## geopdes-users — geopdes users mailng list

You can subscribe to this list here.

 2010 2011 2012 2013 2014 2015 Jan Feb Mar Apr May Jun Jul Aug Sep Oct (1) Nov (1) Dec Jan Feb Mar (1) Apr May (1) Jun (9) Jul (2) Aug (6) Sep (11) Oct (15) Nov (4) Dec (9) Jan (7) Feb (14) Mar (11) Apr May Jun (3) Jul Aug Sep Oct Nov (3) Dec Jan (8) Feb Mar (1) Apr (3) May Jun Jul (3) Aug (2) Sep Oct Nov Dec (4) Jan Feb Mar Apr May (2) Jun (2) Jul Aug Sep (5) Oct (3) Nov (1) Dec (4) Jan (5) Feb (10) Mar (12) Apr (14) May (26) Jun (7) Jul Aug Sep (4) Oct Nov Dec
S M T W T F S

1

2

3

4

5

6

7

8

9

10

11

12

13

14
(3)
15

16
(2)
17

18

19

20

21
(2)
22
(1)
23

24

25

26

27

28

29

30

31

Showing 8 results of 8

 Re: [Geopdes-users] Which control points coordinates and linear functions From: Rafael Vazquez - 2013-01-22 09:47:37 ```Dear Marco, good, I'm glad to hear that it works. For your information, the function "op_u_v" computes the mass matrix also for elasticity. In fact, it works for 2D/3D, and for scalar or vectorial problems. Regards, Rafa Il 21/01/2013 22:10, Marco Brino ha scritto: > Dear Rafa, > > no problem, thank you for the response! > That was one of my problems, that having considered the weights in the > computation of the NURBS functions and derivatives, once I compute the > Jacobian matrix I need Cartesian coordinates. > Other problem I found myself, is the computation of the NURBS > derivatives itself: using the formula with the "squared sum" in the > denominator, this caused me numerical errors. Now I substituted it > with the one with the sum without the square elevation, and everything > works perfect. > > Just for information, I found that NASTRAN 8-nodes brick uses > under-integration to manage shear locking, with also correction of > Poisson's ratio on the off-diagonal terms in the constitutive matrix. > > Now that my code works and after some tests I got interesting results, > efficiency, etc… maybe if you find it interesting we can try > implementing this modal analysis in GeoPDEs. If I'm right, I didn't > find the mass matrix computation in the elasticity toolbox! > > Thank you very much for the help! > > Kindest regards, > Marco > > ```
 Re: [Geopdes-users] Which control points coordinates and linear functions From: Marco Brino - 2013-01-21 21:10:56 ```Dear Rafa, no problem, thank you for the response! That was one of my problems, that having considered the weights in the computation of the NURBS functions and derivatives, once I compute the Jacobian matrix I need Cartesian coordinates. Other problem I found myself, is the computation of the NURBS derivatives itself: using the formula with the "squared sum" in the denominator, this caused me numerical errors. Now I substituted it with the one with the sum without the square elevation, and everything works perfect. Just for information, I found that NASTRAN 8-nodes brick uses under-integration to manage shear locking, with also correction of Poisson's ratio on the off-diagonal terms in the constitutive matrix. Now that my code works and after some tests I got interesting results, efficiency, etc… maybe if you find it interesting we can try implementing this modal analysis in GeoPDEs. If I'm right, I didn't find the mass matrix computation in the elasticity toolbox! Thank you very much for the help! Kindest regards, Marco On Jan 21, 2013, at 2:55 PM, Rafael Vazquez wrote: > Dear Marco, > sorry for the late response. For the weights in the control points, you > can use any of the two approaches: either use the Cartesian coordinates > and the weights separately, or use the homogeneous (weighted) > coordinates. The important thing is that, when you write your function > to compute/refine the geometry, you use that information in the correct > way. The second option is the one that is implemented in the NURBS > toolbox, following the algorithms in the NURBS Book by Piegl and Tiller. > In the book you can also find more explanations about the use of > homogeneous coordinates. I hope this is helpful. > > I have almost no experience with NASTRAN, so I don't know what could be > the difference. Maybe someone else in the list could help with this one. > > Regards, > Rafa > > > Il 17/01/2013 00:18, Marco Brino ha scritto: >> Hi all, >> >> I'm always trying to develop my own 3D IGA code, and I have a couple of >> questions: >> - when dealing with NURBS, which control point coordinates must be used, >> when computing the