Monday, May 30, 2011

2 more days of isolation

...until I finish all the cases. This is the only site I will visit.

Update (30/5, 11:13): run-time error M6110: MATH - floating-point error: stack overflow http://bit.ly/ig7faQ. Seems like there is no direct way out. I will have to split the program.

Update (30/5, 12:52): Devised a bypass for the stack overflow problem. This is not robust at all. Need to get some outputs quick before the internal heap allocation algorithm behaves like crazy.

Update (30/5, 14:22): Had food. Getting ready with pen and paper. Need to jot down all cases I have to run.

Update (30/5, 15:26): Took almost an hour but I have raw data of all the four essential cases. Will compare them and run two more additional cases if things appear fine.

Update (30/5, 15:47): Finished all essential and additional cases. Need some sleep. Will return to the problem as soon as I regain consciousness.


Update (30/5, 18:11): Working on the appendix - two methods to analyze dynamic systems, Have already written one of the methods.

Update(31/5, 12:04): Finished the Appendix (both sections).

Saturday, May 28, 2011

Further micro-level completion plans

Dear readers/myself,

Today's job.
a) Complete the four schematics.
b) Run at least 4-5 cases.

Regards.

Thursday, May 26, 2011

2 days of isolation

I'll give myself two days of isolation - no phones, no connection, no internet ^except this blog , no distraction - just lab, food and night sleep. And skim through the code to see what in the fucking segment is going so wrong. After many futile half-hearted attempts, this appears to be the only way out.

Update (26/5, 13:19): I have obtained a decent bifurcation width for second order cases. The undamped values appear fine but the damped values are off.

Update (26/5, 13:29): Although it is not mentioned anywhere but are these values to be computed using largest possible bifurcation? Each order has two roots and there are two roots in each order. Also at higher orders, the offset errors will diminish.

Update (26/5, 13:52): Test plot looks great. The damped values are still a bit off.

Update (26/5, 14:42): Detected a small error in coding. This factor has very little contribution to the overall effects. The damped values are still off. Must check original formulation by Deolasi.

Update (26/5, 16:00): No sign of improvement. No sign of my guide either. Nothing will work out unless I meet him.

Update (27/5, 10:30): Met my guide. He is fine with whatever results I am obtaining as long as it is legit. Getting back to writing a few crucial pages about the general nature of instability.

Wednesday, May 25, 2011

Working for Dark Project

Dear readers,

While I have been encountering failures after failures for my thesis, things are looking up bright for the work going on for my band, DP.

Yesterday, I sent across a non-automated mixdown of our forthcoming single that made most people happy. Let me shed some light on the thing. For the last one week I have been trying hard to mix the track. I don't use the best gear in the world but I know my gear and their fallacies fairly well - a knowledge indispensable to the mixing engineer. I have been trying hard to implement Michael H. Brauer's ABCD technique using plugins. Sadly I landed up with two versions of crap. Even a frequency analyzer *showed* me how off I was. The low end was suppressed to the point that the mix started sounding tinny. I would have pushed forward even more had I not come across a statement by MHB himself about using his ABCD trick for mixing in-the-box - "For those who think it can be done, do it. But I'm telling you, I've tried what mike is talking about and it works in only one application but not in combination. I'm not going to get into a debate on this issue. As some of the other guys are realizing,my complete multibuss concept was designed to work in the analog world. The feel that i'm looking for get's blurred and the tones get distorted when in the box. Because it is necessary to fully combine the techniques of multibuss compression, parallel compression and the use of send/return compression, the routing in the digital domain causes issues that ADC can't compensate for and causes undesirable results. Meaning it ain't f*cking musical."

I have my share of methods - things that are time tested and work wonders (provided you have well recorded tracks to begin with). This time I used Anesthetize by Porcupine Tree as a frequency and leveling benchmark even if these songs couldn't be more apart stylistically. Also for the first time, I have mixed using REAPER as my primary DAW.

On another side, I have to re-design the official website, too. Yesterday, I submitted a prototype skin to Sudipto. He liked it. The biggest hurdle will be to author the media player. I have my eye on the flowplayer. However, reconstructing the skin will be a difficult job to do. The artwork for the album cycle needs to be as organic as the music itself. In short, I will have to meticulously hand draw them.

I should take a break for a few hours and get back to my PhD problem - one that's giving me the most amount of headaches.

Regards.

P.S. Thanks to Steven Wilson and the journalists who interviewed him and put his mind out in the public.

Tuesday, May 24, 2011

Failures of the n-th order.

Dear readers,

I have been facing failures of the n-th order. The mixes I make are pathetic. The speakers I have are inaccurate and the output doesn't translate well to other systems.

Also, recently caught a bug in the program. Sets me back by a week or so. Here I am, trying to come up with a feasible completion strategy.

Regards.

Monday, May 9, 2011

In Bangalore

Dear readers,

I have safely reached Bangalore today. HWH-YPR Duronto ran pretty well and reached before time - something unimaginable in India. Today is a day of rest, to get rid of the buzz that arises due to imbalance in ampullary cupula - lack of proprioception.

There is also a herculean task of setting up Sudipto's computer with all the necessary software for recording.

Regards.

Tuesday, May 3, 2011

List of things to carry to Bangalore.

  1. Instruments/Equipments
    1. Guitar+Picks+Extra Strings
    2. POD+Controller
    3. Pedals
    4. Cables
    5. Netbook (charged) (+charger)
    6. Backup DVDs 
    7. Phone (charged) (+charger)
  2. Softwares
    1. REAPER
    2. ezDrummer
    3. Waves Plugin
    4. Amplitube / Ampeg VST
    5. Voxengo SPAN, Overtone EQ, Saturator
    6. Nuendo (as a backup)
    7. Production audio tracks
  3. Clothes
    1. 2x Trousers (+Tech Market)
    2. 1x Cargo track.
    3. 5x Tee
    4. Bermudas
    5. Towel
    6. Wash clothes.
  4. Get books from Mehta + Firoze

        Not a very productive day on my part, but....

        ... got a lot of things straightened out with my guide. Now all I need is a micro strategy and a list of possible cases to iterate on. That should do the job. I only have four more days to finish this part before leaving for the recording session.

        Monday, May 2, 2011

        The problem with damping

        To myself,

        The code from the Deolasi formulation is now yielding some results. However the accuracy is doubtful. I have a fair idea that Udar's results might be erroneous. At best, I was able to match only sections. This is especially true if a second order method is used. While dealing with combination resonance frequencies of higher modes, the onset and width of the first-order method correspond to what's reported by Udar and Datta.

        Also, the values match when the damping parameter, $\zeta=0.005$, instead of $\zeta=0.05$.

        Need to have a talk with my guide.

        Regards.