<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
    <title>Alex Mikhalev&apos;s blog</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/" />
    <link rel="self" type="application/atom+xml" href="http://www.sci-blog.com/alex/atom.xml" />
    <id>tag:www.sci-blog.com,2009-07-27:/alex//10</id>
    <updated>2010-05-24T06:30:28Z</updated>
    <subtitle>This is my personal blog for sharing scientific ideas, technologies and experience</subtitle>
    <generator uri="http://www.sixapart.com/movabletype/">Movable Type Pro 4.3-en</generator>

<entry>
    <title>Particle filter and Rao Blackwellised particle filter</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2010/05/particle-filter-and-rao-blackwellised-particle-filter.html" />
    <id>tag:www.sci-blog.com,2010:/alex//10.193</id>

    <published>2010-05-24T06:29:08Z</published>
    <updated>2010-05-24T06:30:28Z</updated>

    <summary><![CDATA[View imageAny improvements?Detection errors-----------------------------&nbsp;PF&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; = 0RBPF&nbsp;&nbsp;&nbsp; = 0&nbsp;&nbsp;Execution time&nbsp; (seconds)-------------------------&nbsp;PF&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; = 6.4465RBPF&nbsp;&nbsp;&nbsp; = 11.6923...]]></summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    <category term="particlefilter" label="particle filter" scheme="http://www.sixapart.com/ns/types#tag" />
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[<a href="http://www.sci-blog.com/alex/assets_c/2010/05/rbpf_pf-39.html" onclick="window.open('http://www.sci-blog.com/alex/assets_c/2010/05/rbpf_pf-39.html','popup','width=512,height=384,scrollbars=no,resizable=no,toolbar=no,directories=no,location=no,menubar=no,status=no,left=0,top=0'); return false">View image</a><br />Any improvements?<br />Detection errors<br />-----------------------------<br />&nbsp;<br />PF&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; = 0<br />RBPF&nbsp;&nbsp;&nbsp; = 0<br />&nbsp;<br />&nbsp;<br />Execution time&nbsp; (seconds)<br />-------------------------<br />&nbsp;<br />PF&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; = 6.4465<br />RBPF&nbsp;&nbsp;&nbsp; = 11.6923<br /><br />]]>
        
    </content>
</entry>

<entry>
    <title>Pingfire 1.0.4 compatible with firefox 3.6</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2010/05/pingfire-104-compatible-with-firefox-36.html" />
    <id>tag:www.sci-blog.com,2010:/alex//10.192</id>

    <published>2010-05-11T00:48:02Z</published>
    <updated>2010-05-11T00:52:36Z</updated>

    <summary>Updated version of pingfire extension. This is my fork with bit.ly supportScreenshot View imagepingfire.xpi...</summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    <category term="firefoxpingfireextensions" label="firefox pingfire extensions" scheme="http://www.sixapart.com/ns/types#tag" />
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[Updated version of pingfire extension. This is my fork with bit.ly support<br />Screenshot <a href="http://www.sci-blog.com/alex/assets_c/2010/05/Screenshot_pingfire-36.html" onclick="window.open('http://www.sci-blog.com/alex/assets_c/2010/05/Screenshot_pingfire-36.html','popup','width=351,height=328,scrollbars=no,resizable=no,toolbar=no,directories=no,location=no,menubar=no,status=no,left=0,top=0'); return false">View image</a><br /><br /><a href="http://www.sci-blog.com/alex/2010/05/11/pingfire.xpi">pingfire.xpi</a><br /><div><br /></div>]]>
        
    </content>
</entry>

<entry>
    <title>Updated pingfire extension</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2010/03/updated-pingfire-extension.html" />
    <id>tag:www.sci-blog.com,2010:/alex//10.191</id>

    <published>2010-03-26T05:15:06Z</published>
    <updated>2010-03-26T05:18:12Z</updated>

