blog post test!
Heading 1 Heading 2 Heading 3
For my MS thesis, I studied the Whisker Frames mechanism through the lens of algebraic geometry, defining a vector model and visualizing my work using Desmos, an interactive graphing software. In this simply-defined geometry, I found a property that would come to redefine the system: all “whisker lines” (green) are tangent to a parabola. This sparked my interest, and led me to better charactarize the geometry, as shown below.
In the vector model, the control frame and face frame are defined as line segments in the {x,y} plane. The face frame is a unit-length line segment on the y-axis, running from the origin (0,0) to the point (0,1). The control frame is a line segment of length s, rotated an angle \(\theta\) from the y-axis and positioned away from the origin by the vector \(\boldsymbol{r}\). Because s cannot be changed in the hardware model, it is considered a constant. The position and length of the face frame is also fixed, attributing all degrees of freedom to the control frame. The configuration of the system is thus defined by 3 variables, forming the following configuration vector \(\boldsymbol{x}\)…
\[\boldsymbol{x}=\left[r_1,r_2,\theta\right]\in\mathbb{R}^3\]
Whisker directions are given by the vector \(\boldsymbol{u}{(y)}\). This vector is defined by vector subtraction of corresponding points \(\boldsymbol{F}\) and \(\boldsymbol{C}\), shown in the figure gallery below.
Rearranged, the equation for \(\boldsymbol{u}\) is defined below.
\[\boldsymbol{u}\left(y\right)=\ \mathbf{F}\left(y\right)-\ \mathbf{C}\left(y\right)\ =\mathbf{w}y-\boldsymbol{r}\] \[\mathbf{w}\ =\left[s\sin{\theta},1-s\cos{\theta}\right]^T\]Going forward, I would see the components of \(\mathbf{w}\) everywhere in the geometry.
In defining an output, I limited my attention to Protraction and Spread, two simple measures of whisking motion.
Protraction is the angle a whisker makes in the dorsal-ventral plane.
Spread is the full span of protractions produced by the whisker array. Rats can control the spread of their whiskers as a function of their sensing behavior. I define spread a bit differently: as the rate of change of protraction with respect to the distance on the face of the rat.
Analytically, protraction and spread are defined below.
\[\Theta(y)=\ \tan^{-1}{\frac{u_2}{u_1}} = \tan^{-1}{\frac{w_2y-r_2}{w_1y-r_1}}\] \[\frac{d\Theta}{dy}(y)\equiv\frac{r_1w_2-r_2w_1}{\left[r_1+w_1y\right]^2+\left[r_2+w_2y\right]^2}\]I started by asking myself the following question: where, on the xy plane, can there exist a whisker line? Up to this point, my analysis only involved the y-axis (for example, for \(\boldsymbol{u}{(y)}\)). So, if we were to make whisker lines in the direction of \(\boldsymbol{u}{(y)}\), where would those lines reach? Would it be everywhere?
My method was to work backwards: start with a point \((a,b) \in (x,y)\) and solve for what value of \(y\) points \(\boldsymbol{u}{(y)}\) in that direction.
\[y=\frac{w_1b-w_2a+r_1}{2w_1}\pm\frac{\sqrt{w_2^2{a}^2+\left(-2w_1w_2\right)a b+w_1^2{b}^2+\left(-2r_1w_2+4r_2w_1\right)a+(-2r_1w_1)b+{r_1}^2}}{-2w_1}\]It’s a bit messy! But there’s one key feature: this is the solution to a quadratic, meaning there are either one, two, or no solutions to whether (a,b) has a whisker running through it. This number of solutions depends on the square root term (discriminant). If we set the inside of the square root to zero, it will show us all values of a and b such that there is only one unique whisker line running though \((a,b)\).
\({\vec{\mathbf{x}}}^TA\vec{\mathbf{x}}=0\), where \(\vec{\mathbf{x}}=xy1\) and \(\ A=\left[\begin{matrix}w_2^2&-w_1w_2&-r_1w_2+2r_2w_1\\-w_1w_2&w_1^2&{-r}_1w_1\\-r_1w_2+2r_2w_1&-r_1w_1&r_1^2\\\end{matrix}\right]\)
Heading 1 Heading 2 Heading 3