Stop Cargo Culting BandGeneral and Plain Numberer

In the early days of OpenSees, perhaps in its G3 infancy, an example Tcl script used BandGeneral for the system and Plain for the numberer. numberer Plain system BandGeneral Who created the script, or why they chose those analysis options, is unimportant now–it could have been me for all I know. But this pair of … Continue reading Stop Cargo Culting BandGeneral and Plain Numberer →

Empty Spaces

Recently, a large OpenSees model was posted in an online forum with the poster asking why the analysis took longer than expected. Short answer: Not only did using a heavy-duty recorder that writes all node, element, and section data take up a big chunk of time, but using OpenSees's default linear equation solver, ProfileSPD, also … Continue reading Empty Spaces →

Reverse Engineering the Equation Numberer

In OpenSees (and any other finite element software), equation numbering is a quiet, behind the scenes analysis option that users do not have to pay any mind. No matter how a user numbers the nodes in their model, e.g., for bookkeeping or from a mesh generator, the equation numberer will clean up any messes. But … Continue reading Reverse Engineering the Equation Numberer →

Choose Your Own Topology

I've been working on a sparse linear equation solver. Not anything new, just implementing the methods presented by Timothy Davis in Direct Methods for Sparse Linear Systems. Why? Because I want to learn how sparse matrix solvers work. So for the sake of discussion, let's call my solver BennySparse. Aside from implementing a standalone sparse … Continue reading Choose Your Own Topology →

The Good, the Not So Bad, and the Full General

Just like shopping for a new refrigerator, picking a linear equation solver in OpenSees (via the system command) can lead to paralysis of choice. And while you can consult Consumer Reports for the pros and cons of refrigerators A, B, and C, the only way to figure out the pros and cons of OpenSees solvers … Continue reading The Good, the Not So Bad, and the Full General →

OpenSees Spy

A previous post on the 'fullGenLapack' eigenvalue solver led me down a rabbit hole of sorting out all the equation solvers (linear and eigen) and equation numberers available in OpenSees. I have yet to emerge from the rabbit hole, but I wanted to show how to create "spy" matrices from OpenSees models. A spy matrix … Continue reading OpenSees Spy →

Make Room for Storage

In a previous post, I showed how equation numberers can reduce the bandwidth of the tangent stiffness matrix. In addition to reducing the solution time for linear systems of equations, a smaller bandwidth reduces the data required to store the tangent stiffness matrix. To discuss matrix storage in this post, I'll use the frame model … Continue reading Make Room for Storage →

Reduce Your Bandwidth

For large structural models, the solution to $latex {\bf K}_T\Delta {\bf U}={\bf R}$ can be the computational bottleneck during an analysis. Although computing speed and algorithms to solve $latex {\bf K}_T\Delta {\bf U}={\bf R}$ are very good, you still want to make sure the solution happens as quickly as possible, particularly when inside the double … Continue reading Reduce Your Bandwidth →