I’ve always found that memorizing proofs or actually doing the exercises (as opposed to taking time to understand the structure of the solutions to some of them, if the main text doesn’t already cover the representative propositions) hits diminishing returns, in most cases anyway, when you are learning for yourself. The details get forgotten too quickly to justify the effort, the useful thing is to get good hold of the concepts (which by the way can be glossed over even with all the proofs and exercises, by relying on brittle algorithm-like technique instead of deeper intuition).
I don’t vote for blind memorization either. However, I think that if one can not reconstruct a proof than it is not understood either. Trying to reconstruct thought processes by heart will highlight the parts with incomplete understanding.
Of course in order to fully understand things one should look at additional consequences, solve problems, look at analogues, understand motivation etc. Still, the reconstruction of proofs is a very good starting point, IMO.
Sure. I’m pointing to the difference between making sure that you can do proofs (not necessarily reconstruct the particular ones from the textbook) and exercises, and actually reconstructing the proofs and doing the exercises. Getting to the point of correctly ruling the former can easily take 10 times less time than the latter. You won’t be as fast at performing the proofs in the coming weeks if need be, but a couple of years pass and you’d be as bad both ways (but you’d still have the concepts!).
I’ve always found that memorizing proofs or actually doing the exercises (as opposed to taking time to understand the structure of the solutions to some of them, if the main text doesn’t already cover the representative propositions) hits diminishing returns, in most cases anyway, when you are learning for yourself. The details get forgotten too quickly to justify the effort, the useful thing is to get good hold of the concepts (which by the way can be glossed over even with all the proofs and exercises, by relying on brittle algorithm-like technique instead of deeper intuition).
I don’t vote for blind memorization either. However, I think that if one can not reconstruct a proof than it is not understood either. Trying to reconstruct thought processes by heart will highlight the parts with incomplete understanding.
Of course in order to fully understand things one should look at additional consequences, solve problems, look at analogues, understand motivation etc. Still, the reconstruction of proofs is a very good starting point, IMO.
Sure. I’m pointing to the difference between making sure that you can do proofs (not necessarily reconstruct the particular ones from the textbook) and exercises, and actually reconstructing the proofs and doing the exercises. Getting to the point of correctly ruling the former can easily take 10 times less time than the latter. You won’t be as fast at performing the proofs in the coming weeks if need be, but a couple of years pass and you’d be as bad both ways (but you’d still have the concepts!).