MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

In perfect elastostatics, the unknown is the displacement $x\mapsto y$, where $x\in\Omega\subset{\mathbb R}^3$ is the reference configuration, and $y\in{\mathbb R}^3$. It obeys to an 2nd-order PDEs. When we rewrite the PDE as a 1st-order system, the unknown becomes the deformation gradient $F(x):=\nabla y$.

A fundamental constraint is that the matter cannot interpenetrate. In particular, one must have $\det F>0$ everywhere. This is highly non-trivial, in the sense that the set $P$ of $3\times3$ real matrices with positive determinant has a non-trivial homotopy type. Thanks to the polar factorization, $P$ is homeomorphic to the product of ${\bf SPD}_3$ (a convex cone, something trivial) with ${\bf SO}_3$ whose fundamental group is ${\mathbb Z}_2$.

This raises the following questions. Given a domain $\Omega\subset{\mathbb R}^3$, how many connected components are there in $C(\partial\Omega;P)$ ? How can we determine if an $f\in C(\partial\Omega;P)$ can be lifted as an $F\in C(\Omega;P)$ ? How can we determine if an $f\in C(\partial\Omega;P)$ can be lifted as an $F\in C(\Omega;P)$ such that $F$ is homotopic to the identity ?

share|cite|improve this question
What is $C(\partial \Omega; P)?$ Is it continuous functions with values in $P?$ – Igor Rivin Sep 25 '11 at 15:17
In elastostatics I think you usually assume that the deformation is gradually applied, which means that $F$ is assumed homotopic to the identity. But, it you don't assume this, suppose that the body $\Omega$ is homeomorphic to a solid ball, then $C(\partial\Omega,P)$ is classified by $\pi_2(SO(3))$, which I am pretty sure(?) is 0, since $S^3$ is the double cover. In other cases of $\Omega$ I don't know. If this comment is all wrong, which is possible, somebody please remove it. – Bob Terrell Sep 25 '11 at 15:44
Yes, $\pi_2(SO(3))=0,$ and you are correct. – Igor Rivin Sep 25 '11 at 18:45

Complementing Dmitri's answer: one can see $SO(3)\cong \mathbb{P^3}(\mathbb{R})$ as the 3-skeleton of $K(\mathbb{Z}/2,1)\cong \mathbb{P^\infty}(\mathbb{R})$. Assuming $\Omega$ and its boundary are non-pathological, they will have the homotopy type of 2-dimensional $CW$-complexes. If $X$ is a 2-dimensional $CW$-complex, one can deform any mapping $X\to K(\mathbb{Z}/2,1)$ into a mapping to the 2-skeleton and any homotopy between such mappings into a mapping to the 3-skeleton. So $H^1(X,\mathbb{Z}/2)\cong [X,K(\mathbb{Z}/2,1)]\cong [X,SO(3)]$.

share|cite|improve this answer
Algori, thanks a lot for the answer! – Dmitri Sep 25 '11 at 16:24
Dmitri -- welcome! – algori Sep 25 '11 at 16:31

The answer to this question can be formulated in the following way. All maps from $\Omega$ to $P$ are classified up to homotopy by the elements $\alpha\in H^1(\Omega,\mathbb Z_2)$, and maps from $\Omega$ to $P$ are classified by the elements $\beta\in H^1(\partial\Omega,\mathbb Z_2)$. Since we have the inclusion $i:\partial \Omega\to \Omega$ we get the map $i^*: H^1(\Omega,\mathbb Z_2)\to H^1(\partial\Omega,\mathbb Z_2)$. In order to be able to extend any map $\partial \Omega\to P$ to the map $\Omega\to P$ we just need to check that $i^*\alpha=\beta$.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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