Transcript Travis Metcalfe

Asteroseismology of Sun-like Stars
Travis Metcalfe (HAO)
The Internal Constitution of the Stars
“At first sight it would seem that the deep
interior of the sun and stars is less accessible
to scientific investigation than any other region
of the universe.
Our telescopes may probe farther and farther
into the depths of space; but how can we ever
obtain certain knowledge of that which is
hidden behind substantial barriers?
Sir Arthur Eddington
(1882 – 1944)
What appliance can pierce through the outer
layers of a star and test the conditions within?”
(written in 1926)
Seismology: seeing with sound
Convection creates acoustic noise – some of it resonates
Motivation
New opportunities to probe the
fundamental physics of solar
and stellar models.
Understanding the solar
structure and evolution in a
broader physical context.
1D oscillations: violin strings
Fundamental
Third overtone
First overtone
Second overtone
2D oscillations: drums
Radial modes
Non-radial modes
2D oscillations: drums
Radial modes
Non-radial modes
3D oscillations: stars
Radial modes
Non-radial modes
3D oscillations: stars
Radial modes
Non-radial modes
Observations
Pulsations cause variations in spectral lines and brightness
Space missions
Space missions will soon revolutionize the observations
Space missions
Space missions will soon revolutionize the observations
Space missions
Space missions will soon revolutionize the observations
Epistemology
Matching models to observations is an optimization problem
Optimization
Easy
Optimization
Hard
Evolution as optimization
“Evolution is cleverer than you are.” – Francis Crick
Evolution as optimization
“Evolution is cleverer than you are.” – Francis Crick
Genetic algorithms
1. Generate N random trial sets of parameter
values.
2. Evaluate the model for each trial and
calculate the variance.
3. Assign a “fitness” to each trial, inversely
proportional to the variance.
4. Select a new population from the old one,
weighted by the fitness.
5. Encode-Breed-Mutate-Decode
6. Loop to step 2 until the solution converges.
Evolutionary operators
Parallel computing
• Genetic algorithms are
intrinsically parallelizable
• Each iteration typically has
128 model evaluations
• Number of processors sets
the number of models that
can be evaluated in parallel
• Also need multiple runs with
different random initialization
Stellar parameters
• Total Mass
• Surface Temperature
• Chemical composition
• Convective efficiency
• Internal chemical gradients
• Rotation rate
• Age / Evolutionary status
The Future: Eddington et al.
2007 and beyond: a flood of unprecedented observations