and τg(s) a geodesic torsion of γ(s), respectively. In terms of these quantities, the geodesics, curvatures lines, and asymptotic lines on a smooth surface may be characterized, as loci along which κg=0, τg=0, and κn=0, respectively.
Now, we introduce a new rotation minimizing frame using a common tangent vector field to a surface along a curve and call this frame the rotation minimizing Darboux frame.
Definition 3.1. A moving orthogonal frame {ξ1, ξ3, ξ3}, along a space curve r(s), is called rotation minimizing frame (RMF) with respect to a certain reference direction if its angular velocity ω has no component along that direction.
Although the Darboux frame is not RMF with respect to ei(i=1, 2, 3), one can easily derive such a RMF from it. New plane vectors (ξ1, ζ2) are specified through a rotation of (e2, e3) according to
(ζζ1ζ2)=(1000cosϑsinϑ0−sinϑcosϑ)(e1e2e3), | (3.1) |
with a certain angle ϑ(s). Here, we will call the set {ζ, ζ1, ζ2}} as rotation minimizing Darboux frame (RMDF). Therefore, we have the alternative frame equations:
(ζ′ζ′1ζ′2)=(0κ1κ2−κ100−κ200)(ζζ1ζ2)=ω×(ζζ1ζ2), | (3.2) |
where ω(s)=−κ2ζ1+κ1ζ2 is the Darboux vector. One can show that:
{κ1(s)=κgcosϑ+κnsinϑ,κ2(s)=κgsinϑ−κncosϑ,√κ21+κ22=√κ2g+κ2n=κ(s),ϑ(s)=−∫s0τgds+ϑ0,ϑ0=ϑ(0). | (3.3) |
From Eqs (2.2) and (3.2) we observe that the relative velocity is
˜ω(s)−ω(s)=τge1. |
This shows that, the Darboux frame involves an additional rotation about the tangent, whose speed equals the geodesic torsion τg(s). This examination explains the integral formula of Eq (3.3) for computing the RMDF by rectifying the unwanted rotation of the Darboux frame. Hence, the Darboux frame is conformable with the RMDF for curvature lines, that is, τg=0. This is in analogous with Klok's result[1].
Proposition 3.1. Under the above notations we state the following:
(1) If γ(s) is a geodesic curve on M, then the curvatures κ1(s), and κ2(s) satisfy the following:
tan−1(κ1κ2)=−ϑ(s),withϑ(s)=−∫s0τgds+ϑ0,ϑ0=ϑ(0). |
(2) If γ(s) is an asymptotic curve on M, then the curvatures κ1(s), and κ2(s) satisfy the following:
cot−1(κ1κ2)=ϑ(s), withϑ(s)=−∫s0τgds+ϑ0,ϑ0=ϑ(0). |
In this subsection, we give the parametric representations of sweeping surface through the spine curve γ(s) on the surface M in the following: The sweeping surface associated to γ(s), is the envelope of the family of unit spheres, with the center on the curve γ(s)∈M.
Remark 3.1. Clearly, if γ(s) is a straight line, thus the sweeping surface is just a circular cylinder, having γ(s) as symmetry axis. If, on the other hand, γ(s) is a circle, then the corresponding sweeping surface is a torus.
Now, it is easy to see that the connect between the spheres from the family and the sweeping surface is a great circle of the unit sphere, lying in the subspace Sp{ζ1(s),ζ2(s)}, of the spine curve γ(s). Let us indicate by q the position vector attaching the point from the curve γ(s) with the point from the sweeping surface. Then, clearly, we have
Υ:q=γ(s)+r, | (3.4) |
where the unit vector r itself lies in the same subspace Sp{ζ1(s),ζ2(s)}. Let us indicate by the angle u between the vectors r and ζ1. Then, as one can see immediately, we have
r(u)=cos(u)ζ1(s)+sin(u)ζ2(s), | (3.5) |
which is the characteristic circles of sweeping surface. Combining Eqs (3.4) and (3.5), we see that we obtained a representation of the sweeping surface,
Υ:q(s,u)=γ(s)+cosuζ1(s)+sinuζ2(s), | (3.6) |
This representation of Υ excludes sweeping surfaces with stationary vector ζ, because its geometrical properties that is not very important and very easy to be studied.