    <summary>Pingfire firefox extension compatible with 3.6.2 firefox. Usual disclaimer: I do it for my own use so no guarantee it will work for you.pingfire.xpiReport issues on bitbucket...</summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    <category term="extension" label="extension" scheme="http://www.sixapart.com/ns/types#tag" />
    <category term="firefox" label="firefox" scheme="http://www.sixapart.com/ns/types#tag" />
    <category term="pingfire" label="pingfire" scheme="http://www.sixapart.com/ns/types#tag" />
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[Pingfire firefox extension compatible with 3.6.2 firefox. Usual disclaimer: I do it for my own use so no guarantee it will work for you.<br /><a href="http://www.sci-blog.com/alex/2010/03/26/pingfire.xpi">pingfire.xpi</a><br /><br />Report issues on <a href="http://bitbucket.org/alex_mikhalev/pingfire/issues/">bitbucket</a><br />]]>
        
    </content>
</entry>

<entry>
    <title>Using forloop counter in django pagination</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2010/03/using-forloop-counter-in-django-pagination.html" />
    <id>tag:www.sci-blog.com,2010:/alex//10.190</id>

    <published>2010-03-25T23:40:48Z</published>
    <updated>2010-03-25T23:40:48Z</updated>

    <summary>In order to use forloop.counter with pagination use:...</summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[<p>In order to use forloop.counter with pagination use:</p>
]]>
        <![CDATA[<p>{% for object in object<em>list %}
    {{ page</em>obj.start_index|add:forloop.counter0 }}
{% endfor %}</p>
]]>
    </content>
</entry>

<entry>
    <title>Building nginx on ubuntu karmic</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2010/02/building-nginx-on-ubuntu-karmic.html" />
    <id>tag:www.sci-blog.com,2010:/alex//10.189</id>

    <published>2010-02-12T06:33:43Z</published>
    <updated>2010-02-12T06:36:00Z</updated>

    <summary>Building nginx on ubuntu...</summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[<p>Building nginx on ubuntu</p>
]]>
        <![CDATA[<p>sudo aptitude -R install build-essential libpcre3 libpcre3-dev libpcrecpp0 libssl-dev zlib1g-dev</p>

<pre>
./configure --prefix=/etc/nginx \
  --sbin-path=/usr/sbin/nginx \
  --conf-path=/etc/nginx/nginx.conf \
  --error-log-path=/var/log/nginx/error.log \
  --pid-path=/var/run/nginx.pid \
  --lock-path=/var/lock/nginx.lock \
  --user=www-data \
  --group=www-data \
  --http-log-path=/var/log/nginx/access.log \
  --http-client-body-temp-path=/var/lib/nginx/body \
  --http-proxy-temp-path=/var/lib/nginx/proxy \
  --http-fastcgi-temp-path=/var/lib/nginx/fastcgi \
  --with-cc-opt="-O2 -fno-strict-aliasing" \
  --with-http_gzip_static_module \
  --with-http_ssl_module \
  --with-http_geoip_module
</pre>
]]>
    </content>
</entry>

<entry>
    <title>Another magazine with my face</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2010/01/another-magazine-with-my-face.html" />
    <id>tag:www.sci-blog.com,2010:/alex//10.188</id>

    <published>2010-01-18T22:21:52Z</published>
    <updated>2010-01-18T22:24:44Z</updated>

    <summary><![CDATA[Does publication of my face in the magazine (large pdf), can be counted as academic? I am on page 23.&nbsp;...]]></summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    <category term="magazine" label="magazine" scheme="http://www.sixapart.com/ns/types#tag" />
    <category term="publication" label="publication" scheme="http://www.sixapart.com/ns/types#tag" />
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[Does publication of my face in the <a href="http://www.science.mod.uk/codex/documents/codex_issue5.pdf">magazine</a> (large pdf), can be counted as academic? I am on page 23.&nbsp; ]]>
        
    </content>
</entry>

<entry>
    <title>My PhD completed - Hurray </title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2010/01/my-phd-completed-hurray.html" />
    <id>tag:www.sci-blog.com,2010:/alex//10.187</id>

    <published>2010-01-18T20:09:10Z</published>
    <updated>2010-01-18T20:21:20Z</updated>

    <summary><![CDATA[Two of the brightest fellows, Dr Evan Hughes&nbsp; and Dr David Salmond (DSTL), who were my&nbsp; examiners, recommended the award of the PhD.&nbsp;...]]></summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[Two of the brightest fellows, <a href="http://www.cranfield.ac.uk/cds/staff/page11380.jsp">Dr Evan Hughes</a>&nbsp; and Dr David Salmond (DSTL), who were my&nbsp; examiners, recommended the award of the PhD.&nbsp; ]]>
        
