lisitsyn

# Tag public

template <typename T>
class Tag 
{
public:
    // registers tag in object
    Tag(SGObject* object, const char* name);
};

debasishg

1. Courses and MOOCs

• Introduction to Convex Optimization MITOCW by Prof. Stephen Boyd
• Convex Analysis and Optimization MITOCW by Prof. Dimitri Bertsekas
• Convex Optimization MOOC from Stanford Online
• Convex Optimization at CMU Spring 2015

2. Books

• Convex Optimization – Boyd and Vandenberghe - downloadable book
• Convex Optimization Theory - by Dimitri P. Bertsekas
• Theory of Convex Optimization for Machine Learning - download

karlnapf

from pylab import *

from shogun.Mathematics import LogDetEstimator
from shogun.Mathematics import ProbingSampler
from shogun.Library import SerialComputationEngine
from shogun.Mathematics import LogRationalApproximationCGM
from shogun.Mathematics import RealSparseMatrixOperator
from shogun.Mathematics import LanczosEigenSolver
from shogun.Mathematics import CCGMShiftedFamilySolver
from shogun.Mathematics import Statistics

gist:6381240

from pylab import *

from shogun.Mathematics import LogDetEstimator
from shogun.Mathematics import ProbingSampler
from shogun.Library import SerialComputationEngine
from shogun.Mathematics import LogRationalApproximationCGM
from shogun.Mathematics import RealSparseMatrixOperator
from shogun.Mathematics import LanczosEigenSolver
from shogun.Mathematics import CCGMShiftedFamilySolver
from shogun.Mathematics import Statistics

sonney2k

/*
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 3 of the License, or
 * (at your option) any later version.
 * 
 * Written (W) 2013 Soumyajit De
 */

#include <shogun/lib/common.h>

lambday

function_that_estimates_logdet
{
	CSparseMatrixOperator<float64_t> *op
	CSerialComputationEngine *e
	CLogRationalApproximationCOCGM log_op(op, e) 
	// or CLogRationalApproximationIndividual log_op(op, e)
	CProbingSampler *sampler
	CLogDetEstimator est(log_op, e, sampler)
	SGVector<float64_t> samples=est.sample(num_logdet_estimates)
}

karlnapf

function [V,colours] = probe( colours)
%% function [V] = probe(A,p)
%   Creates a probing matrix based on a graph colouring of A using the
%   algorithm in "J. Tang and Y. Saad, A probing method for computing the diagonal of the matrix
%   inverse" (2010).
%
%   REQUIRES: Matgraph toolbox http://www.ams.jhu.edu/~ers/matgraph/
%
%    Input:  colours   - A graph colouring (of A^p probably)
%            

lambday

/*
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 3 of the License, or
 * (at your option) any later version.
 *
 * Written (W) 2013 Heiko Strathmann
 */

#include <shogun/lib/config.h>

karlnapf

% samples from N(mu, C) in case type=='cov', and from N(mu, C^(-1)) in case
% type=='prec'. z is optional and must be drawn from N(0,I), inds is a
% permutations vector that is applied to C before sampling (doesnt effect
% the output, just efficiency, smyamd is used as default)
function sample=gaussian_sampler_symamd(mu, C, type, z, inds)
if nargin<3
    type='prec';
end

if nargin <4

jboner

Latency Comparison Numbers (~2012)
----------------------------------
L1 cache reference                           0.5 ns
Branch mispredict                            5   ns
L2 cache reference                           7   ns                      14x L1 cache
Mutex lock/unlock                           25   ns
Main memory reference                      100   ns                      20x L2 cache, 200x L1 cache
Compress 1K bytes with Zippy             3,000   ns        3 us
Send 1K bytes over 1 Gbps network       10,000   ns       10 us
Read 4K randomly from SSD*             150,000   ns      150 us          ~1GB/sec SSD