The q′s tangent vectors are
{qu(s,u)=−sinuζ1+cosuζ2,qs(s,u)=(1−κ1cosu−κ2sinu)ζ. | (3.7) |
Then,
u(s,u)=qs×qu‖qs×qu‖=cosuζ1+sinuζ2, | (3.8) |
which shows that surface normal is included in the subspace Sp{ζ1(s),ζ2(s)}, since it is orthogonal to ζ. The coefficients of the first fundamental form g11, g12 and g22 are
{g11=<qs,qs>=(1−κ1cosu−κ2sinu)2,g12=<qs,qu>=0, g22=<qu,qu>=1. | (3.9) |
For the second fundamental form, we have
{qss=−(κ′1cosu+κ′2sinu)ζ+(1−κ1cosu−κ2sinu)(κ1ζ1+κ2ζ2),qsu=(κ1sinu+κ2cosu)ζ,quu=−cosuζ1−sinuζ2. |
Hence, the elements of the second fundamental form h11, h12, and h22 are
{h11=<qss,u>=−(1−κ1cosu−κ2sinu)(κ1sinu+κ2cosu),h12=<qsu,u>=0,h22=<quu,u>=−1. |
Then, the u-and s curves are curvature lines, that is, g12 and h12 vanish identically (g12=h12=0). Thus, the Gaussian and mean curvature are calculated as
K(s,u)=−κ1cosu+κ2sinu1−(κ1cosu+κ2sinu), |
and
H(s,u)=1−2(κ1cosu+κ2sinu)2−2(κ1cosu+κ2sinu). |
Corollary 3.1. The sweeping surface defined by Eq (3.6) has a constant Gaussian curvature iff
κ1cosu+κ2sinu=cc−1 |
for some real constant c.
Corollary 3.2. The sweeping surface defined by Eq (3.6) has a constant mean curvature iff
κ1cosu+κ2sinu=2c−1c−1 |
for some real constant c.
Corollary 3.3. The sweeping surface defined by Eq (3.6) is a (K(s,u),H(s,u))-Weingarten surface.
Corollary 3.4. The sweeping surface defined by Eq (3.6) is a (K(s,u),H(s,u))-linear Weingarten surface iff
κ1cosu+κ2sinu=c+c12(c1+c2−c) |
where c, c1 and c2 are not all zero real numbers.
On the other hand, the isoparametric curve
π(u):β(u):=q(u,s0)=γ(s0)+cosuζ1(s0)+sinuζ2(s0), | (3.10) |
is a planar unit speed curvature line. Equation (3.10) define a set of one-parameter set of planes in E3. The unit tangent vector to β(u) is
tβ(u)=−sinuζ1(s0)+cosuζ2(s0), |
and thus the unit principal normal vector of β(u) is given by
nβ=ζ(s0)×tβ(u)=cosuζ1+sinuζ2=u(s0,u). |
Thus, the surface normal u(s0,u) is parallel to the principal normal nβ, that is, the curve β(u) is a geodesic planar curvature line, and cannot be asymptotic curve.
Surfaces whose parametric curves are curvature lines have several implementations in geometric design[1,2,3,4]. In the situation of sweeping surfaces, one has to figure the offset surfaces qf(s,u)=q(s,u)+f u(s,u) of a given surface q(s,u) at a certain distance f. In consequence of this equation, the offsetting process for sweeping surface can be reduced to the offsetting of planar profile curve, which is considerably easier to deal with. Hence, we can state the following proposition:
Proposition 3.2. Consider a sweeping surface Υ defined by Eq (3.6). Let xf(u) be the planar offset of the profile r(u) at constant distance f. Then the offset surface qf(s,u) is still a sweeping surface, generated by the spine curve γ(s) and profile curve rf(u).
Singularities and convexity are useful for grasping the ownerships of sweeping surfaces and are investigated in the following:
The sweeping surface Υ has singular points iff the first derivatives are linearly dependent, that is,
qu×qs=(1−κ1cosu−κ2sinu)u=0. | (3.11) |
Since u is a nonzero unit vector, then 1−κ1cosu−κ2sinu=0, that is,
sinu=−κ2±κ1√|κ22+κ21−1|κ22+κ21, | (3.12) |
and
cosu=−κ1±κ2√|κ22+κ21−1|κ22+κ21. | (3.13) |
Hence, there are two singular points on every generating circle. Connecting these two sets of singular points gives two curves that contain all the singular points of a sweeping surface. From Eqs (3.6) and (3.12) it follows that the expression of the two curves is
Γ(s)=γ(s)+−κ1±κ2√|κ22+κ21−1|κ22+κ21ζ1+−κ2±κ1√|κ22+κ21−1|κ22+κ21ζ2. | (3.14) |
From the above analysis it can easily be seen that:
Corollary 3.5. The sweeping surface Υ represented by Eq (3.6), has no singular points if the condition
1−κ1(−κ1±κ2√|κ22+κ21−1|κ22+κ21)−κ2(−κ2±κ1√|κ22+κ21−1|κ22+κ21)≠0, |
is satisfied.
