I mean I want to make all the element which is larger than a value x to be x.

For a lower bound, if A is an Array:

A = A.max(x);

and if M is A Matrix:

M = M.array().max(x);

EDIT: For an upper bound, replace max by min.

ggael wrote:If A is an Array:

A = A.max(x);

and if M is A Matrix:

M = M.array().max(x);

Thank you! So, is that a usual way to avoid inf before using M.array().exp()?

ggael wrote:If A is an Array:

A = A.max(x);

and if M is A Matrix:

M = M.array().max(x);

It is strange, when I use max it acts like min, when I use min it acts like max.

Code: Select all

        MatrixXd x(2, 2);
   x(0, 0) = 800;
   x(0, 1) = 200;
   x(1, 0) = 30;
   x(1, 1) = 700;
   x = x.array().min(600);
   cout << x << endl;


the result is:
600 200
30 600

Code: Select all

        MatrixXd x(2, 2);
   x(0, 0) = 800;
   x(0, 1) = 200;
   x(1, 0) = 30;
   x(1, 1) = 700;
   x = x.array().max(600);
   cout << x << endl;


the result is
800 600
600 700

Of course, if you want an upper bound you have to replace "max" by "min" in my answer. min (resp. max) return the minimum (resp. maximum) coefficient wise.