mapping between the physical domain and the >> parametric domain? I mean, the "original" coordinates, or the >> coordinates multiplied by the weights, like we have to do with the >> "nrbmak" function? >> >> - I understand that using linear B-spline functions causes the shape >> functions to be equal to the ones used by standard FEM. So also the >> stiffness matrix is equal to the one computed with standard FEM. So I >> tried to compare the stiffness matrix computed with GeoPDEs, with the >> one computed by NASTRAN, of a beam composed by 5 hexa8 elements. Well, I >> found that the matrices are sligthly different, so that GeoPDEs gives a >> stiffer matrix. >> I tried using my 3D IGA code and I obtained the same matrix of GeoPDEs. >> >> Did someone of you find similar differencies? >> >> Thank you very much in advance! >> >> Kindest regards, >> Marco >> >> ------------------------------------------------------------------------------ >> Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS, >> MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current >> with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft >> MVPs and experts. ON SALE this month only -- learn more at: >> http://p.sf.net/sfu/learnmore_122712 >> _______________________________________________ >> Geopdes-users mailing list >> Geopdes-users@... >> https://lists.sourceforge.net/lists/listinfo/geopdes-users >> > > > ------------------------------------------------------------------------------ > Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS, > MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current > with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft > MVPs and experts. SALE \$99.99 this month only -- learn more at: > http://p.sf.net/sfu/learnmore_122412 > _______________________________________________ > Geopdes-users mailing list > Geopdes-users@... > https://lists.sourceforge.net/lists/listinfo/geopdes-users -- Marco Brino Politecnico di Torino - DIGEP Corso Duca degli Abruzzi, 24 I-10129 Torino, Italy +39 011 090 7205 +39 393 554 4537 marco.brino@... ```
 Re: [Geopdes-users] Which control points coordinates and linear functions From: Rafael Vazquez - 2013-01-21 13:55:42 ```Dear Marco, sorry for the late response. For the weights in the control points, you can use any of the two approaches: either use the Cartesian coordinates and the weights separately, or use the homogeneous (weighted) coordinates. The important thing is that, when you write your function to compute/refine the geometry, you use that information in the correct way. The second option is the one that is implemented in the NURBS toolbox, following the algorithms in the NURBS Book by Piegl and Tiller. In the book you can also find more explanations about the use of homogeneous coordinates. I hope this is helpful. I have almost no experience with NASTRAN, so I don't know what could be the difference. Maybe someone else in the list could help with this one. Regards, Rafa Il 17/01/2013 00:18, Marco Brino ha scritto: > Hi all, > > I'm always trying to develop my own 3D IGA code, and I have a couple of > questions: > - when dealing with NURBS, which control point coordinates must be used, > when computing the mapping between the physical domain and the > parametric domain? I mean, the "original" coordinates, or the > coordinates multiplied by the weights, like we have to do with the > "nrbmak" function? > > - I understand that using linear B-spline functions causes the shape > functions to be equal to the ones used by standard FEM. So also the > stiffness matrix is equal to the one computed with standard FEM. So I > tried to compare the stiffness matrix computed with GeoPDEs, with the > one computed by NASTRAN, of a beam composed by 5 hexa8 elements. Well, I > found that the matrices are sligthly different, so that GeoPDEs gives a > stiffer matrix. > I tried using my 3D IGA code and I obtained the same matrix of GeoPDEs. > > Did someone of you find similar differencies? > > Thank you very much in advance! > > Kindest regards, > Marco > > ------------------------------------------------------------------------------ > Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS, > MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current > with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft > MVPs and experts. ON SALE this month only -- learn more at: > http://p.sf.net/sfu/learnmore_122712 > _______________________________________________ > Geopdes-users mailing list > Geopdes-users@... > https://lists.sourceforge.net/lists/listinfo/geopdes-users > ```