In Computer Aided Geometric Design, conditions that guarantee the convexity of a surface are required in various applications (such as manufacturing of sculptured surfaces, or layered manufacturing). In the case of the sweeping surface Υ, however, the convexity can be controlled with the help of the differential geometric properties. Therefore, we discuss the Gaussian curvature K(s,u)=χ1χ2; χi(s,u) (i=1,2) are the principal curvatures, as follows:
Since g12=h12=0, the value of one principal curvature is
χ1(s0,u):=‖drdu×d2rdu2‖‖drdu‖−3=1. | (3.15) |
The other principal curvature is easy to get
χ2(s,u)=K(s,u)χ1(s0,u)=−κ1cosu+κ2sinu1−(κ1cosu+κ2sinu). | (3.16) |
On the other hand, the curvature for the s-curves (u-constant) can be obtained as:
χ(s,u0)=‖qs×qss‖‖qs‖3=κ1−κ1cosu−κ2sinθ. | (3.17) |
In view of Meusnier's theorem, the expression of χ2(s,u) in Eq (3.16) is
χ2(s,u)=χ(s,u0)cosψ. | (3.18) |
Here ψ=cos−1<n,u>. Thus, the the Gaussian curvature K(s,u) can be obtained as:
K(s,u)=χ(s,u)cosψ. | (3.19) |
We now concentrate on the curves on Υ that are created by parabolic points, that is, points with vanishing Gaussian curvature. These curves separate elliptic (K>0, locally convex) and hyperbolic (K<0, hence non-convex) parts of the surface. From Eq (3.18), it follows that
K(s,u)=0⇔χ(s,u)cosψ=0. |
It can be seen that there are two main cases that cause parabolic points:
Case 3.1. When χ(s,u)=0, that is κ=0. Thus, a planar point of the spine curve γ generates a parabolic curve u = const. on Υ. In other words, the spine curve γ is degenerate to a straight line. Therefore, an inflection or flat point of the spine curve gives a parabolic curve u = const.
Case 3.2. When ψ=π2(<n,u>=0), that is, the osculating plane of γ at each point coincides with the tangent plane to the surface Υ at that point. Then, the spine curve γ is not only a curvature line but also an asymptotic curve on Υ.
In fact we have the following:
Corollary 3.6. Consider a sweeping surface represented by Eq (3.6) with spine and profile curves have non-vanishing curvatures anywhere. If the spine curve γ is a curvature line and also asymptotic curve, then Υ has parabolic points.
This part exmaine in what conditions the sweeping surfaces are developable surfaces. Therefore, we analyze the case that the profile curve r(u) degenerates into a line. Then, we have the following developable surface
S:P(s,u)=γ(s)+uζ2(s),u∈R. | (3.20) |
We also have that
S⊥:P⊥(s,u)=γ(s)+uζ1(s),u∈R | (3.21) |
It is easy to show P(s,0)=α(s) (resp. P⊥(s,0)=α(s)), 0≤s≤L, that is, the surface S (resp. S⊥) interpolate the curve γ(s). Furthermore, since
Ps×Pu:=−(1−uκ2)ζ1(s), | (3.22) |
then S⊥ is the normal developable surface of S along γ(s). Hence, the curve γ(s) is a curvature line on S (resp. S⊥).
Theorem 3.1. Let Υ be the sweeping surface expressed by Eq (3.6). Then we have the following:
(1) the developable surfaces S and S⊥ intersect along γ(s) at a right angle,
(2) the curve γ(s) is a curvature line on S and S⊥.
Theorem 3.2. (Existence and uniqueness). Under the above notations there exists a unique developable surface represented by Eq (3.19).
Proof. For the existence, we have the developable represented by Eq (3.19). Furthermore, since S is a ruled surface, we may write that
{S:P(s,u)=γ(s)+uζ(s),u∈R,η(s)=η1(s)ζ1+η2(s)ζ2+η3(s)ζ,‖η(s)‖2=η21+η22+η23=1, η′(s)≠0. | (3.23) |
It can be immediately seen that S is developable iff
det(γ′,η,η′)=0⇔η1η′2−η2η′1+η3(η1κ2−η2κ1)=0. | (3.24) |
On the other hand, in view of Eq (3.21), we have
(Ps×Pu)(s,u)=−ψ(s,u)ζ1, | (3.25) |
where θ=θ(s,u) is a differentiable function. Furthermore, the normal vector Ps×Pv at the point (s,0) is
(Ps×Pu)(s,0)=−η2ζ1+η1ζ2. | (3.26) |
Thus, from Eqs (3.24), and (3.25), one finds that:
η1=0, and η2=θ(s,0), | (3.27) |
which follows from Eq (3.23) that η2η3κ1=0, which leads to η2η3=0, with κ1≠0. If (s,0) is a regular point (i.e., θ(s,0)≠0), then η2(s)≠0, and η3=0. Therefore, we obtain η(s)=ζ2. This means that uniqueness holds.
