Sunday, January 4, 2015

Simple RK4 Code

January 4, 2015 10:11 pm 

Thought I'd start off the New Year with a simple freebie, a 4th Order Runge-Kutta class for Delphi, Windows and Mac OS. Should also work on Android and iOS mobile platforms and with Free Pascal. The code below shows how you'd use it. First declare the set of differential equations that need to be solved (TDoubleDynArray type can be found in Types):

 TMySystem = class (TObject)
    class procedure func (time : double; y, p, dydt : TDoubleDynArray);
 end;

class procedure TMySystem.func (time : double; y, p, dydt : TDoubleDynArray);
begin
  dydt[0] := p[0] - p[1]*y[0];
  dydt[1] := p[1]*y[0] - p[2]*y[1];
end;

Next, create a Runge-Kutta object:

// First argument = number of ODEs
// Second argument = number of parameters, if any
// Third argument = ODE Function
rk4 := TRK4.Create (2, 3, TMySystem.func);
rk4.p[0] := vo; rk4.p[1] := k1; rk4.p[2] := k2;

The above code include the option to declare and initialize some parameters that are part of the differential equations. To actually integrate the system by one step use the line:

// Return new time point, assign to startTime ready for next time
startTime := rk4.eval (startTime, y, stepSize);

The y argument will contain the updated solution. Run the line again to do the next step etc. Note that the eval method takes three arguments, that current value for the independent variable (time), an array that contains the current values for the dependent variable, and the step size to use.

Download the code here. RK4.zip The download include a Windows exe that you can try right away.

Tuesday, October 21, 2014

My First Mac OS Application

October 21, 2014 8:36 pm

I finished my first Mac OS application. Two screenshots are shown below, I'll be making it available download at the end of the week. This was a test application to see how easy it would be use write a portable application using Firemonkey. The application in question is a specialist tool that allows users to load systems biology models expressed in SBML models and simulate them. It also has some limited steady state functionality. It users libRoadRunner as the backend simulator (libroadrunner.org). Most of the underling UI business code is new but the graphing component is one I resurrected from an old Windows project I had. Took me a week or two to extract the code from the old VCL Delphi application and swap in Firemonkey canvas methods. The biggest issue I had was remembering that Firemonkey now uses single precision values to represent canvas coordinates (VCL uses integer values).

I developed the application first on Windows using XE6. Once it was working on Windows I compiled it for Mac OS. Other than a few very minor issues (eg forgetting to remove Windows in the uses clause) getting it running on the Mac was surprisingly uneventful. It took me about 20 mins to get it working on the Mac, I was impressed. Pserver, which is the conduit for moving a Delphi application to the Mac makes the process child's play. It copies over the relevant files and metadata and creates the bundle on the Mac's hard drive ready to be distributed.

Some gotchas to watch out for:

1. Scroll bars don't show automatically on the Mac in scrollable controls, to make them show up include the following line in your main form OnCreate event:

memo1.AniCalculations.AutoShowing:=false;

where memo1 is the name of a memo control.

2. Use the Deployment option in the Project menu to specify any Mac OS dylibs you want to distribute. pserver will handle the copying and bundling of these automatically.

3. Loading dll and dylibs. Not really a gotcha but I thought it would be tricky. This was surprisingly easy and in fact no different from Windows.

DllHandle :=LoadLibrary(PWideChar (fullpath));

Obviously the path to the libraries will change because the files extensions are different.

but I need this code to deal with any errors in the loading:

{$IFDEF POSIX}
errStr := string(dlerror());
{$ELSE}
errStr := SysErrorMessage(Winapi.Windows.GetLastError);
{$ENDIF}

Use of GetProcAddress is identical in Windows and Mac OS.

4. I couldn't find the equivalent of GetDeviceCaps on the Mac so assumed the same size for the Mac and Windows.

5. I didn't realize the current version of Delphi (even XE7) generates 32-bit Mac OS binaries which means it can't load any 64-bit libraries. This posed a bit of a problem initially as we only distribute libRoadRunner as a 64 bit library and the current guardian of libRoadrunner was unable to create a 32-bit version. Luckily one of my other students, (Kyle Medley) was savvy enough to compile a 32-bit version which saved the day.

6. Bugs in XE6:

a) Don't use a separate stylebook for each form because closing and freeing a second form will corrupt the buttons or other visual element on the calling form.

b) TextHeight returns incorrect an value for an empty string

7. The Firemonkey TStringGrid is severely limited or at least it seems that way. It's currently a poor cousin of this VCL partner. Not sure how to get round its limitations yet.

