VBM long term planning and bug report

Brief explanation

This is a page with internal destination for the LCVM2 lab and its close collaborators. Anyone outside these categories is not concerned.

The purpose of this page is to keep everyone informed with the progress, plans and bugs in VBM otherwise than by emails that people may not read and keep. Everyone that has a bug to report about VBM or a suggestion to make about future development of the software is encouraged to send an email to

Your suggestion or bug report should rapidly appear on this page, in one of the 3 sections below.

The features on this list that have already been implemented are displayed in red.


Plan for VBM 1.0

Test the configure scripts on as many different machines as possible. (Randy will do this)

The rod version of the new data_probe needs to be finished and documentation written. (Dana will do this)

The demos need to be updated to work with the new module.  This is mainly a matter of documentation. (Randy will figure out exactly what modules are needed with each demo, and then Randy and Dana will do the documentation updates).

Get rid of Tix.  (Randy will do this).

Make setup report the current values of the parameters, not those stored in the file. Be able to store the changes in the file too.

Correct the problem about the working directory, which causes a continuation module to delete all files in it on quit. Possible patch for it is to check that this directory is empty to begin with, otherwise create a subdirectory that can be deleted at the end.

Add a third option to the data panel (general and rod-specific) combining the fort8 and IVP options. With this option, one can first read some data from the fort8 file, then add new columns to it by running the IVP solver.

Add a special tag bloc for the reconstruction. It means that the reconstruction may use other right hand sides than the continuation.

Add the vbm filename to the bifurcation diagram and to the dataprobes in the title.

Plan for VBM 1.1

Indexing the fort.8 files.

Custom color maps. Be able to introduce your own color map in VBM, and/or to set the range for the colormap from VBM.

Including the new IVP solver from Julian de Siebenthal.

Other computer engines-
          Parallel "C" AUTO

In the dataprobe(s), being able to visualize a smaller number of points than the ones computed, either by selecting a range, or by culling the data by a factor.

Interacting with the solution, like, can I see the xyz values in this point. Could be mixed with the pointwise frame.

Improve the computational speed for the markers. First, if one only changes geometrical aspects of the markers (like the sphere size), the markers should not be recomputed, which is sometimes the case now. Second, maybe the computation itself could be improved, but the probability is not high.

The markers color does not get saved in the saveview operation.

Set the pointer to the 0 0 group and point when loading a new file. This sometimes causes VBM to crash because of index out of range when loading another file (it does not happen to the first file that gets loaded).

Animate the bifurcation diagram: walk along the branch and see the solution change.

Combine multiple bifurcation diagrams in 1 view: reconstruct using different uhats and Ks.

Combine several solutions in 1 view.

Axes locator - scaling and centering the axes automatically with the data.

Center the solution with the center button.

Open questions and bugs

Hardcopy to postcript.

Color map discrete: how to assign colors to integers. It can be turned down to defining a range for the existing discrete colormap.

How to automatically postprocess in the bifurcation diagram ? How can this be combined with continuation inside VBM?

If the VBM file is compressed, decompress it automatically on load. This is a difficult question because each platform has its own compressing and decompressing programs, whose options may be different too.

Transparency (more of a DV thing then a VBM thing, but it would be useful in what Sebius and Randy are working on).

Add text labels to the axes.

Interpolate the uhats from the uhat data with a smoother function (polynomial).

Add a 2D dataprobe for publication purposes, probably using Matlab.

Last updated: September 5, 2000, dana.vrajitoru@epfl.ch.