As an application (such as flank milling or cylindrical milling), during the movement of the RMDF along γ, let a cylindrical cutter be rigidly linked to this frame. Then the equation of a set of cylindrical cutters, which is defined by the movement of cylindrical cutter along γ(s), can be gotten as follows:
Sf:¯P(s,u)=P(s,u)+ρζ1(s), | (3.28) |
where ρ denotes cylindrical cutter radius. This surface is a developable surface offset of the surface P(s,u). The equation of Sf, can therefore be written as:
Sf:¯P(s,u)=γ(s)+uζ2(s)+ρζ1(s). | (3.29) |
The normal vector of cylindrical cutter can be represented as
uf(s,0)=¯PsׯPu‖¯PsׯPu‖=ζ1(s). | (3.30) |
Also, from Eq (3.27), we have
S:P(s,u)=¯P(s,u)−ρζ1(s). | (3.31) |
The derivative of Eq (3.30) with respect to s can be derived as follows
¯Ps(s,u)=Ps(s,u)−(ρω)×ζ1. | (3.32) |
Equation (3.22) shows that the vector ¯Ps(s,u) is orthogonal to the normal vector ζ1. And, the vector ζ1 is orthogonal to the tool axis vector ζ(s). Subsequently, the envelope surface of the cylindrical cutter and the developable surface P(s,u) have the common normal vector and the distance between the two surfaces is cylindrical cutter radius ρ. Hence, we can draw a conclusion as follows:
Proposition 3.3. Consider a developable surface S defined by Eq (3.19). Let Sf be the envelope surface of cylindrical cutter at distance ρ. Then the two surfaces S and Sf are offset developable surfaces.
As it is will known, there are three types of developable surfaces, the given curve can be distributed into three kinds correspondingly[12,13,14,15]. In what follows, we will discuss the relationship between the given curve γ(s)∈M and its isoparametric developable surface. The first case is when,
ζ2×ζ′2=0⇔κ2ζ1=0. | (3.33) |
In this situation, S is referred to as a cylindrical surface. Since ζ1 is a nonzero unit vector, then Υ is a cylindrical surface iff
κ2=0⇔κgcosϑ−κnsinϑ=0⇔ϑ=tan−1(κgκn),κn≠0. | (3.34) |
Similarly, we can also have ζ2×ζ′2≠0. In this situation, S is referred to as a non-cylindrical surface. Therefore, the first derivative of the directrix is
γ′(s)=c′(s)+σ(s)ζ′2(s)+σ′(s)ζ2(s), | (3.35) |
where c′ is the first derivative of the striction curve, σ(s) is a smooth function. By an immediate calculation, we can show that
det(c′,ζ2,ζ′2)=<c′,ζ×ζ′2>=0. | (3.36) |
Then the next two situations hold:
The first situation is when the first derivative of the striction curve is c′=0. Geometrically this position implies that the striction curve degenerates to a point, and S becomes a cone; the striction point of a cone is generally referred to as the vertex. By using Eqs (3.2) and (3.34) we have that S is a cone iff there exists a fixed point c and a function σ(s) such that σκ2=−1, σ′=0, which imply that
σ(s)=const.=1κ2⇔κgcosϑ−κnsinϑ=κg0cosϑ0−κn0sinϑ0. | (3.37) |
The second situation is when c′≠0, that is,
σ(s)=const.≠1κ2⇔κgcosϑ−κnsinϑ≠κg0cosϑ0−κn0sinϑ0. | (3.38) |
From Eq (3.35), we have <c′,ζ2×ζ′2>=0, that is, c′∈Sp{ζ2, ζ′2}. The condition for c to be striction curve is therefore equivalent to c′ and ζ′2 are perpendicular to each other. Therefore, we may conclude that the ruling is parallel to the first derivative of the striction curve, which is also the tangent of the striction curve. This ruled surface is referred too as a tangent ruled surface. So, the surface S is a tangent surface iff there exists a curve c(s) so that σ(s)=const.≠1κ2.
In this subsection, as an application of our main results, we give the following examples.
Example 1. Let M be a hyperboloid of one sheet defined by
M:R(s,u)=(coss−u√2coss,sins+u√2sins,u√2). |
It is easy to see that γ(s)=(coss,sins,0) lies on M. Then, we have the Darboux frame as follows:
e1(s)=(−sins,coss,0),e2(s)=(0,0,1),e3(s)=(coss,sins,0). |
The normal curvature, the geodesic curvature, and the geodesic torsion of γ(s) on M, respectively, are
κn=−1, and κg=τg=0. |
Then ϑ(s)=ϑ0 is a constant, moreover,
ζ1(s)=(sinϑ0coss,sinϑ0sins,cosϑ0),ζ2(s)=(cosϑ0coss,cosϑ0sins,−sinϑ0),κ1(s)=−sinϑ0, and κ2(s)=cosϑ0. |
1) If we take ϑ0=π2(κ2(s)=0), then we immediately obtain a sweeping surface given by
Υ:q(s,u)=((1+cosu)coss,(1+cosu)sins,−sinu). |
The graphs of the surfaces M, Υ, and M∪Υ are shown in Figure 1; 0≤u,s≤2π. Obviously, κ2(s)=0 satisfies Eq (3.33), and the developable surface
S:P(s,u)=(coss,sins,−u)+u(0,0,−1)=(coss,sins,−u) |
is a cylinder with γ(s) as a curvature line; 0≤s≤2π, and −1≤u≤1 (Figure 2).