8. The trackar is a limited control, there is no way to alter the thumb size or shape, track width, etc.

9. Various visual artifacts on some of the styles, eg Auric has problems with thumbs in tack bars.

10. The online documentation on menu handling between Mac OS and Windows appears to be completely incorrect.

Screenshots from a Mac Pro:







Tuesday, September 9, 2014

What do 17 year old science students know (or not)?

Originally Posted on  by hsauro

This summer I had the opportunity to host five high school students at a UW science camp. I decided to give them a project where they could build circuits that could do addition and multiplication just using resistors.

I’ve been a high school teacher in the past so I though I knew the general level of education for 17 year olds. Let me remind everyone that we’re talking about students who are one year away from entering University. Let me say that again, they are one year away from entering University. This is what I found out:

1. They’ve never seen or used a multimeter before
2. They didn’t know what the symbol was for a battery or a resistor
3. They didn’t understand what the lines were between the circuit symbols (wires!)
4. They’d never heard of a capacitor
5. They’d never seen a breadboard before and had no idea what to with it.
6. They didn’t know the difference between DC and AC (or even what they were)

What is surprising was they did know about ohms law but didn’t know the symbol for an ohm. I am certain that if I asked them to wire up a battery and a light they would have been lost.

Is this the same country that landed men on the moon?


Saturday, August 16, 2014

Roman Fort at Clunderwen?

Originally Posted on  by hsauro

About ten years ago, a roman road was discovered tracing a path west of Carmarthen. Currently the road is known to reach as far as Wiston where a roman fort was recently found in 2013. This was the first roman fort ever found in Pembrokeshire. Excavation at Whitland as a result of road improvements also confirmed the presence of the road (and witnessed by the writer).

While looking along the route of the roman road on Google maps I recently noticed a rectangular structure very close to the roman road as it passed south of Clunderwen. The image below was taken from Google Maps. The rectangular structure is shown inside the red square (SN12311893). The red line is the rough alignment for the roman road. North is at the top of the image. The houses to the left of the enclosure mark the southern part of Clunderwen with the railway line arcing just at the top of center of the image. The Royal Commission marks this site as a generic ‘ENCLOSURE;DEFENDED ENCLOSURE;HENGE’ but the rectangular dimensions do not match the usual shape for an iron age or post roman site. The site was only recorded in 2011 and even the Royal Commission is unsure how to interpret the site and suggests a possible roman origin. Of more interest is that the distance between Wiston where there is a confirmed roman fort and the one suggested here is between 9 and 10 miles which is a common roman distance between marching camps. The forts that extend from Edinburgh to the north of Scotland are roughly 10 miles apart. The size of the enclosure is roughly 1.3 acres. (94 meters by 54 meters). This is small for a marching camp but not unheard of. For example Knockcross in Cumbria is only 1.5 acres.  The Wiston camp is a little bigger at about 130 by 160 meters. 

The circumstantial evidence is therefore suggestive that the rectangular enclosure south of clunderwen is in fact another Pembrokeshire roman fort. Only excavation can confirm this hypothesis. 




Rhod Kemp says:

“GAER Y;FFYNNON BRODYR

Primary Reference Number (PRN) : 3729
Trust : Dyfed
Community : Clynderwen
NGR : SN12311893
Site Type (preferred type first) : Neolithic Henge / Bronze Age Henge
Legal Protection : scheduled ancient monument

Summary :
Large oval-shaped enclosure with the lowered bowl-like interior giving the impression of an amphitheatre. Defined by an exterior bank with a possible entrance on the SSE side. The field is still cultivated and the banks have been spread by ploughing but the monument remains impressive on the landscape. This site is similar to the henge at Castell Garw (1024) and is likely to be a ritual rather than domestic/defended enclosure, dating to the late Neolithic/early Bronze age. NB 2000

Description :
Character as described by Greives, & unlike any settlement enclosure I have ever seen. Lowered interior gives impression of a large amphitheatre. Similar to Fynnon Neueydd. Nantgaredig, in respect of this lowered interior, resulting drop to inside of entrance shape. Under alternating arable/pasture regime. GH Williams 1980″

from Archwilio website (http://www.cofiadurcahcymru.org.uk/arch/dyfed/english/dyfed_interface.html) 04/01/16

Thursday, August 7, 2014

Skynet Edges Closer

Originally Posted on  by hsauro

There is a fascinating article in science this week about the construction in hardware of a neural chip. This isn’t a new idea but scale and flexibility is novel. The chip is made by researchers from IBM and Cornell and can emulate 1,000,000 ‘neurons’. The developers claim that the approach is scalable, and efficient. Apparently chips can be cascaded to make even bigger networks.

I did a back of the envelope calculation to guestimate how many transistors it might take to emulate the brains from different organisms since we know how many transistors it took to make the chip.

