Comments on: When explaining isn’t helping http://www.orangewallaby.net/wall/2007/when-explaining-isnt-helping personal and professional notes from a naive chemistry professor Wed, 20 May 2009 18:19:07 +0000 hourly 1 http://wordpress.org/?v=3.1.3 <img class='wavatar' src='http://www.gravatar.com/avatar.php?gravatar_id=5c33b55faf2677fe8d90af73d81d2f6b.jpg&s=80&d=http%3A%2F%2Fwww.orangewallaby.net%2Fwall%2Fwp-content%2Fcache%2Fwavatars%2F5c33b55faf2677fe8.png' width='80' height='80' alt='Wavatar' />By: Wallaby http://www.orangewallaby.net/wall/2007/when-explaining-isnt-helping/comment-page-1#comment-10842 Wavatar Wallaby Tue, 18 Sep 2007 21:29:37 +0000 http://www.orangewallaby.net/wall/2007/when-explaining-isnt-helping#comment-10842 Good guess. The whole thing's been working for weeks now. Thanks for explaining that "tweaking for the biggest signal is usually enough". You're right, of course. The whole point of this post was not so much my (passing) frustration with an experiment that didn't work (then). Rather, the point was that telling someone to "just tweak for the biggest signal" - when they already know to do so, and realize it's a matter of time and careful work to get there - isn't useful at all. An example? Suppose someone wants his startup company to double its sales in 2008. Growth's been a bit flat lately, so he hires a consultant to come in and point things in the right direction. The consultant says "just find more people who want to buy your product, or develop more products that people want to buy". And then sends a bill for $50k. Sure, it's the right idea, but there's no actionable information the company didn't already have. (Still more useful than trying to drum up business by looking for postdocs blogging about trying to measure pulse width.) Good guess. The whole thing’s been working for weeks now.

Thanks for explaining that “tweaking for the biggest signal is usually enough”. You’re right, of course. The whole point of this post was not so much my (passing) frustration with an experiment that didn’t work (then). Rather, the point was that telling someone to “just tweak for the biggest signal” – when they already know to do so, and realize it’s a matter of time and careful work to get there – isn’t useful at all.

An example? Suppose someone wants his startup company to double its sales in 2008. Growth’s been a bit flat lately, so he hires a consultant to come in and point things in the right direction. The consultant says “just find more people who want to buy your product, or develop more products that people want to buy”. And then sends a bill for $50k. Sure, it’s the right idea, but there’s no actionable information the company didn’t already have.

(Still more useful than trying to drum up business by looking for postdocs blogging about trying to measure pulse width.)

]]> <img class='wavatar' src='http://www.gravatar.com/avatar.php?gravatar_id=dccd9f7bd2597a25dff3a9a65bdd8aef.jpg&s=80&d=http%3A%2F%2Fwww.orangewallaby.net%2Fwall%2Fwp-content%2Fcache%2Fwavatars%2Fdccd9f7bd2597a25d.png' width='80' height='80' alt='Wavatar' />By: Dan http://www.orangewallaby.net/wall/2007/when-explaining-isnt-helping/comment-page-1#comment-10821 Wavatar Dan Tue, 11 Sep 2007 20:32:00 +0000 http://www.orangewallaby.net/wall/2007/when-explaining-isnt-helping#comment-10821 OK, it sounds like you are doing an interferometric autocorrelation. (Colinear beams hitting the photodiode?) This gives more information than interferometry with a linear detector, which gives only the pulse spectrum. It provides the autocorrelation plus the second harmonic spectrum. As you say, the experiment should be the guide as to the information you require. For example, in the case of DFWM, a chirped pulse will reduce the amount of signal you see and may skew conclusions. If you just need a big hammer, and there is no physics (or chemistry) in the interaction of the pulse with the sample, just tweaking for the biggest signal is usually enough. Dan OK, it sounds like you are doing an interferometric autocorrelation. (Colinear beams hitting the photodiode?) This gives more information than interferometry with a linear detector, which gives only the pulse spectrum. It provides the autocorrelation plus the second harmonic spectrum. As you say, the experiment should be the guide as to the information you require. For example, in the case of DFWM, a chirped pulse will reduce the amount of signal you see and may skew conclusions. If you just need a big hammer, and there is no physics (or chemistry) in the interaction of the pulse with the sample, just tweaking for the biggest signal is usually enough.

Dan

]]> <img class='wavatar' src='http://www.gravatar.com/avatar.php?gravatar_id=5c33b55faf2677fe8d90af73d81d2f6b.jpg&s=80&d=http%3A%2F%2Fwww.orangewallaby.net%2Fwall%2Fwp-content%2Fcache%2Fwavatars%2F5c33b55faf2677fe8.png' width='80' height='80' alt='Wavatar' />By: Wallaby http://www.orangewallaby.net/wall/2007/when-explaining-isnt-helping/comment-page-1#comment-10791 Wavatar Wallaby Fri, 17 Aug 2007 13:44:06 +0000 http://www.orangewallaby.net/wall/2007/when-explaining-isnt-helping#comment-10791 Coincidence I'm sure that your company sells the stuff you recommend, but at least it's helpful info... Ahem. The detector is a two-photon photodiode. We have the equipment for FROG, but that's not the direction this particular project is going. Interferometry actually works better for my purposes, which don't require pulse characterization at that level. Thanks for reading, and for the suggestion! Coincidence I’m sure that your company sells the stuff you recommend, but at least it’s helpful info… Ahem.

The detector is a two-photon photodiode. We have the equipment for FROG, but that’s not the direction this particular project is going. Interferometry actually works better for my purposes, which don’t require pulse characterization at that level.

Thanks for reading, and for the suggestion!

]]> <img class='wavatar' src='http://www.gravatar.com/avatar.php?gravatar_id=a147e538c16e89053a1c56a223f0bf4b.jpg&s=80&d=http%3A%2F%2Fwww.orangewallaby.net%2Fwall%2Fwp-content%2Fcache%2Fwavatars%2Fa147e538c16e89053.png' width='80' height='80' alt='Wavatar' />By: Daniel J. Kane http://www.orangewallaby.net/wall/2007/when-explaining-isnt-helping/comment-page-1#comment-10789 Wavatar Daniel J. Kane Thu, 16 Aug 2007 20:19:52 +0000 http://www.orangewallaby.net/wall/2007/when-explaining-isnt-helping#comment-10789 How do you plan to measure an ultrafast laser pulse with just an interferometer? You can't do it. An interferometer will only give you the pulse spectrum. You need to include a nonlinearity (second harmonic generation crystal) to give you an autocorrelation. (A blobby representation of the pulse at best.) Spectrally resolving the autocorrelation yields a "FROG trace." The using an iterative phase retrieval algorithm characterizes the pulse: Temporally and spectrally. You have not only the temporal shape, but the spectral phase--where the frequencies are in the pulse. Dan How do you plan to measure an ultrafast laser pulse with just an interferometer? You can’t do it. An interferometer will only give you the pulse spectrum. You need to include a nonlinearity (second harmonic generation crystal) to give you an autocorrelation. (A blobby representation of the pulse at best.) Spectrally resolving the autocorrelation yields a “FROG trace.” The using an iterative phase retrieval algorithm characterizes the pulse: Temporally and spectrally. You have not only the temporal shape, but the spectral phase–where the frequencies are in the pulse.

Dan

]]>