Howdy,

Trying to follow along "iterative dynamics with temporal coherence, catto 2005" and I've some parts that need clarification:

1) Algorithm 1 on page 7 has: J_sp(i, 1)V(b_1) and Algorithm 4 on page 17 has: J_sp(i, 1) * B(1, i)

Presumably both are dot products between 6-vectors or does the dot, represented by * above, on page 17 signify some different operation?

2) Looking at equation (35) on page 14:

\eta = (\zeta / \delta t) - J((V1 / \delta t) + M^-1 * F_ext)

It would seem eta is an s-by-1 matrix? F_ext is 6n-by-1, M^-1 is 6n-by-6n, V1 is 6n-by-1, J is s-by-6n and \zeta is s-by-1? On page 16 it however states that elements of \eta are 6-vectors and I am not sure where does that follow?

3) What about \zeta, \lambda, C and C'? As far as I can tell the are s-by-1 matrices but what is their element type; 6-vector or real?

// ville

## iterative dynamics with temporal coherence clarifications?

### Re: iterative dynamics with temporal coherence clarification

I have received a reply from Erin Catto and have permission to post it here for others' benefit:

"

1) Right, they are both dot products. JV doesn't need the dot because you can view J as a row vector and V as a column vector, whereas J and B are both row vectors.

2) That is a typo, eta is s-by-1.

3) They are s-by-1 with scalar elements, one for each scalar constraint.

"

// ville

"

1) Right, they are both dot products. JV doesn't need the dot because you can view J as a row vector and V as a column vector, whereas J and B are both row vectors.

2) That is a typo, eta is s-by-1.

3) They are s-by-1 with scalar elements, one for each scalar constraint.

"

// ville