 [Geopdes-users] Which control points coordinates and linear functions From: Marco Brino - 2013-01-16 23:20:34 ```Hi all, I'm always trying to develop my own 3D IGA code, and I have a couple of questions: - when dealing with NURBS, which control point coordinates must be used, when computing the mapping between the physical domain and the parametric domain? I mean, the "original" coordinates, or the coordinates multiplied by the weights, like we have to do with the "nrbmak" function? - I understand that using linear B-spline functions causes the shape functions to be equal to the ones used by standard FEM. So also the stiffness matrix is equal to the one computed with standard FEM. So I tried to compare the stiffness matrix computed with GeoPDEs, with the one computed by NASTRAN, of a beam composed by 5 hexa8 elements. Well, I found that the matrices are sligthly different, so that GeoPDEs gives a stiffer matrix. I tried using my 3D IGA code and I obtained the same matrix of GeoPDEs. Did someone of you find similar differencies? Thank you very much in advance! Kindest regards, Marco -- Marco Brino Politecnico di Torino - DIGEP Corso Duca degli Abruzzi, 24 I-10129 Torino, Italy +39 011 090 7205 +39 393 554 4537 marco.brino@... ```
 [Geopdes-users] Which control points coordinates and linear functions From: Marco Brino - 2013-01-16 23:19:15 ```Hi all, I'm always trying to develop my own 3D IGA code, and I have a couple of questions: - when dealing with NURBS, which control point coordinates must be used, when computing the mapping between the physical domain and the parametric domain? I mean, the "original" coordinates, or the coordinates multiplied by the weights, like we have to do with the "nrbmak" function? - I understand that using linear B-spline functions causes the shape functions to be equal to the ones used by standard FEM. So also the stiffness matrix is equal to the one computed with standard FEM. So I tried to compare the stiffness matrix computed with GeoPDEs, with the one computed by NASTRAN, of a beam composed by 5 hexa8 elements. Well, I found that the matrices are sligthly different, so that GeoPDEs gives a stiffer matrix. I tried using my 3D IGA code and I obtained the same matrix of GeoPDEs. Did someone of you find similar differencies? Thank you very much in advance! Kindest regards, Marco ```
 Re: [Geopdes-users] Order of DoF in the 3D elesticity From: Marco Brino - 2013-01-14 11:30:20 ```Thank you Rafa, so you list the DoF by components, as I guessed. I know about the boundaries, and actually I have the problem of the incoherence of results AFTER I apply the symmetry. To do it, I do the same that in FEM is done with RBE2. Anyway, I was comparing both the versions with and without the glued boundaries. Thank you very much, I will try to fit with your listing order. Kindest regards, Marco On 14/01/2013 12:06, Rafael Vázquez wrote: > Dear Marco, > in GeoPDEs_elasticity the dofs for a vector component are listed > consecutively: assume that you have N dofs for the scalar problem, you > will have 3*N for the elasticity problem. The dofs for the first component > are 1:N, for the second they are N+1:2*N, and for the third 2*N+1:3*N. I'm > not sure to have understood correctly your ordering, but if your code is > implemented in a coherent way, this should not give any error. > > By the way, are you modelling the whole cylinder without symmetry? Because > a possible source of error is what you do at the boundaries. Notice that > when you apply the parametrization, two of the boundaries of the square > collapse into a single internal edge. You will have to set that the dofs > of both sides are the same, similar to what is done in multipatch. This > topic was already discussed in the mailing list. > > Regards, > Rafa > >> Hi all, >> >> I'm trying to build Matlab IGA own codes, just for having full knowledge >> and control of the relative routines. Obviously I use GeoPDEs as a >> reference for benchmark and results. >> >> In particular, I have a 3D hollow cylinder, and I'm trying to compute >> its modal analysis. I obtained strange results so i tried to compare my >> stiffness matrix with the one calculated by GeoPDEs. >> >> The layout of the matrices appears very different, and this could be >> related to the order of the DoF are listed. >> >> In my