robryk @robryk@qoto.org

An amusing way to explain that garbage input produces garbage output.

I am going to CCC (#39c3). If everything works, I will be reachable by phone at 7275.

If you perform the following procedure:
1. Draw A from Poisson with mean x.
2. Sum up A i.i.d. +1/-1 Bernoulli variables with mean y.

You'll get a Gaussian random variable, with -- expectedly -- mean of x*y, but somewhat surprisingly with variance x.

It's somewhat nice (and surprising for me) that the variance is independent of the Bernoulli distribution skew.

@eventphone Werden die Kurztextnachrichten über DECT während 39c3 funktionieren? (Mitel's Doku scheint zu sagen, dass das selbstverständlich innerhalb ein OMM ist, aber ich kenne mich mit DECT-Infrastruktur nicht gut aus.)

Random, potentially useful fact: powered wheelchairs seem to reasonably often be convertible to push mode. Look for a lever somewhere around one of the motors which switches between powered operation&break enabled and freewheeling (with brake disabled! note that they're potentially heavy).

@_thegeoff do you have experience or ideas on how to mask off infrared leakage from an emitter to a detector that are next to each other (the intent is to detect reflections/diffuse reflections)?

I currently have 5mm IR LEDs and a photodiode next to each other, in black heat shrink that extends down to the slight ridge in the base, but don't overlap it. I still get significant signal when there's nothing in front and adding more layers of heat shrink doesn't help. My best guess is that I am seeing emission going backwards and scattering off the pcb into the photodiode, but am somewhat unsure how to deal with that given that I need something ir-opaque, nonconductive, and able to withstand the heat of soldering.

Aaand the media file got lost.

This is not a sound recording; it's a recording of intensity of reflected IR. :)

This is not a sound recording; it's a recording of intensity of reflected IR. :)

It seems that black heatshrink and aluminium foil are much less IR-translucent than heatshrink of other colors and cardstock. Heatshrink of other colors isn't that surprising, but I was really surprised to see how translucent cardstock is.

WTF Osram, why do your LEDs have longer cathodes than anodes? (https://look.ams-osram.com/m/6985efa866041336/original/SFH-4547.pdf)

```
LOG_NLS_V | info | Start solving Non-Linear System 207 (size 8) at time 0 with Hybrd Solver
| | | | | 1. der(reactor.coolant_vessel.medium.h) = 0.000000
| | | | | | nominal = 250000000.000000
| | | | | | old = 0.000000
| | | | | | extrapolated = 0.000000
| | | | | 2. test_loop.flow_restr_2.flowModel.states[2].h = 100000.000000
| | | | | | nominal = 500000.000000
| | | | | | old = 100000.000000
| | | | | | extrapolated = 0.000000
| | | | | 3. test_loop.pump.heatTransfer.states[1].h = 84030.539315
| | | | | | nominal = 500000.000000
| | | | | | old = 84030.539315
| | | | | | extrapolated = 0.000000
| | | | | 4. test_loop.pump_2.heatTransfer.states[1].p = 119900.000000
| | | | | | nominal = 100000.000000
| | | | | | old = 119900.000000
| | | | | | extrapolated = 0.000000
| | | | | 5. test_loop.pump_2.heatTransfer.states[1].h = 84030.539315
| | | | | | nominal = 500000.000000
| | | | | | old = 84030.539315
| | | | | | extrapolated = 9701582846615680389965170410065635635592644909986107348146830517680841797018933994035582096881053530010197810704725168063739672624651808469562738004095429049456876177299072376746747482865664.000000
| | | | | 6. test_loop.pump.heatTransfer.states[1].p = 119900.000000
| | | | | | nominal = 100000.000000
| | | | | | old = 119900.000000
| | | | | | extrapolated = 1395545749263273562353094300355912123666420127868569549243039153200782685590445388375565229326362995465733389002326171356874244432514685387165896028648662914933266970943160745918382187935715352284279593816494624266522555068710912.000000
| | | | | 7. test_loop.pump.N = 1500.000000
| | | | | | nominal = 1.000000
| | | | | | old = 1500.000000
| | | | | | extrapolated = 107477949771266220580127046916236898947795277663670987074868389930580837650472937515744177753777718814327714610808858911987699308800252211914642615367406474047816442643162274605062751179869984619432932445505357721667305696715179033483470650708221688317554936849744913235968.000000
| | | | | 8. test_loop.pump_2.N = 1500.000000
| | | | | | nominal = 1.000000
| | | | | | old = 1500.000000
| | | | | | extrapolated = 24371644343455766312317374891129151099810235543819179569928328023469746077106736735078443514579514619241838070512462790807003317472949993331128066179656919744985893279289689732107804348505109936163901058751957917554869003401675407360.000000
```

Well, ok, but _why_? _What_ is the system of equations that leads to this absurd outcome during search for valid initial values?

Let's assume we have a fixed container filled with some amount of a liquid and its vapor. If we heat it up, obviously the fraction of mass that's in vapor form will increase (being in gas is higher energy state, fraction of stuff that's in a higher energy state increases as energy increases in a thermodynamic equilibrium; I might be wrong if there's something really wonky going on with surface tension). What about volume fraction? At first glance it seems that it can change in either direction.

https://en.wikipedia.org/wiki/Nusselt_number#Dittus%E2%80%93Boelter_equation says:

> n = 0.4 for the fluid being heated, and n = 0.3 for the fluid being cooled.

WTAF. Why do we have different power laws for heat transfer between a solid and liquid when the flow is turbulent _depending on the direction of heat transfer_? I can't think of any simple mean field approximation of the process that would yield that.

#physics #fluiddynamics