This sentence was taken from Ars Technica: “The new processor, which the team is calling TrueNorth, takes a radically different approach. Its 5.4 billion transistors include over (1 million neurons) 4,000 individual cores” Assuming 100 billion neurons in a human brain that means it would take:

54 Trillion transistors to make a human brain, that is or roughly equivalent to 5,200 modern PCs, assuming 10 billion transistors per PC.

In other words not currently possible (In addition the fact that the chip only talks to 256 or neurons whereas the each brain neuron talks to about 7000). The IBM team would have to connect over 5000 chips to make a human brain, not that many it seems. Some other numbers:

Jelly Fish: ~ 4 Million transistors
Pond Snail Brain: ~ 60 Million transistors
Fruit Fly Brain: ~ 540 Million transistors
Honey Bee Brain: ~5.2 Billion transistors
Cockroach: ~5.3 Billion transistors 
Frog Brain: ~86 Billion transistors
Rat Brain: ~1 Trillion transistors
Human Brain: ~54 Trillion transistors

The single chip could just about simulate a Cockroach brain.


Brain data from http://en.wikipedia.org/wiki/List_of_an … of_neurons1 and
http://en.wikipedia.org/wiki/Neuron2



Friday, March 28, 2014

Reversibility and Product Inhibition

 Originally Posted on  by hsauro

It is common in biochemistry to categorize reactions as either reversible or irreversible. What do we mean by these terms? Let’s look at a simple reaction such as:

  \[A \rightarrow B\]

where A is transformed to B. The rate of reaction for this transformation can be assumed to be first-order, that is:

  \[v = k_1 A\]

Mathematically this is an irreversible reaction because the reaction only goes from reactant A to product B. We can easily make the reaction reversible by adding the reverse reaction:

  \[A \rightleftharpoons B\]

If we again assume first-order kinetics for the reverse process then the reaction rate is now given by:

  \[v = k_1 A - k_2 B\]

Depending on the ratio of k_1 to k_2, if the concentration of B is high enough the reaction rate, v will actually go negative. This is the main characteristic of a reversible reaction, the potential for the forward reaction rate to go negative. If k_2 were very small relative to k_1 we might find that the concentration of B required to make the reaction go in reverse is unrealistically high. For example, if k_1 = 1 and k_2 = 0.0001, and the concentration of A was 1 mM, then the concentration of B required to reverse the reaction would be 10 M. In a biological cell such a concentration would be highly unlikely. In this situation we would probably assume for all practical purposes that our reaction is irreversible.

Another way to look at this is to recall that the ratio of the forward to backward rate constant (k_1/k_2) is equal to the equilibrium constant, K_{eq} = B/A. In our example the equilibrium constant was 10,000. One may think that such equilibrium constants are quite rare but as Cornish-Bowden and Cardenas reported, the equilibrium constant for pyruvate kinase is of the order of 100,000.

An important question is how should we represent reactions that have high equilibrium constants in our computer models? Do we simply use irreversible rate laws to model such reactions? The short answer to this is probably no.

What we often forget is that the product of an enzyme catalyzed reaction can reversible bind to the enzyme and when it does it will compete with the substrate. This means that the reaction rate is likely to decrease as product accumulates simply due to the competitive effect of product. This effect is called product inhibition. For a Michaelis-Menten like rate law it is easy to adjust the equation to take into account product inhibition, we just treat the product as a competitive inhibitor:

  \[v = \frac{V_m S}{S + K_m \left( 1 + \frac{P}{K_p} \right)}\]

where V_m is the maximal rate, K_m the substrate concentration with no product that gives half the maximal rate and K_p is the dissociation constant for the product. Depending on the value for K_p it is quite possible for the product to greatly affect the forward rate. The reaction is irreversible but the forward reaction can be reduced via product inhibition.


Sunday, December 29, 2013

Converting Excel Tables to LaTeX Tables

Originally Posted on  by hsauro

Today I had the problem of converting an Excel table of values into a LaTeX table. I could have done it by hand but the table had over 400 entries so that would have been a bit tedious. Instead I did the usual and hunted around for any automated possibilities and came across two:

1. Excel2LaTeX.

This is a Excel Add-In that converts a selected set of Excel rows and columns into a LaTeX table. However as per usual for Microsoft it isn’t trivial to add Add-Ins. After some fuzzing around I discovered these instructions on stackoverflow. One would think that adding an Add-In to Excel would be a case of simply loading the Add-In file, but of course not, it requires a minimum of at least 8 different steps. But it doesn’t end there. The final step, number 9, is to close Excel, yes I said close Excel, restart Excel and the Add-In is available (hopefully). After that the Add-In is fairly trivial to use and offers a number of useful formating options..

