add docs

peterdsharpe · Nov 30, 2023 · 2846841 · 2846841
1 parent a44a21e
commit 2846841
Show file tree

Hide file tree

Showing 3 changed files with 12 additions and 0 deletions.
diff --git a/tutorial/01 - Optimization and Math/01 - Optimization Benchmark Problems/README.md b/tutorial/01 - Optimization and Math/01 - Optimization Benchmark Problems/README.md
@@ -14,10 +14,22 @@ This problem is chosen here because it shares many difficult aspects with engine
 
 The performance of AeroSandbox (with CasADi backend) is compared against existing methods using black-box optimization techniques. AeroSandbox offers faster practical and asymptotic optimization performance than existing black-box optimization methods, demonstrating the magnitude of acceleration that is possible.
 
+Exact code implementation of each method is [here](./nd_rosenbrock/run_times.py).
+
 ![benchmark_nd_rosenbrock](./nd_rosenbrock/benchmark_nd_rosenbrock.png)
 
 ## AeroSandbox vs. Disciplined Optimization Methods
 
+Here, we compare the performance of AeroSandbox (with CasADi backend) to existing methods using disciplined optimization techniques. AeroSandbox offers faster practical and asymptotic optimization performance than existing disciplined optimization methods, demonstrating the magnitude of acceleration that is possible.
+
+One disciplined optimization method used in aircraft design is geometric programming, with the [GPkit](https://github.com/convexengineering/gpkit) package an example implementation of this. Here, we compare the performance of AeroSandbox to GPkit on the GP-compatible [beam problem](https://gpkit.readthedocs.io/en/latest/examples.html#simple-beam) given in the GPkit documentation. 
+
+This problem is a structural analysis of a cantilever beam with a distributed load. As described in the GPkit documentation:
+
+> In this example we consider a beam subjected to a uniformly distributed transverse force along its length. The beam has fixed geometry so we are not optimizing its shape, rather we are simply solving a discretization of the Euler-Bernoulli beam bending equations using GP.
+
 In this chart, runtime is used instead of function evaluations, because the GPkit API doesn't easily expose this information from the underlying solver.
 
+Exact code implementation of each method is [here](./gp_beam/run_times.py).
+
 ![benchmark_gp_beam](./gp_beam/benchmark_gp_beam.png)
diff --git a/... Optimization and Math/01 - Optimization Benchmark Problems/gp_beam/benchmark_gp_beam.pdf b/... Optimization and Math/01 - Optimization Benchmark Problems/gp_beam/benchmark_gp_beam.pdf
diff --git a/...n and Math/01 - Optimization Benchmark Problems/nd_rosenbrock/benchmark_nd_rosenbrock.pdf b/...n and Math/01 - Optimization Benchmark Problems/nd_rosenbrock/benchmark_nd_rosenbrock.pdf