fserra

I guess @mueldgog would have to have a look at the issue. He is currently unavailable but I think he would be interested to fix this. After all is very...

I think we might have thought of 2 things: 1. We were always afraid of making pyscipopt deviate too much from SCIP. I guess this qualified as deviating too much...

To see the memory leak, you have to compile scip with `NOBLKMEM=true` and then run `make valgrind` on CSIP The leak is because of `&expr[i]`, but when creating the `CONST`...

The problem is that `POW` doesn't know `CONST`, nobody knows `CONST`. Here ```C exponent = values[children[begin[children[begin[i] + 1]]]]; ``` the code is getting the exponent directly from the values array....

Yes, it is just that we actually don't need the `CONST` for creating a `POW`... I can't remember why I did like that :(... maybe was simpler for julia... anyway,...