2) In the case of ϑ0=π4(κ2(s)=const.), we obtain a sweeping surface given by
Υ:q(s,u)=((1+sinu+cosu)coss,(1+sinu+cosu)sins,(cosu−sinu)). |
The graphs of the surfaces M, Υ, and M∪Υ are shown in Figure 3; 0≤u,s≤2π. Also, the developable surface
S:P(s,u)=(coss,sins,−u)+u(0,0,−1)=(coss,sins,−u) |
is a cone with γ(s) as a curvature line; 0≤s≤2π, and −5≤u≤2 (Figure 4).
Example 2. Consider the tubular surface parameterized by
M:R(s,u)=(coss√2+√2sinu,sins√2+√2cosu,s√2), |
where 0≤u,s≤2√2π. It is clear that γ(s)=(coss√2,sins√2,s√2) lies on M. By a similar procedure as in Example 1, we have
e1(s)=1√2(−sins√2,coss√2,1√2),e2(s)=1√2(−coss√2+1√2sins√2,−sins√2−1√2coss√2,1√2),e3(s)=1√2(coss√2+1√2sins√2,sins√2−1√2coss√2,1√2),κn=κg=1√2, and τg=12. |
Then ϑ(s)=−s2+ϑ0. If we choose ϑ0=0, for example, we have
ζ1(s)=(1√2[cos(s√2−s2)+sin(s√2−s2)](cos(s2)−sin(s2))sin(s√2)−1√2(cos(s2)+sin(s2))cos(s√2)1√2(cos(s2)+sin(s2))),ζ2(s)=(1√2[cos(s√2−s2)−sin(s√2−s2)](cos(s2)+sin(s2))sin(s√2)−1√2(cos(s2)−sin(s2))cos(s√2)−1√2(cos(s2)−sin(s2))),κ1(s)=12√2(sins2+coss2), and κ2(s)=12√2(sins2−coss2). |
Therefore, the sweeping surface is
Υ:q(s,u)=(coss√2+1√2[cos(s√2−s2+u)+sin(s√2−s2+u)]sins√2+[(cos(s2−u)−sin(s2+u))sin(s√2)−1√2(cos(s2−u)+sin(s2+u))cos(s√2)]s√2−1√2(cos(s2−u)+sin(s2−u))). |
The graphs of the surfaces M, Υ, and M∪Υ are shown in Figure 5; 0≤u,s≤2√2π. In view of κ2(s)≠κ2(s0) and Eq (3.37), the developable surface
S:P(s,u)=(coss√2+u√2[cos(s√2−s2)+sin(s2−s√2)]sins√2+u[(coss2+sins2)sins√2−1√2(coss2−sins2)coss√2]s√2−u√2(coss2−sins2)) |
is a tangent surface with γ(s) as a curvature line; 0≤s≤2√2π, and −1≤u≤1 (Figure 6).
This paper introduce and study sweeping surface with a new RMDF associated with a curve on the surface. The paper further investigated the problem of requiring the sweeping surface is a developable surface. There are several opportunities for further work. An analogue of the problem addressed in this paper may be consider for surfaces in Minkowski 3-space. We will study this problem in the future.
All of the data are available within the paper. The authors have no conflicts of interest.
