Forum Replies Created
-
Posts
-
I think SOFA is a good tool and also a great community to study FEM.
And I want to learn deeply, so I often ask questions about papers just for real understanding.
NO other special intend AT ALL. I am sorry if my questions have made you uncomfortable.There is only one place related to firction in the paper(Duriez et al,2006). I can’t see where it implements the coulomb’s law. There is only signorini’s law. Am I miss something important?
Bellow is respect to the algorithm 1 in 2006 version:
I am pretty sure there are some errors in 2 ways:
1.Through experiments: It’s easy to implement algorithm 1 strictly follow the process. But I will always get “NaN” in the inverse matrix of “Knn”. Try that with little time and one probably get the same result.
2.Theoretically: I feel difficult to understand the paper in the way it provides, so i can not determine where the logic is wrong exactly. But instead, I rewrite the problem all by pure math, and get a diffierent version of the inverse matrix of “knn”. And it turns out works very well.
I think the best way to be sure about this(if you want) is doing the experiment of Algorithm 1. Maybe just some typing errors, but it indeed affects the result a lot.
It turns out to be a normal question in field of robot control or computer graphics, i see now. Thanks all the same.
i have solved the problem, not by looking at the source code though. it turns out a normal question in area which need to compute the orientation of a object, like computer graphic or robot control etc,and so many others.
and by the way,how to @ someone in a topic?
link below is the cpomputional process, i still can not see in which way could get the formula (8)?
you mean the implemention is fine which strictly follow the algorithm of the paper?
what kind of fine?It’s also strange that there is a negative sign in the second fomula of formula(7) . It seems should not be.
Hi Hugo,
Here is the link of 2005 :https://link.springer.com/chapter/10.1007/11566489_66.
You can download from the page.Imagine a series of linked beams. As far as i konw, from the “KU=P” equitions, we can get the displacement vector “U”, but the radian in “U” is not the same as euler angles, so one can not just use euler angle to change a node’s rotation. so how to roate a node?
Hi Hugo，
There seems to be some mistakes in the formulars both in the 2005 and 2006 version. 2005 version for example:
1.”[Hi”] should interchange place with”[Hi]T” int the formula (8) .
2.In “Algorithm 2”, to compute the “u_node”,a negative sign should be added to the right.
3.The second equation in formula (7) lost the subscript “ii”;My question is am i right about the 3 errors?
thanks for your quick reply. i have already solved the problem. it just needs check the box of “Advance” in cmake gui. by the way,i have send an email for the license of a plugin, servral days past and i got no reply, is something wrong?
