Analog Computational Matrix
<< Back to the Capstone Index.
I believe there may be value in defining complexity classes for analog configurations. Currently, I tend to sort things into three general categories:
- Fixed visual applications, such as the butterfly wing assignments
- Fixed computational applications, such as XOR and other digital logic approximations
- Dynamic computational applications, such as the proposed RED approximation and my work toward evolution in analog
Two axes are immediately apparent to me: fixed vs dynamic applications and visual vs. compuational applications. Plotted as a matrix, I believe a simple classification system emerges:
| Fixed | Dynamic | |
|---|---|---|
| Visual | [1] Butterfly Wing | [3] ? |
| Computational | [2] Digital Logic | [4] Alife |
As you can see, quadrant (3) appears empty. I don't know of any dynamic visual applications. Perhaps the analog retina or cyclotron beam controller fit into this category, perhaps not. I tend to envision some kind of continuous waveform analyzer as the instantiation of this category.
In any case, I believe higher quandrants represent more interesting problems. That is, purely fixed visual applications (those where we are primarily interested in the shape of the gradient) are less interesting than dyanamic computational applications (those where we are primarily interested in the values of LLAs).
Whether this means anything or not is open to debate. I find it useful to distinguish between different types of analog applications, but that may just be me.