[1] |
C. R. Nastase, D. J. Mavriplis, A parallel hp-multigrid solver for three-dimensional discontinuous Galerkin discretizations of the Euler equations, 45th AIAA Aerospace Sciences Meeting and Exhibit, (2007), 512. https://doi.org/10.2514/6.2007-512 doi: 10.2514/6.2007-512
![]() |
[2] |
H. Luo, J. D. Baum, R. Löhner, A p-multigrid discontinuous Galerkin method for the Euler equations on unstructured grids, J. Comput. Phys., 211 (2006), 767–783. https://doi.org/10.1016/j.jcp.2005.06.019 doi: 10.1016/j.jcp.2005.06.019
![]() |
[3] |
R. Hartmann, P. Houston, Adaptive discontinuous Galerkin finite element methods for the compressible Euler equations, J. Comput. Phys., 183 (2002), 508–532. https://doi.org/10.1063/1.5033621 doi: 10.1063/1.5033621
![]() |
[4] |
F. Bassi, S. Rebay., High-order accurate discontinuous finite element solution of the 2D Euler equations, J. Comput. Phys., 138 (1997), 251–285. https://doi.org/10.1006/jcph.1997.5454 doi: 10.1006/jcph.1997.5454
![]() |
[5] |
F. Bassi, S. Rebay., A high-order accurate discontinuous finite element method for the numerical solution of the compressible Navier-Stokes equations, J. Comput. Phys., 131 (1997), 267–279. https://doi.org/10.1006/jcph.1996.5572 doi: 10.1006/jcph.1996.5572
![]() |
[6] |
L. Chen, M. B. Ozakin, R. Zhao, H. Bagci, A locally-implicit discontinuous Galerkin time-domain method to simulate metasurfaces using generalized sheet transition conditions, IEEE Trans. Antennas Propag., 71 (2023), 869–881. https://doi.org/10.1109/AP-S/USNC-URSI47032.2022.9887215 doi: 10.1109/AP-S/USNC-URSI47032.2022.9887215
![]() |
[7] |
G. S. Baruzzi, Wagdi Habashi, A second order finite element method for the solution of the transonic Euler and Navier-Stokes equations, Int. J. Numer. Methods Fluids., 20 (1995), 671–693. https://doi.org/10.1002/fld.1650200802 doi: 10.1002/fld.1650200802
![]() |
[8] |
M. Gurris, D. Kuzmin, S. Turek., Implicit finite element schemes for the stationary compressible Euler equations, Int. J. Numer. Methods Fluids., 69 (2012), 1–28. https://doi.org/10.1002/fld.2532 doi: 10.1002/fld.2532
![]() |
[9] |
A. Jameson, D. Mavriplis, Finite volume solution of the two-dimensional Euler equations on a regular triangular mesh, AIAA J, 24 (1986), 611–618. https://doi.org/10.2514/3.9315 doi: 10.2514/3.9315
![]() |
[10] |
W. K. Anderson, Comparison of finite volume flux vector splittings for the Euler equations, AIAA J., 24 (2015), 1453–1460. https://doi.org/10.2514/3.9465 doi: 10.2514/3.9465
![]() |
[11] |
L. Acedo, S. B. Yuste, An explicit finite difference method and a new von Neumann-type stability analysis for fractional diffusion equations, SIAM J. Numer. Anal., 42 (2005), 1862–1874. https://doi.org/10.1137/030602666 doi: 10.1137/030602666
![]() |
[12] |
R. F. Warming, B. J. Hyett, The modified equation approach to the stability and accuracy of finite difference method, J. Comput. Phys., 14 (1974), 159–179. https://doi.org/10.1016/0021-9991(74)90011-4 doi: 10.1016/0021-9991(74)90011-4
![]() |
[13] |
U. Baur, P. Benner, L. Feng, Model order reduction for linear and nonlinear systems: A system-theoretic perspective, Arch. Comput. Methods Eng., 21 (2014), 331–358. https://doi.org/10.1007/s11831-014-9111-2 doi: 10.1007/s11831-014-9111-2
![]() |
[14] |
Z. Bai, Krylov subspace techniques for reduced-order modeling of large-scale dynamical systems, Appl. Numer. Math., 43 (2002), 9–44. https://doi.org/10.1016/S0168-9274(02)00116-2 doi: 10.1016/S0168-9274(02)00116-2
![]() |
[15] |
S. Renee, M. Laura, P. Benjamin, W. Karen, Projection-based model reduction: Formulations for physics-based machine learning, Comput Fluids., 179 (2019), 704–717. https://doi.org/10.1016/j.compfluid.2018.07.021 doi: 10.1016/j.compfluid.2018.07.021
![]() |
[16] |
J. Jiang, Y. Chen, N. Akil, Offline-enhanced reduced basis method through adaptive construction of the surrogate training set, J Sci Comput., 73 (2017), 853–875. https://doi.org/10.1007/s10915-017-0551-3 doi: 10.1007/s10915-017-0551-3
![]() |
[17] |
Z. Peng, Y. Chen, Y. Cheng, F. Li, A reduced basis method for radiative transfer equation, J Sci Comput., 91 (2022), 5. https://doi.org/10.1007/s10915-022-01782-2 doi: 10.1007/s10915-022-01782-2
![]() |
[18] |
D. Binion., X. Chen, A Krylov enhanced proper orthogonal decomposition method for efficient nonlinear model reduction, Finite Elem. Anal. Des., 47 (2011), 728–738. https://doi.org/10.1016/j.finel.2011.02.004 doi: 10.1016/j.finel.2011.02.004
![]() |
[19] |
L. Sirovich., Turbulence and the dynamics of coherent structures part Ⅰ: Coherent structures, Appl. Math., 45 (1986), 561–571. https://doi.org/10.1090/qam/910464 doi: 10.1090/qam/910464
![]() |
[20] |
C. L. Pettit, P. S. Beran., Application of proper orthogonal decomposition to the discrete Euler equations, Int. J. Numer. Methods Eng., 55 (2002), 479–497. https://doi.org/10.1002/nme.510 doi: 10.1002/nme.510
![]() |
[21] |
J. Goss, K. Subbarao., Inlet shape optimization based on POD model reduction of the Euler equations, 12th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, (2008), 5809. https://doi.org/10.2514/6.2008-5809 doi: 10.2514/6.2008-5809
![]() |
[22] |
G. Kerschen, J. C. Golinval, A. F. Vakakis, L. A. Bergman, The method of proper orthogonal decomposition for dynamical characterization and order reduction of mechanical systems: an overview, Nonlinear Dyn., 41 (2005), 147–169. https://doi.org/10.1007/s11071-005-2803-2 doi: 10.1007/s11071-005-2803-2
![]() |
[23] |
Q. Wang, N. Ripamonti, J. S. Hesthaven, Recurrent neural network closure of parametric POD-Galerkin reduced-order models based on the Mori-Zwanzig formalism, J Comput Phys., 410 (2020), 109402. https://doi.org/10.1016/j.jcp.2020.109402 doi: 10.1016/j.jcp.2020.109402
![]() |
[24] |
T. Akman, Error estimates for space–time discontinuous Galerkin formulation based on proper orthogonal decomposition, Appl. Anal., 96 (2017), 461–482. https://doi.org/10.1080/00036811.2016.1143930 doi: 10.1080/00036811.2016.1143930
![]() |
[25] |
C. Gräßle, M. Hinze, POD reduced order modeling for evolution equations utilizing arbitrary finite element discretizations, Adv. Comput. Math., 44 (2018), 1941–1978. https://doi.org/10.1007/s10444-018-9620-x doi: 10.1007/s10444-018-9620-x
![]() |
[26] |
Z. D. Luo, F. Teng, J. Chen, A POD-based reduced-order Crank-Nicolson finite volume element extrapolating algorithm for 2D Sobolev equations, Math. Comput. Simul., 146 (2018) 118–133. https://doi.org/10.1016/j.matcom.2017.11.002 doi: 10.1016/j.matcom.2017.11.002
![]() |
[27] |
S. F Zhu, L. Dedè, A. Quarteroni, Isogeometric analysis and proper orthogonal decomposition for parabolic problems, Numer Math., 135 (2017), 333–370. https://doi.org/10.1007/s00211-016-0802-5 doi: 10.1007/s00211-016-0802-5
![]() |
[28] |
R. C. Li, Q. B. Wu, S. F. Zhu, Proper orthogonal decomposition with SUPG-stabilized isogeometric analysis for reduced order modelling of unsteady convection-dominated convection-diffusion-reaction problems, J Comput Phys., 387 (2019), 280–302. https://doi.org/10.1016/j.jcp.2019.02.051 doi: 10.1016/j.jcp.2019.02.051
![]() |
[29] |
S. F Zhu, L. Dedè, A. Quarteroni, Isogeometric analysis and proper orthogonal decomposition for the acoustic wave equation, ESAIM: M2AN, 51 (2017), 1197–1221. https://doi.org/10.1051/m2an/2016056 doi: 10.1051/m2an/2016056
![]() |
[30] |
S. Jun, K. H Park, H. M Kang, D. H Lee, M. Cho, Reduced order model of three-dimensional Euler equations using proper orthogonal decomposition basis, J Mech Sci Technol., 24 (2010), 601–608. https://doi.org/10.1007/s12206-010-0106-0 doi: 10.1007/s12206-010-0106-0
![]() |
[31] |
R. Hartmann, Error estimation and adjoint based refinement for an adjoint consistent DG discretisation of the compressible Euler equations, Int J Comput Sci Mat, 1 (2007), 207–220. https://doi.org/10.1504/IJCSM.2007.016532 doi: 10.1504/IJCSM.2007.016532
![]() |
[32] |
X. Zhang, C. Shu, Positivity-preserving high order discontinuous Galerkin schemes for compressible Euler equations with source terms, J Comput Phys., 230 (2011), 1238–1248. https://doi.org/10.1016/j.jcp.2010.10.036 doi: 10.1016/j.jcp.2010.10.036