    </content>
</entry>

<entry>
    <title>Cloud computing security, common questions and my answers</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2010/01/cloud-computing-security-common-questions-and-my-answers.html" />
    <id>tag:www.sci-blog.com,2010:/alex//10.186</id>

    <published>2010-01-15T23:58:31Z</published>
    <updated>2010-01-18T20:24:40Z</updated>

    <summary>I have a several friends, who have a great product ideas, based on complex scientific algorithms. Each time when I suggest using cloud computing for developing product we go through a round of questions. I will put them down here,...</summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    <category term="cloudcomputing" label="cloud computing" scheme="http://www.sixapart.com/ns/types#tag" />
    <category term="security" label="security" scheme="http://www.sixapart.com/ns/types#tag" />
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[<p>I have a several friends, who have a great product ideas, based on complex scientific algorithms. Each time when I suggest using cloud computing for developing product we go through a round of questions. I will put them down here, so I can reference it.</p>
]]>
        <![CDATA[<p>If you want to make online product in 2010, your solution should be scalable. It should be able to handle 10 users as well as 10000, one million. One of the ways to achieve it for startup is to leverage cloud computing architecture. Other way is to raise a lot of money and put your servers in secure datacenter under nuclear station in Switzerland. <br />
Cloud computing is insecure if you compare it with switch off computer with hard drive locked in the safe. Compared to other solutions it is comparably secure and security risks can be mitigated.  </p>

<p><em>Q: If we deploy it on somebody else infrastructure, we basically gave them away code with a lot of work behind.</em>  </p>

<p>A: This statement is only partially correct. Yes, member of the staff will theoretically be able to access running virtual appliance  from host server, yes, he theoretically can dump memory, then retrieve encryption keys and then debug your code into working algorithm. But, if you application so &#8220;interesting&#8221;, I bet you already have very secure storage - switched off computer and secure safe in some vault (and you work in finance, MOD HQ, MI5 or MI6 etc.), because if you are not, you are making this up and everything else you are doing potentially less secure then submitting job in the cloud.  </p>

<p><em>Q: Well, in this sense my home server is more secure in then the cloud.</em></p>

<p>A: When defending some data against malicious access, the only question matter &#8220;How much it costs?&#8221; see <a href="http://xkcd.com/538/" title="Security">this comic</a> for a simple example. It is race about how much you are ready to spend and how much your competitor will spend in order to get your data. Depending on security of your house, it may cost hundred of pounds/dollars to get your server (physically). Compare it to hacking into amazon cloud architecture or bribing their sysadmin (which one? in what zone?). I bet odds are not in favour of the home server. I am really against home server as home server doesn&#8217;t have scalability and have a high latency, which will be crucial for your users.  </p>

<p><em>Q: Well, if my home server will be stolen, there will be a trace like physical intrusion and then insurance can pay the damage.</em> </p>

<p>A: Patent your algorithm or application. Provisional patent will cost around 250  pounds (100$ for US) and will protect you for a year. This have a big advantage as you can talk about your application loudly, which may attract investors. </p>

<p>Amazon Elastic cloud pros:</p>

<ol>
<li>Strong firewall by default. Instance is not automatically visible to outside would. You have to configure firewall in order to be able to login to your own box. </li>
<li>Instance can be started only using X509 certificates or logging in using <a href="http://aws.amazon.com/security/#5" title="MFA">MFA</a>. MFA is a convenient standard way for online protection (my bank uses it). </li>
<li>Proper designed backend servers are invisible to outside world and can be located anywhere (eu or us), making attack difficult.</li>
<li>You can scale up and DOWN, which is very important for startup (leverage).</li>
<li>Traffic between nodes can be encrypted using products like VPN Cubed.</li>
</ol>

<p>I believe if you are paranoid about your application, you can encrypt some part of the virtual machine which contains protected application (cryptfs etc.), and then setup key server with rotating keys which will be used to decrypt &#8220;especially secret&#8221; part of the code during runtime. Of cause all configuration is secure as long as your key distribution server is secure (somewhere in Switzerland, under nuclear station, you know the drill), thus I think patenting is a better option. The bottom line is that you have product and your customers, or you don&#8217;t. Great application, sitting on switched off computer with hard drive doesn&#8217;t make any money. Cloud computing is a very cheap way to create great scalable architecture and compared to alternatives - home server, co-location, your own laptop, is no less secure.</p>
]]>
    </content>
</entry>

