### 3.2.2 logxy

You may want to choose among log, semi-log, or ordinary linear scale axes. Use the logxy command:

logxy, 1, 1 // log-log plot logxy, 1, 0 // semi-log plot, x logarithmic logxy, 0, 1 // semi-log plot, y logarithmic logxy, 0, 0 // linear axis scales |

You can also omit the x or y flag to leave the scaling of that axis unchanged:

logxy, , 1 |

changes the y axis to a log scale, leaving the x axis scaling unchanged.

The flags returned by the limits function include the current logxy settings, if you need them in a program.

Zero or negative values in your data have no catastrophic effects with log axis scaling; Yorick takes the absolute value of your data and adds a very small offset before applying the logarithm function. As a side effect, you lose any indication of the sign of your data in a log plot. If you insist you need a way to get log axis scaling for oscillating data, write a function to treat negative points specially. For example, to draw positive-y portions solid and negative-y portions dashed, you might use a function like this:

func logplg(y, x) { s = sum(y>=0.); /* number of positive-y points */ n = numberof(y); if (s) plg, max(y,0.), x, type="solid"; if (s<n) plg, min(y,0.), x, type="dash"; } |

You always have the option of plotting the logarithm of a function, instead of using log axes:

logxy,0,1; plg,y,x |

plots the same curve as

logxy,0,0; plg,log(y),x |

The log axis scaling merely changes the position of the ticks and their labels; the y axis ticks will look like a slide rule scale in the first case, but like an ordinary ruler in the second.