ExcelToLatex1

2. The second option is an online tool by Eric Wood This was very easy to use, just drag your excel file on the the browser window. If your data is on a particular sheet, make sure to select that sheet on the provided dropdown list.

ExcelToLatex2


Converting Excel Tables to LaTeX Tables

December 29, 2013 6:31 pm

Today I had the problem of converting an Excel table of values into a LaTeX table. I could have done it by hand but the table had over 400 entries so that would have been a bit tedious. Instead, I did the usual and hunted around for any automated possibilities and came across two:

1. Excel2LaTeX.

This is a Excel Add-In that converts a selected set of Excel rows and columns into a LaTeX table. However as per usual for Microsoft it isn't trivial to add Add-Ins. After some fuzzing around I discovered these instructions on stackoverflow. One would think that adding an Add-In to Excel would be a case of simply loading the Add-In file, but of course not, it requires a minimum of at least 8 different steps. But it doesn't end there. The final step, number 9, is to close Excel, yes I said close Excel, restart Excel and the Add-In is available (hopefully). After that, the Add-In is fairly trivial to use and offers a number of useful formatting options..


2. A second option is an online tool by Eric Wood This was very easy to use, just drag your excel file on the browser window. If your data is on a particular sheet, make sure to select that sheet on the provided dropdown list.



Monday, October 28, 2013

Bottlenecks and Slow Steps - again

October 28, 2013 5:55 pm

I've written a few articles on this blog about the irrationality of bottlenecks and slow steps in metabolism, hoping that one day my fellow scientists will feel the same. Sadly it is not true and maybe I should write more because I week or two ago I came from a talk about metabolic engineering which made some egregious statements that I wouldn't even expect my undergraduates to make. Apparently *the* bottleneck (note singular) in a metabolic pathway is the 'slowest' step and that's the one to engineer. I should add that this not quite right. Unfortunately I've no idea what they meant by the slowest step and I didn't have the heart to ask them given how lost they were. If I did ask, I suspect they'd become even more confused.

I think its time to write a blog on "What's the deal with slow steps"'

Thursday, September 12, 2013

How to Increase MikTeX 2.9 Memory

Originally Posted on  by hsauro

I was recently running some large stochastic simulations and was generating very large pgfplot data files. MikTeX out of the box was unable to compile these plots because it kept running out of main memory. To increase main memory (or other memory limits) run the following line from the DOS prompt:

initexmf --edit-config-file pdflatex

This will open the configuration file for pdflatex into an editor (probably notepad). Add the following lines to this configuration file:

pool_size=5000000
main_memory=6000000
extra_mem_bot=2000000
font_mem_size=2000000

I needed lots of memory so these numbers are quite bit. I’ve got 8GB of RAM so let’s be generous (My first computer was the MK14 which had 256 bytes of RAM, things have moved on a little).

Save the file (file name is provided) and then remake the format files. To do this use the command line:

initexmf --dump=pdflatex

I happen to use pdflatex all the time so note that I am specifying pdflatex in each of the commands. If you’re using straight latex then substitute pdflatex with latex.

Talking about big tikz/pfdplot figures, don’t forget to use the external command:

\usepackage{tikz}
\usetikzlibrary{external}
\tikzexternalize % activate!

This forces tikz to reuse previously stored figures, saves regenerating them all the time. See section 32 of the TikZ manual (Version 2.10 !) for more details.

Saturday, July 13, 2013

Filming Paramecium bursaria

July 13, 2013 5:16 pm

I got myself a camera attachment from AmScope for the relatively cheap microscope I have (A OMAX 40X-2000X Digital Binocular Biological Compound Microscope - find it on Amazon). I used my wife's Cannon 3Ti to take the video. Got some water form one of my plant trays and had a look. Not much life, a few unidentified and quite small ciliates, a few Euglena, some unidentified filamentous algae and what looked like some spirochetes. But what caught my attention was one Paramecium and not any old Paramecium but a Paramecium called Paramecium bursaria. What is interesting about this particular Paramecium is that it has a symbiotic relationship with green algae called Zoochlorella.

The video below (taken at roughly 600x to 800x) shows a number of interesting features, some of which I've never seen myself before and it was very interesting when I realized what was going on. In the video you'll see two contractile vacuoles pumping water out of the cell, a large nucleus near the center and an oral grove to the left of the nucleus with some beating cilia. Also notice the many symbiotic algae in the cell. Also around the outside of the Paramecium you'll also see beating cilia. The entire cell is about 100 to 150 microns long. It took a while for the cell to settle down so that I could photograph it. One question, in the first image you'll see some long thin algae (?) like cells drifting from the top to the bottom left corner. I have not been able to identify these but the sample was full of them and I suspect is what gave the water a slightly green color.