<entry>
    <title>Mysql DB and unicode utf8</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2009/12/mysql-db-and-unicode-utf8.html" />
    <id>tag:www.sci-blog.com,2009:/alex//10.184</id>

    <published>2009-12-25T00:25:21Z</published>
    <updated>2009-12-25T00:25:21Z</updated>

    <summary>Do you expect newest Ubuntu server 9.10 (karmic) to support utf8 in mysql server out of the box? Well, it doesn&#8217;t....</summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[<p>Do you expect newest Ubuntu server 9.10 (karmic) to support utf8 in mysql server out of the box? Well, it doesn&#8217;t.</p>
]]>
        <![CDATA[<p>Don&#8217;t forget to put small file <em>/etc/mysql/conf.d/charset.cnf</em>:</p>

<pre><code>[mysqld]
character-set-server=utf8
default-collation=utf8_unicode_ci
character_set_client=utf8
[mysql]
default-character-set=utf8
</code></pre>
]]>
    </content>
</entry>

<entry>
    <title>DIY Amazon EC2 Spot Instances price chart </title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2009/12/diy-amazon-ec2-spot-instances-price-chart.html" />
    <id>tag:www.sci-blog.com,2009:/alex//10.183</id>

    <published>2009-12-17T02:16:03Z</published>
    <updated>2009-12-17T18:32:49Z</updated>

    <summary>This is a short example how to build very simple chart for EC2 Spot Instances using Timeplot...</summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    <category term="amazonec2" label="amazon ec2" scheme="http://www.sixapart.com/ns/types#tag" />
    <category term="chart" label="chart" scheme="http://www.sixapart.com/ns/types#tag" />
    <category term="price" label="price" scheme="http://www.sixapart.com/ns/types#tag" />
    <category term="spotinstances" label="spot instances" scheme="http://www.sixapart.com/ns/types#tag" />
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[This is a short example how to build very simple chart for EC2 Spot Instances using <a href="http://www.simile-widgets.org/timeplot">Timeplot</a>]]>
        <![CDATA[<p>Following <a href="http://alestic.com/2009/12/ec2-spot-instance-prices">Eric Hammond example </a><br />
We are going to plot ec2 c1.medium prices against c1.medium in eu-west-1, to see how more expensive amazon ec2 instances in Europe</p>

<ol>
<li><p>Grab EC2 spot instance price:</p>

<p>ec2-describe-spot-price-history -t c1.medium -d Linux/UNIX |
 perl -ane 'print "$F[2],$F[1]\n"'&gt;&gt;data_ec2.txt </p></li><li>Add --region eu-west-1 to the command above so: <br /></li></ol><blockquote>ec2-describe-spot-price-history --region eu-west-1 -t c1.medium -d Linux/UNIX |
 perl -ane 'print "$F[2],$F[1]\n"'&gt;&gt;data_ec2_eu.txt<br /></blockquote>3. You also need to use two files <a aiotitle="time-plot.js" href="http://www.sci-blog.com/alex/2009/12/17/time-plot.js">time-plot.js</a> and <a href="http://www.sci-blog.com/alex/2009/12/17/ec2_plotter.html">html file</a><br />4. Save them in one directory (or use correct file paths)<br />5. That's it, folks. It should look like <a href="http://www.sci-blog.com/alex/2009/12/17/ec2_plotter.html" target="top"> this</a>. <br /><br /><br /><ol>

</ol>

]]>
    </content>
</entry>

<entry>
    <title>Mail count on Ubuntu Evolution</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2009/12/mail-count-on-ubuntu-evolution.html" />
    <id>tag:www.sci-blog.com,2009:/alex//10.182</id>

    <published>2009-12-09T00:15:34Z</published>
    <updated>2009-12-09T00:16:56Z</updated>

