On the nature of certain things

Replies to This Discussion

Permalink Reply by Athianarchist on August 12, 2009 at 12:25pm

Well integrate dy/dx = ky and you'll get y = be^(kx) like you said, but this is of course an exponential and y --> 0 as x --> -inf so technically y can never really "equal" zero in this situation. And when you set the derivative equal to a constant you of course won't be able to integrate it properly at all, it becomes dy/dx = c, ==>; y = cx, and you won't get the exponential at all. But I think what you meant to say in the middle part was:

y = be^(kx), k > 0
y = 0, k = 0
y = -be^(kx), k < 0

instead of y

Permalink Reply by Rick Taylor on May 8, 2012 at 10:55am

If you're solving this differential equation using separation of variables, you would start by writing:

dy/y = dx

Notice when you do this, you are dividing both sides of the equation by y. That means you are assuming that y is not 0 (because division by 0 is not defined). Therefore, the logic of your derivation is not valid when y=0, and you must test that as a separate case. In general, whenever a derivation involves dividing both sides of an equality by a quantity that may be zero, you must test what happens when the quantity is zero as a separate case.