12345qiupeng
/
gtsam
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Ordering minor fixes

release/4.3a0
p-zach 2025-04-17 10:28:17 -04:00
parent 11f7742259
commit 2f0bff14b7
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
gtsam/inference/doc/Ordering.ipynb
View File

@ -17,7 +17,7 @@
"source": [ "source": [
"An `Ordering` specifies the order in which variables are eliminated during inference (e.g., Gaussian elimination, multifrontal QR). The choice of ordering significantly impacts the computational cost and fill-in (sparsity) of the resulting Bayes net or Bayes tree.\n", "An `Ordering` specifies the order in which variables are eliminated during inference (e.g., Gaussian elimination, multifrontal QR). The choice of ordering significantly impacts the computational cost and fill-in (sparsity) of the resulting Bayes net or Bayes tree.\n",
"\n", "\n",
"GTSAM provides several algorithms to compute good orderings automatically, such as COLAMD and METIS (if available), or allows you to specify a custom ordering." "GTSAM provides several algorithms to compute good orderings automatically, such as COLAMD and METIS, or allows you to specify a custom ordering."
] ]
}, },
{ {
@ -40,7 +40,7 @@
}, },
"outputs": [], "outputs": [],
"source": [ "source": [
"%pip install gtsam" "%pip install --quiet gtsam-develop"
] ]
}, },
{ {
@ -139,7 +139,7 @@
"GTSAM provides algorithms to automatically compute an elimination ordering from a factor graph. Two common algorithms are:\n", "GTSAM provides algorithms to automatically compute an elimination ordering from a factor graph. Two common algorithms are:\n",
"\n", "\n",
"1. **COLAMD (Column Approximate Minimum Degree):** A greedy algorithm that aims to minimize *fill-in* at each elimination step. It typically produces orderings that are good for sparse direct methods executed sequentially.\n", "1. **COLAMD (Column Approximate Minimum Degree):** A greedy algorithm that aims to minimize *fill-in* at each elimination step. It typically produces orderings that are good for sparse direct methods executed sequentially.\n",
"2. **METIS:** A graph partitioning algorithm (requires METIS library to be installed when compiling GTSAM). It aims to find orderings that partition the graph well, often leading to more balanced elimination trees which can be beneficial for parallel computation and sometimes reduce overall fill-in compared to purely local greedy methods like COLAMD, especially on large, structured problems.\n", "2. **METIS:** A graph partitioning algorithm. It aims to find orderings that partition the graph well, often leading to more balanced elimination trees which can be beneficial for parallel computation and sometimes reduce overall fill-in compared to purely local greedy methods like COLAMD, especially on large, structured problems.\n",
"\n", "\n",
"Let's illustrate the difference using a 2D grid factor graph." "Let's illustrate the difference using a 2D grid factor graph."
] ]
@ -391,7 +391,7 @@
"source": [ "source": [
"### COLAMD Ordering and Resulting Bayes Net\n", "### COLAMD Ordering and Resulting Bayes Net\n",
"\n", "\n",
"Now, we compute the COLAMD ordering and eliminate the variables according to this order. We then visualize the resulting Symbolic Bayes Net. The structure of the Bayes Net (specifically, the cliques formed by the conditional dependencies) reflects the structure of the elimination tree (or Bayes Tree)." "Now, we compute the COLAMD ordering and eliminate the variables according to this order."
] ]
}, },
{ {