import distl
import numpy as np

Math with Integers/Floats

g = distl.gaussian(10, 2)

out = g.plot(show=True)

png

multiplying a Gaussian distribution by a float or integer treats that float or integer as a Delta distribution. In this case, it is able to return another Gaussian distribution. When unable to return a supported distribution type, a Composite distribution will be returned instead.

print(2*g)

<distl.gaussian loc=20.0 scale=4.0>

print(g*2)

<distl.gaussian loc=20.0 scale=4.0>

out = (g*2).plot(show=True)

png

Note that this gives us the same resulting distribution as if we defined a Delta distribution manually and multiplied by that.

d = distl.delta(2)

print(g*d)

<distl.gaussian loc=20.0 scale=4.0>

out = (g*d).plot(show=True)

png

If we multiply in the other order, we'll get a Composite distribution, but see that it gives us the same resulting sample.

print(d*g)

<distl.delta loc=2.0> * <distl.gaussian loc=10.0 scale=2.0>

out = (d*g).plot(show=True)

png

Note that since the floats/integers are treated as distributions, 2*g is not equivalent to g+g (note the change in the x-limit scale on the plot below compared to those above).

print(g+g)

<distl.gaussian loc=10.0 scale=2.0> + <distl.gaussian loc=10.0 scale=2.0>

out = (g+g).plot(show=True)

png

Supported Operators

  • multiplication
  • division
  • addition
  • subtraction
  • np.sin, np.cos, np.tan

(for and/or logic, see these examples)

g = distl.gaussian(2*np.pi, np.pi/6)

print(np.sin(g))

sin(<distl.gaussian loc=6.283185307179586 scale=0.5235987755982988>)

out = g.plot_pdf(show=True)

png

out = np.sin(g).plot_pdf(show=True)

png

out = np.sin(g).plot_cdf(show=True)

png

out = np.sin(g).plot(show=True)

png

out = np.cos(g).plot(show=True)

png

out = np.tan(g).plot(show=True)

png