Skip to content

Latest commit

 

History

History
75 lines (62 loc) · 2.14 KB

lec26.md

File metadata and controls

75 lines (62 loc) · 2.14 KB

Lecture #26: 3D in matplotlib, testing and continuous integration

Lecture Objectives

Textbook Reference

Activities

  • jupyter notebook with:
    • seq_driver imported and seq_dict = seq_driver.seq_driver_incr(length, length)
import mpl_toolkits.mplot3d.axes3d as p3
import numpy as np

fig = plt.figure()
ax = p3.Axes3D(fig)

for key, seq in seq_dict.items():
   ax.plot(key[0]*np.ones(length),range(length),seq, label=str(key))

ax.legend()
  • surface plot
    • re-arrange data into a matrix
Z = np.ndarray((len(initial),length))
for key, seq in seq_dict.items():
    Z[key[0]-2,:] = seq
  • X & Y grid data for each point in matrix
initial = [key[0] for key in seq_dict]
X,Y = np.meshgrid(range(length),initial)
  • plot surface:
fig = plt.figure()
ax = p3.Axes3D(fig)

ax.plot_surface(X,Y,Z)

  • testing

    • testing is a vital part of all modern software projects
      • confirms/demonstrates/defines expected behavior
      • demontrates bugs and then demonstrates that bugs have been fixed
      • prevents regression
    • unit tests
      • specific to individual functions
      • test behavior under many possible conditions, including error conditions if errors are a possible response
      • should be fast and automated so that there is no disincentive for running them often
      • can include input/output of files
    • consider a function you are writing and describe some tests
    • integration tests
      • ensure that an complete software project does what is expected
      • often related to verification and validation
    • verification
      • are the equations implemented correctly
    • validation
      • are the correct equations implemented
    • uncertainty quantification
      • characterization of the uncertainty in the result of a simulation
      • related to validation

  • Continuous integration

    • fundamentally: merging updates to the mainline on a frequent basis
    • key requirement:
      • automated building of software (less relevant for Python)
      • automated testing of software as it's proposed
      • testing in a realistic environment
    • services exist to assist with this, e.g. CircleCI