[Sparsepop-users] Release SparsePOP 2.95

[Hayato W. <wa...@cs...>]

Dear All

We are happy to inform you that SparsePOP 2.95 is released.

<-- Changes in SparsePOP 2.95 -->

(1) For POPs which have constraints "x in {0,1}" and/or
"x in {-1, 1}", SparsePOP can reduce the size of the
resulting SDP relaxation problem by exploiting their
constraints. This reduction is based on the following paper:

J.B.Lasserre, "An explicit equivalent positive semidefinite
program for nonlinear 0-1 programs", SIAM J.Optim., 12,
pp. 756--769.

* For the reduction for a POP which has constraint "x in {0, 1}",
  set param.binarySW = 1.
* For the reduction for a POP which has constraint "x in {-1, +1}",
  set param.SquareOneSW = 1.

However, these reductions do not work when user computes error
bounds of an approximated solution obtained by SparsePOP.

(2) The reductions executed by reduceMomentMatSW = 1 and/or
reduceAMatSW = 1 work when user computes error bounds of an
approximated solution obtained by SparsePOP.

(3) We move "class pop_params" from conversion.h into Parameters.h.

(4) Some bugs are fixed.


Best regards
Hayato Waki