    <summary>I have to check exchange mailbox and I really like numbers appear today in my mailbox...</summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    <category term="evolutiongnome" label="evolution gnome" scheme="http://www.sixapart.com/ns/types#tag" />
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[I have to check exchange mailbox and I really like numbers appear today in my mailbox<img alt="Screenshot.png" src="http://www.sci-blog.com/alex/Screenshot.png" class="mt-image-right" style="margin: 0pt 0pt 20px 20px; float: right;" height="54" width="473" />]]>
        
    </content>
</entry>

<entry>
    <title>But why?</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2009/11/but-why.html" />
    <id>tag:www.sci-blog.com,2009:/alex//10.181</id>

    <published>2009-11-22T07:57:30Z</published>
    <updated>2009-11-22T07:57:30Z</updated>

    <summary>One of the questions which drives me crazy recently - why. There is particle filter (bootstrap, condensation, simulated annealing) algorithm which is popular and works well(sometimes). But why? There is a Kalman filter algorithm - which doesn&apos;t work on geolocation...</summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[<p>One of the questions which drives me crazy recently - why.   <br />
There is particle filter (bootstrap, condensation, simulated annealing) algorithm which is popular and works well(sometimes). But why?<br />
There is a Kalman filter algorithm - which doesn't work on geolocation problem at all (EFK as well), but still there are papers on the topic. But why?<br />
There are genetic algorithms and they perform reasonably well. But why?<br />
There is a genetic programming and it works (sometimes). But why?<br />
There is an assumption that the processes can be described as Gaussian (and very few can). But why?<br />
There is  BAYESIAN law. which works (occasionally). But why?</p>]]>
        
    </content>
</entry>

<entry>
    <title>My old presentation for students about presentation</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2009/11/my-old-presentation-for-students-about-presentation.html" />
    <id>tag:www.sci-blog.com,2009:/alex//10.180</id>

    <published>2009-11-12T08:25:11Z</published>
    <updated>2009-11-12T08:26:59Z</updated>

    <summary>presentation.pdfMost credit goes to Garr Reynolds...</summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    <category term="presenation" label="presenation" scheme="http://www.sixapart.com/ns/types#tag" />
    <category term="science" label="science" scheme="http://www.sixapart.com/ns/types#tag" />
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[<a href="http://www.sci-blog.com/alex/2009/11/12/presentation.pdf">presentation.pdf</a><br />Most credit goes to <a href="http://www.presentationzen.com/">Garr Reynolds </a><br />]]>
        
    </content>
</entry>

<entry>
    <title>Changed MT installation from sqlite to mysql</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2009/10/changed-mt-installation-from-sqllite-to-mysql.html" />
    <id>tag:www.sci-blog.com,2009:/alex//10.179</id>

    <published>2009-10-08T03:31:17Z</published>
    <updated>2009-10-08T03:33:50Z</updated>

    <summary>Some things broke. I am really dissapointed by mt backup/restore procedure. It doesn&apos;t restore properly....</summary>
    <author>
        <name>Alex Mikhalev</name>
        
    </author>
    
    <category term="mtinstallationweb" label="mt installation web" scheme="http://www.sixapart.com/ns/types#tag" />
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        Some things broke. I am really dissapointed by mt backup/restore procedure. It doesn&apos;t restore properly. 
        
    </content>
</entry>

<entry>
    <title>Bandsharing capacity</title>
    <link rel="alternate" type="text/html" href="http://www.sci-blog.com/alex/2009/09/bandsharing-capacity.html" />
    <id>tag:sci-blog.com,2009:/alex//2.150</id>

    <published>2009-09-07T01:09:04Z</published>
    <updated>2009-09-07T01:10:43Z</updated>

    <summary>http://mashable.com/2009/09/03/iphone-att/My main project is to develop algorithms and methods for wireless bandsharing. It is estimated that it will be an issue by 2017. I think it started to hit us now....</summary>
    <author>
        <name>getfriday</name>
        <uri>http://www.media-den.co.uk/</uri>
    </author>
    
    <category term="science" label="science" scheme="http://www.sixapart.com/ns/types#tag" />
    
    <content type="html" xml:lang="en-us" xml:base="http://www.sci-blog.com/alex/">
        <![CDATA[http://mashable.com/2009/09/03/iphone-att/<br />My main project is to develop algorithms and methods for wireless bandsharing. It is estimated that it will be an issue by 2017. I think it started to hit us now.<br /><br /><br />]]>
        
    </content>
</entry>

</feed>
