Problem

Visual studio like to complain about this. He is right. God I love VS :)

Problem

Most of the files are in io. This is wrong. As not all of these things deal with io.

Acceptance Criteria

• everything makes sense.
• everything looks neat.
• make CMake happy.

Problem

qsts currently only works with single char variables names. This might be a problem for some people.

Suggested Solution

Implement a "scanner" of some kind that will scan a string or terminal input and generate a list (maybe) of parse_tokens (maybe). Alter the infix,postfix types accordingly.

Acceptance Criteria

• calling expression("hello + ginger * zing - (a + b-d)") will generate a valid expression.
• test, as always

Problem

Suppose we have a node [+, A,B]. The first component is a parent and the second and third are children. We want [+, A,B] == [+, B,A] to be true. As this agrees with our understanding of mathematics. Currently this is not implemented. One little pitfall is that we do NOT want [/, A,B] == [/, B,A] to be true as 9/4 != 4/9 etc.

Acceptance Criteria

• add the logic that solves the above problem.
• tests added that show the nodes are now unique for appropriate cases.

Problem

We rarely use template classes in qsts. So we should consider splitting these .hpp files into implementation (.cpp) and definition (.hpp) files for (almost) every class in the .hpp files. Within each .hpp file there are several classes which have now become quite bloated. Alternatively we could just split into more .hpp files

Acceptance Criteria

• Clean seperation of files
• More tests to be added
• All tests pass.

Problem

no unary functions

Suggested Solution

This is a two stage process. The first stage is getting qsts to parse unary operators, i.e. functions, and form graphs that contain them. The second stage is being able to differentiate unary functions.

For the first stage you will have to investigate how to deal with infix->postfix conversion where unary functions are part of the infix expression. This is not as simple as it looks and google will be your friend. The differentiation part will be quite simple once you can form expression graphs with functions.

AC

• Correctly obtain gradients of expressions containing unary functions.
• Tested with examples.