code, they are: >> - u_1x >> - u_1y >> - u_1z >> - u_2x >> - u_2y >> - u_2z >> - ... >> >> Which is the order of the DoF in the e lasticity package of GeoPDEs? >> >> Thank you very much for the help! >> >> Kindest regards, >> Marco >> >> -- >> Marco Brino >> Politecnico di Torino - DIGEP >> Corso Duca degli Abruzzi, 24 >> I-10129 Torino, Italy >> +39 011 090 7205 >> +39 393 554 4537 >> marco.brino@... >> >> >> ------------------------------------------------------------------------------ >> Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS, >> MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current >> with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft >> MVPs and experts. SALE \$99.99 this month only -- learn more at: >> http://p.sf.net/sfu/learnmore_122412 >> _______________________________________________ >> Geopdes-users mailing list >> Geopdes-users@... >> https://lists.sourceforge.net/lists/listinfo/geopdes-users >> > -- Marco Brino Politecnico di Torino - DIGEP Corso Duca degli Abruzzi, 24 I-10129 Torino, Italy +39 011 090 7205 +39 393 554 4537 marco.brino@... ```
 Re: [Geopdes-users] Order of DoF in the 3D elesticity From: Rafael Vázquez - 2013-01-14 11:27:58 ```Dear Marco, in GeoPDEs_elasticity the dofs for a vector component are listed consecutively: assume that you have N dofs for the scalar problem, you will have 3*N for the elasticity problem. The dofs for the first component are 1:N, for the second they are N+1:2*N, and for the third 2*N+1:3*N. I'm not sure to have understood correctly your ordering, but if your code is implemented in a coherent way, this should not give any error. By the way, are you modelling the whole cylinder without symmetry? Because a possible source of error is what you do at the boundaries. Notice that when you apply the parametrization, two of the boundaries of the square collapse into a single internal edge. You will have to set that the dofs of both sides are the same, similar to what is done in multipatch. This topic was already discussed in the mailing list. Regards, Rafa > Hi all, > > I'm trying to build Matlab IGA own codes, just for having full knowledge > and control of the relative routines. Obviously I use GeoPDEs as a > reference for benchmark and results. > > In particular, I have a 3D hollow cylinder, and I'm trying to compute > its modal analysis. I obtained strange results so i tried to compare my > stiffness matrix with the one calculated by GeoPDEs. > > The layout of the matrices appears very different, and this could be > related to the order of the DoF are listed. > > In my code, they are: > - u_1x > - u_1y > - u_1z > - u_2x > - u_2y > - u_2z > - ... > > Which is the order of the DoF in the e lasticity package of GeoPDEs? > > Thank you very much for the help! > > Kindest regards, > Marco > > -- > Marco Brino > Politecnico di Torino - DIGEP > Corso Duca degli Abruzzi, 24 > I-10129 Torino, Italy > +39 011 090 7205 > +39 393 554 4537 > marco.brino@... > > > ------------------------------------------------------------------------------ > Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS, > MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current > with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft > MVPs and experts. SALE \$99.99 this month only -- learn more at: > http://p.sf.net/sfu/learnmore_122412 > _______________________________________________ > Geopdes-users mailing list > Geopdes-users@... > https://lists.sourceforge.net/lists/listinfo/geopdes-users > ```
 [Geopdes-users] Order of DoF in the 3D elesticity From: Marco Brino - 2013-01-14 09:41:18 ```Hi all, I'm trying to build Matlab IGA own codes, just for having full knowledge and control of the relative routines. Obviously I use GeoPDEs as a reference for benchmark and results. In particular, I have a 3D hollow cylinder, and I'm trying to compute its modal analysis. I obtained strange results so i tried to compare my stiffness matrix with the one calculated by GeoPDEs. The layout of the matrices appears very different, and this could be related to the order of the DoF are listed. In my code, they are: - u_1x - u_1y - u_1z - u_2x - u_2y - u_2z - ... Which is the order of the DoF in the e lasticity package of GeoPDEs? Thank you very much for the help! Kindest regards, Marco -- Marco Brino Politecnico di Torino - DIGEP Corso Duca degli Abruzzi, 24 I-10129 Torino, Italy +39 011 090 7205 +39 393 554 4537 marco.brino@... ```

Showing 8 results of 8