Hi, Does anybody know of a logarithmic interval implementation in Squeak where the steps in the interval go by logarithmic scale instead of linear ? Best, Karl |
Yes, But i don't remember if published somewhere... I'll check this evening or tomorrow Le dim. 3 févr. 2019 à 16:43, karl ramberg <[hidden email]> a écrit :
|
I used a workaround that probably is more flexible since I have several variables that need the logarithm steps := 100. xLog := 0.02 log: steps. yLog := 1.894 log: steps 1 to: steps do:[:i| x := i raisedTo: xLog. y := i raisedTo: yLog. ] Best, Karl On Sun, Feb 3, 2019 at 5:10 PM Nicolas Cellier <[hidden email]> wrote:
|
Hi karl, This is not a logarithmic interval. The progression should be geometric, that is with a constant ratio. x3/x2 = (x2/x1). The ith element of geometric serie should be something like (ratio raisedTo: i-1) * start. Le dim. 3 févr. 2019 à 18:15, karl ramberg <[hidden email]> a écrit :
|
You are right. It would be nice to have a logarithmic interval. Best. Karl On Sun, Feb 3, 2019 at 6:37 PM Nicolas Cellier <[hidden email]> wrote:
|
Hi Karl, I published GeometricInterval along with some GeometricIntervalTests at It is a port of VW SYSEXT-GeometricInterval from Cincom public store, a bit revisited. Le dim. 3 févr. 2019 à 21:18, karl ramberg <[hidden email]> a écrit :
|
And I have added a LogarithmicInterval which is more like numpy.logspace. For a LogarithmicInterval, bounds are specified by their logarithm (in whatever base). If you want 100 points from 0.1 to: 10.0, then you can write: (0.1 log logTo: 10 log size: 100 base: 10). or: (-1 logTo: 1 size: 100 base: 10) or you can omit the base and use ln/exp (0.1 ln logTo: 10 ln size: 100). Le ven. 8 févr. 2019 à 15:56, Nicolas Cellier <[hidden email]> a écrit :
|
Hi, Cool. I will try it soon. I have a cold at the moment :-( Cheers, Karl On Fri, Feb 8, 2019 at 10:28 PM Nicolas Cellier <[hidden email]> wrote:
|
I made test with a Mandelbrot set zoom with the logarithmic interval and it works really well :-) The zoom scale get down to ridiculously small numbers and it is only possible to do using logarithmic steps. Cheers, Karl On Sat, Feb 9, 2019 at 5:04 PM karl ramberg <[hidden email]> wrote:
|
Free forum by Nabble | Edit this page |