"... In PAGE 4: ... Moreover, answers to pressing questions need to be addressed. How do effects like wrinkling, thermal loading, and manufacturing variability affect the vibrational response of membrane/inflatable space structures in general, and solar sails in particular? Can adaptive control methods use vibration information for local and global state estimates in such structures? 8,9 Table1 lists the types of measurements needed for validating structural analytical models and modeling techniques of solar sails. Also listed are estimated ranges of the measurement parameters and an accuracy target for each parameter assuming a sail size of 70 meters.... ..."

"... In PAGE 15: ... 5.4 Comparison of ParaSails and MultiSails In Table6 , we give some comparison results between MultiSails and ParaSails for solving a few sparse matrices. We see that when constructing a preconditioner, MultiSails usually spends less time than ParaSails to reach the same amount of sparsity ratio.... In PAGE 15: ... So the whole computational cost of these sparse matrices is also small, compared with computing a single sparse matrix with comparable density in the case of ParaSails. Also the data in Table6 show that with the same amount of memory cost (sparsity ratio), MultiSails usually has better convergence performance than ParaSails does. For... ..."

"... In PAGE 4: ... Such a sail could reach a large fraction of all NEOs within one year. Table1 contains the size of square sails necessary to return a 500 kg meteoroid with two different desired characteristic accelerations and two different sail masses per unit area. Even with a very low characteristic acceleration, the sails need to be nearly 200 meters on a side.... ..."

"... In PAGE 14: ... Sparsity ratio and iteration number. Table4 gives some results from using Multi- Sails with di erent steps to solve the FIDAP031 matrix. We note that 0 step here can be regarded as a standard sparse approximate inverse algorithm, i.... In PAGE 14: ... This will obviously increase both computational cost and memory cost for MultiSails with a large number of steps. In Table4 we notice that in the 5th step case, the preconditioner converges in 342 iterations, but the total computational cost is 5 times as much as that in the 1st step case. The reduction in the solution time does not compensate for the increase in the setup time.... In PAGE 14: ... So unless it does not converge with a lower number of steps or in the case of solving one matrix with many right hand sides, usually we do not recommend too many steps in real applications, even though that may yield better convergence results. In Table4 , we think that 1 or 2 steps is a good compromise between reasonable computational cost and good convergence results.... ..."