• Categories

  • Archives

Revamped DifferentialOperator and pull request cleanup

The first task that I handled this week was to generalize DifferentialOperator so that it can handle more complicated expressions like a Laplacian or something similar. Its constructor now operates in three different modes: (see the commit here)

  1. A single symbol or string: this represents the variable to differentiate with respect to. This a simple \frac{d}{dx} type operator.
  2. A tuple with two entries: the first entry of the tuple is the same as in 1), but the second is an integer representing the order of the derivative to be taken. Therefore, DifferentialOperator((“x”, 2)) would represent \frac{d^2}{dx^2}
  3. Two arguments, an Expr and a tuple: the first entry is a general expression involving a Function. The second argument is a tuple. The first entry in the tuple is the symbol of the Function in argument 1 which is to be replaced by the Wavefunction we apply this operator to. The second entry in the tuple is the symbol which the function is evaluated with before replacement. So, another way of expressing a single derivative would be DifferentialOperator(Derivative(f(x), x), (f, x)) were f = Function(‘f’) and x = Symbol(‘x’). This allows us to put any arbitrary expression in and have it applied to the Wavefunction accordingly.

Once the new DifferentialOperator was implemented, it was time to add this functionality to As a result, it is now possible to represent X operators in the momentum basis and vice versa!

In addition to the DifferentialOperator changes, this week was about coming up with more tests and trying to find bugs in corner cases of representations. One bug related to integration in sympy was discovered when testing out representations that involve DifferentialOperator. DifferentialOperator is non-commutative, and as_independent, which is called by the integration routine, would treat these operators as if they did not depend on the variable of integration even when they did. I opened issue 2549 here and then shortly after had a pull request here with a solution that eventually got merged in. This meant that DifferentialOperators were now integrated over correctly, so more tests were added that took advantage of this new functionality.

I had a brief discussion with Brian on IRC about outstanding issues in the PR, and one thing that we both agreed on was the the enumerate_states function of would not work for very general kets. The way it was previously structured was to simply take the labels of kets and append indices to the end of them. This works for continuous bases, and is useful when trying to form an inner product like \langle x_1 | x \rangle, but it’s completely useless for kets of discrete bases. This commit factors the actual enumeration of the states out to an internal ket function, leaving’s enumerate_states as the public interface. I think this design is ideal, because this means that each ket can decide for itself how to handle a request for a certain number of kets from the basis.

Hopefully Brian and I can discuss any leftover outstanding issues with my PR soon and we can get it merged in. After that happens, I’m going to start implementing many different coordinate systems and lay the groundwork for being able to handle many more example systems. This point in my schedule is when I already wanted to be implementing the coordinate systems, so there’s a chance that I may fall a bit behind that schedule. Luckily, I incoporated a 3 week buffer into the end of the timeline, just in case that happened, so I think I’m still on track to get a nice working product by the end of the summer.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: