Starting on rigde opt

docs/source/conf.py

@@ -69,7 +69,7 @@
 # List of patterns, relative to source directory, that match files and
 # directories to ignore when looking for source files.
 # This pattern also affects html_static_path and html_extra_path .
-exclude_patterns = []
+exclude_patterns = ['**.ipynb_checkpoints']
 
 # The name of the Pygments (syntax highlighting) style to use.
 pygments_style = 'sphinx'

docs/source/index.rst

@@ -10,17 +10,10 @@ Welcome to Parameter Space Dimension Reduction's documentation!
    :maxdepth: 2
    :caption: Contents:
 
+   install
    tutorial
+   api
 
-   domain 
-   basis
-   function
-   pools
-   samplers
-   subspace
-   surrogate
-   opt
-   demos
 
 Indices and tables
 ==================

docs/source/install.rst

@@ -3,9 +3,9 @@ Installation
 
 
 As this project is currently in active development, I recommend installing from
-github directly:
+github directly
 
-code ::
+::
 	
 	pip install git+https://github.com/jeffrey-hokanson/PSDR#egg=psdr
 

psdr/__init__.py

@@ -14,4 +14,6 @@
 from .redis_pool import RedisJob, RedisPool, RedisWorker
 #from opt.shared import LinProgException, InfeasibleConstraints
 from .pool import Pool, SequentialJob, SequentialPool
-from .sample import maximin_sample  
+from .sample import maximin_sample 
+
+

psdr/function.py

@@ -47,16 +47,19 @@ class Function(BaseFunction, Domain):
 	fun_kwargs: dict, default: empty
 		Keyword arguments to pass to the functions when evaluating function 
 	"""
+
+	# TODO: Implement fancy pool-features
+
 	def __init__(self, funs, domain, grads = None, fd_grad = None, vectorized = False, fun_kwargs = None):
 		self.funs = funs
-		self.domain = domain
+		self.domain_app = domain
 		self.domain_norm = domain.normalized_domain()
 		self.fd_grad = fd_grad
 
 
 	def eval(self, X_norm):
 		X_norm = np.atleast_1d(X_norm)
-		X = self.domain.unnormalize(X_norm)
+		X = self.domain_app.unnormalize(X_norm)
 
 		if len(X.shape) == 1:
 			x = X.flatten()

psdr/polyridge.py

@@ -190,7 +190,8 @@ class PolynomialRidgeApproximation(PolynomialRidgeFunction):
 	leaving an optimization problem posed over the Grassmann manifold alone.
 
 	For both the 1-norm and the :math:`\infty`-norm,
-	this implementation uses ........
+	this implementation uses a sequential linear program with a trust region
+	coupled with a nonlinear trajectory through the search space.
 
 	Parameters
 	----------

psdr/ridge.py

@@ -54,7 +54,8 @@ def shadow_plot(self, X = None, fX = None, dim = None, ax = None):
 				linewidths = 0.5)
 
 
-
+		else: 
+			raise NotImplementedError("Cannot draw shadow plots in more than two dimensions")	
 
 
 

psdr/ridgeopt.py

@@ -2,19 +2,45 @@
 
 
 import numpy as np
+from function import Function
 from polyridge import PolynomialRidgeApproximation 
 
 
 class RidgeOptimization:
+	r""" Ridge-based nonlinear optimization
 
+	Given a vector valued function :math:`\mathbf{f}: \mathcal{D}\subset \mathbb{R}^m\to \mathbb{R}^n`,
+	this class solves the optimization problem
+
+	.. math::
+
+		\min_{\mathbf{x}\in \mathcal{D}} &\ f_0(\mathbf{x}) \\
+		\text{such that} & \ f_i(\mathbf{x}) \le 0
+
+	by constructing a sequence of bounding surrogates.
+
+	Parameters
+	----------
+	func : Function
+		Function to optimize	
+
+	"""
 	def __init__(self, func, X = None, fX = None, pool = None):
+		self.func = func
+		self.domain = func.domain_norm 
 		self.X = []
 		self.fX = []
 
-
-
 	def step(self):
+
 		pass
 
 if __name__ == '__main__':
+	from demos import golinksi_volume, golinksi_design_domain
+
+	domain = golinksi_design_domain()
+	func = Function(golinksi_volume, domain)
+
+	x = func.sample(1)
+	print func(x)
 	pass