- *
- * Normalisation:
- *
- * We want the minimum energy to remain unchanged with changes in
- * triangle densitiy, when the vertices lie evenly spaced on
- * circles, and we do this by normalising the force ie the
- * derivative of the energy with respect to linear motions of the
- * vertices.
- *
- * We consider only the force on Q due to PQR, wlog. (Forces on
- * P qnd R due to PQR are equal and opposite so normalising
- * forces on Q will normalise them too.)
- *
- * Force on Q is in the plnae PQR and normal to PR, so we can
- * consider it only linearly in that dimension. WLOG let that be
- * the x dimension. So with f' representing df'/dx_Q:
- *
- * , d
- * F = cost = --
- * Q,e Q,e err looks like we can only do
- * this if we make some kind of
- * assumption about delta or
- * something give up
- *
- *
- * Interposing M and N so that we have P-M-Q-N-R
- * generates half as much delta for each vertex. So
- *