![]() |
[33] |
M Boizard, R Boyer, G Favier, P. Larzabal, Fast multilinear Singular Values Decomposition for higher-order Hankel tensors, 2014 IEEE 8th Sensor Array and Multichannel Signal Processing Workshop (SAM), (2014), 437–440. https://doi.org/10.1109/SAM.2014.6882436 doi: 10.1109/SAM.2014.6882436
![]() |
[34] |
P. Batten, N. Clarke, C. Lambert, D. M. Causon, On the choice of wavespeeds for the HLLC Riemann solver, SIAM J Sci Comput, 18 (1997), 1553–1570. https://doi.org/10.1137/S1064827593260140 doi: 10.1137/S1064827593260140
![]() |
[35] |
S. Simon, J. C. Mandal, A cure for numerical shock instability in HLLC Riemann solver using antidiffusion control, Comput Fluids., 174 (2018), 144–166. https://doi.org/10.1016/j.compfluid.2018.07.001 doi: 10.1016/j.compfluid.2018.07.001
![]() |
[36] |
S. Simon, J. C. Mandal, A simple cure for numerical shock instability in the HLLC Riemann solver, J Comput Phys., 378 (2019), 477–496. https://doi.org/10.1016/j.jcp.2018.11.022 doi: 10.1016/j.jcp.2018.11.022
![]() |
[37] | Y. Zhu., A. C. Cangellaris, Multigrid Finite Element Methods for Electromagnetic Field Modeling, New York: John Wiley & Sons, 2006. |
[38] |
B. Cockburn, S. Hou, C. W Shu, TVD Runge-Kutta local projection discontinuous Galerkin finite element method for conservation laws Ⅳ: the multidimensional case, Math Comp, 55 (1990), 545–581. https://doi.org/10.1090/S0025-5718-1990-1010597-0 doi: 10.1090/S0025-5718-1990-1010597-0
![]() |
[39] |
H. Zhang, J. Y Yan, X. Qian, X. M Gu, S. H Song, On the preserving of the maximum principle and energy stability of high-order implicit-explicit Runge-Kutta schemes for the space-fractional Allen-Cahn equation, Numer. Algorithms, 88 (2021), 1309–1336. https://doi.org/10.1007/s11075-021-01077-x doi: 10.1007/s11075-021-01077-x
![]() |
[40] |
H. Luo, J. Baum, R. Löhner, A fast p-Multigrid Discontinuous Galerkin Method for Compressible Flows at All Speeds, AIAA J, 46 (2008), 635–652. https://doi.org/10.2514/1.28314 doi: 10.2514/1.28314
![]() |
[41] |
J. Burkardt, M. Gunzburger, H. C. Lee, POD and CVT-based reduced-order modeling of Navier-Stokes flows, Comput. Methods Appl. Mech. Eng., 196 (2006), 337–355. https://doi.org/10.1016/j.cma.2006.04.004 doi: 10.1016/j.cma.2006.04.004
![]() |
[42] |
Z. Luo, J. Gao, A POD reduced-order finite difference time-domain extrapolating scheme for the 2D Maxwell equations in a lossy medium, J. Math. Anal. Appl., 444 (2016), 433–451. https://doi.org/10.1016/j.jmaa.2016.06.036 doi: 10.1016/j.jmaa.2016.06.036
![]() |
[43] |
K. Kunisch, S. Volkwein, Galerkin proper orthogonal decomposition methods for parabolic problems, Numer. Math., 90 (2001), 117–148. https://doi.org/10.1007/s002110100282 doi: 10.1007/s002110100282
![]() |
[44] | V. Schmitt, Pressure distributions on the ONERA M6-wing at transonic Mach numbers, experimental data base for computer program assessment, AGARD AR138, (1979). |
1. | Emad Solouma, Ibrahim Al-Dayel, Meraj Ali Khan, Youssef A. A. Lazer, Characterization of imbricate-ruled surfaces via rotation-minimizing Darboux frame in Minkowski 3-space E31, 2024, 9, 2473-6988, 13028, 10.3934/math.2024635 | |
2. | Emad Solouma, Ibrahim Al-Dayel, Meraj Ali Khan, Mohamed Abdelkawy, Investigation of Special Type-Π Smarandache Ruled Surfaces Due to Rotation Minimizing Darboux Frame in E3, 2023, 15, 2073-8994, 2207, 10.3390/sym15122207 | |
3. | Gökhan Köseoğlu, Mustafa Bilici, Involutive sweeping surfaces with Frenet frame in Euclidean 3-space, 2023, 9, 24058440, e18822, 10.1016/j.heliyon.2023.e18822 | |
4. | Özgür Boyacıoğlu Kalkan, Süleyman Şenyurt, Mustafa Bilici, Davut Canlı, Sweeping surfaces generated by involutes of a spacelike curve with a timelike binormal in Minkowski 3-space, 2025, 10, 2473-6988, 988, 10.3934/math.2025047 | |
5. | Yangke Deng, Yanlin Li, Süleyman Şenyurt, Davut Canlı, İremnur Gürler, On the Special Viviani’s Curve and Its Corresponding Smarandache Curves, 2025, 13, 2227-7390, 1526, 10.3390/math13091526 |